{"aliases": [], "authors": ["Panupong Pasupat", "Tian-Shun Jiang", "Evan Zheran Liu", "Kelvin Guu", "Percy Liang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "f368116f70e987cde111835a05b5b36e0d345dabb4e8351610e2bbfb691b3440", "submitted_at": "2018-08-28T06:09:39Z", "updated_at": "2018-10-01T03:22:58Z", "version": 2, "versioned_url": "https://arxiv.org/abs/1808.09132v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=1802.08802%2C1808.09132%2C2005.03776%2C2101.09465%2C2103.16057%2C2104.08560%2C2105.13231%2C2107.03374%2C2107.13731%2C2108.03353%2C2202.02312%2C2202.08137%2C2204.01691%2C2205.11029%2C2207.01206%2C2209.08199%2C2209.08655%2C2209.14927%2C2209.15099%2C2210.02663%2C2210.03629%2C2210.03945%2C2211.07615%2C2301.10165%2C2301.13280%2C2302.04761%2C2303.07280%2C2303.08774%2C2303.17491%2C2303.17580%2C2304.04370%2C2304.07061%2C2305.04091%2C2305.05662%2C2305.08144%2C2305.10601%2C2305.11854%2C2305.12487%2C2305.14257%2C2305.14325%2C2305.16504%2C2305.16653%2C2305.19308%2C2306.00245%2C2306.06070%2C2306.07863%2C2307.10088%2C2307.12856%2C2307.13854%2C2308.01990&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=1802.08802%2C1808.09132%2C2005.03776%2C2101.09465%2C2103.16057%2C2104.08560%2C2105.13231%2C2107.03374%2C2107.13731%2C2108.03353%2C2202.02312%2C2202.08137%2C2204.01691%2C2205.11029%2C2207.01206%2C2209.08199%2C2209.08655%2C2209.14927%2C2209.15099%2C2210.02663%2C2210.03629%2C2210.03945%2C2211.07615%2C2301.10165%2C2301.13280%2C2302.04761%2C2303.07280%2C2303.08774%2C2303.17491%2C2303.17580%2C2304.04370%2C2304.07061%2C2305.04091%2C2305.05662%2C2305.08144%2C2305.10601%2C2305.11854%2C2305.12487%2C2305.14257%2C2305.14325%2C2305.16504%2C2305.16653%2C2305.19308%2C2306.00245%2C2306.06070%2C2306.07863%2C2307.10088%2C2307.12856%2C2307.13854%2C2308.01990&max_results=50"}], "id": "arxiv:1808.09132", "identifiers": {"arxiv": "1808.09132"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:20Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/1808.09132"}], "name": "Mapping Natural Language Commands to Web Elements", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2018-08-28", "relations": [], "review_status": "curated", "scope": "core", "summary": "Defines language grounding on webpages as choosing the UI element targeted by a natural-language command. A collection of more than 50,000 commands and three baseline models study functional references, relationships between elements, and visual position cues.", "topics": ["grounding", "data-generation", "evaluation"], "updated_at": "2018-10-01", "url": "https://arxiv.org/abs/1808.09132"}
{"aliases": [], "authors": ["Yang Li", "Jiacong He", "Xin Zhou", "Yuan Zhang", "Jason Baldridge"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "3af7b45455ffbd616e8cdf5612b19275ecb7bf07b1065af9f5b724ae19e96fb4", "submitted_at": "2020-05-07T21:41:40Z", "updated_at": "2020-06-05T02:11:56Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2005.03776v2"}, "publication": {"authors": ["Yang Li", "Jiacong He", "Xin Zhou", "Yuan Zhang", "Jason Baldridge"], "date_basis": "Explicit publisher metadata: 2020/7; precision=month.", "date_parts": [2020, 7], "doi": "10.18653/v1/2020.acl-main.729", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2020.acl-main.729/"], "identity_basis": "Same work: exact normalized landing title, corresponding author list, and matching specific contribution. Publisher and arXiv versions remain distinct.", "issue": null, "notes": "Publisher-hosted proceedings record; distinct from the retained arXiv preprint snapshot.", "pages": "8198-8210", "pdf_url": "https://aclanthology.org/2020.acl-main.729.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "Mapping Natural Language Instructions to Mobile UI Action Sequences", "type": "conference-paper", "url": "https://aclanthology.org/2020.acl-main.729/", "venue": "Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics", "volume": null}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=1802.08802%2C1808.09132%2C2005.03776%2C2101.09465%2C2103.16057%2C2104.08560%2C2105.13231%2C2107.03374%2C2107.13731%2C2108.03353%2C2202.02312%2C2202.08137%2C2204.01691%2C2205.11029%2C2207.01206%2C2209.08199%2C2209.08655%2C2209.14927%2C2209.15099%2C2210.02663%2C2210.03629%2C2210.03945%2C2211.07615%2C2301.10165%2C2301.13280%2C2302.04761%2C2303.07280%2C2303.08774%2C2303.17491%2C2303.17580%2C2304.04370%2C2304.07061%2C2305.04091%2C2305.05662%2C2305.08144%2C2305.10601%2C2305.11854%2C2305.12487%2C2305.14257%2C2305.14325%2C2305.16504%2C2305.16653%2C2305.19308%2C2306.00245%2C2306.06070%2C2306.07863%2C2307.10088%2C2307.12856%2C2307.13854%2C2308.01990&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:53:48+00:00", "role": "primary", "sha256": "c196294e19378d08d71a7cfd41c2e368f5a6b49ee7c3601257537b9ae5c42add", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.publisher", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue"], "url": "https://aclanthology.org/2020.acl-main.729/"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=1802.08802%2C1808.09132%2C2005.03776%2C2101.09465%2C2103.16057%2C2104.08560%2C2105.13231%2C2107.03374%2C2107.13731%2C2108.03353%2C2202.02312%2C2202.08137%2C2204.01691%2C2205.11029%2C2207.01206%2C2209.08199%2C2209.08655%2C2209.14927%2C2209.15099%2C2210.02663%2C2210.03629%2C2210.03945%2C2211.07615%2C2301.10165%2C2301.13280%2C2302.04761%2C2303.07280%2C2303.08774%2C2303.17491%2C2303.17580%2C2304.04370%2C2304.07061%2C2305.04091%2C2305.05662%2C2305.08144%2C2305.10601%2C2305.11854%2C2305.12487%2C2305.14257%2C2305.14325%2C2305.16504%2C2305.16653%2C2305.19308%2C2306.00245%2C2306.06070%2C2306.07863%2C2307.10088%2C2307.12856%2C2307.13854%2C2308.01990&max_results=50"}], "id": "arxiv:2005.03776", "identifiers": {"arxiv": "2005.03776"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:20Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2005.03776"}, {"rel": "publication", "url": "https://aclanthology.org/2020.acl-main.729/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2020.acl-main.729.pdf"}], "name": "Mapping Natural Language Instructions to Mobile UI Action Sequences", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2020-05-07", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces PIXELHELP for predicting mobile UI action sequences from instructions, together with action-phrase annotations and synthetic UI action descriptions for training. Separate Transformers extract instruction steps and ground object descriptions using interface content and screen position.", "topics": ["mobile-navigation", "grounding", "data-generation", "action-prediction"], "updated_at": "2020-06-05", "url": "https://arxiv.org/abs/2005.03776"}
{"aliases": [], "authors": ["Jingwen Fu", "Xiaoyi Zhang", "Yuwang Wang", "Wenjun Zeng", "Sam Yang", "Grayson Hilliard"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "8788e419d1a901f8ad9dd6089a12fe129baee764b1c285bbeb4ea60d7046afa2", "submitted_at": "2021-05-25T13:45:54Z", "updated_at": "2022-04-14T11:00:58Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2105.11941v2"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T07:57:26Z", "role": "primary", "sha256": "3e89566cdb4521a8b3b0d99c514c3401645eda5b5c0de9720eb726b1ffe0a11f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=1511.03745%2C1602.02261%2C1809.04860%2C1811.06521%2C1811.07871%2C1812.03426%2C1901.11528%2C1909.00031%2C1912.03098%2C2003.00431%2C2004.14973%2C2006.16241%2C2007.04428%2C2010.09670%2C2010.12083%2C2012.12350%2C2104.08244%2C2105.11941%2C2106.09422%2C2106.13500%2C2107.05911%2C2108.03298%2C2108.05308%2C2110.10661%2C2201.04100%2C2201.11676%2C2203.10897%2C2204.05239%2C2205.10816%2C2206.08853%2C2206.12839%2C2206.15407%2C2208.06032%2C2208.08080%2C2209.14981%2C2210.01379%2C2210.03735%2C2210.08073%2C2210.10275%2C2210.11435%2C2211.07912%2C2211.09527%2C2212.13326%2C2301.10677%2C2301.13779%2C2302.12173%2C2303.00001%2C2303.11366%2C2303.12027%2C2304.03442&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:11:47Z", "role": "primary", "sha256": "41c4da71e457ea23725a8070b0da64c234f03613841b95edea3a5f3c95dd8c68", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/abs/2105.11941v2"}], "id": "arxiv:2105.11941", "identifiers": {"arxiv": "2105.11941"}, "kind": "paper", "last_verified_at": "2026-10-17T09:11:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2105.11941"}], "name": "Understanding Mobile GUI: from Pixel-Words to Screen-Sentences", "notes": "Reviewed against the exact-version official arXiv abstract page; its normalized abstract SHA-256 matches the retained bibliographic snapshot. The original API response body was not locally available for this review. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2021-05-25", "relations": [], "review_status": "curated", "scope": "core", "summary": "Models mobile screens as atomic text and graphic components whose relationships are learned by a Screen Transformer. The work introduces RICO-PW annotations and evaluates metadata-free screen understanding on relation, clickability, retrieval, and app-classification tasks.", "topics": ["screen-understanding", "grounding", "data-generation"], "updated_at": "2022-04-14", "url": "https://arxiv.org/abs/2105.11941"}
{"aliases": [], "authors": ["Gang Li", "Gilles Baechler", "Manuel Tragut", "Yang Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d971ee06cafd46fe27a01d40bc16e3604e1d153c38ee92be970328d9935b0708", "submitted_at": "2022-01-11T17:52:40Z", "updated_at": "2022-01-13T17:53:31Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2201.04100v2"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T07:57:26Z", "role": "primary", "sha256": "3e89566cdb4521a8b3b0d99c514c3401645eda5b5c0de9720eb726b1ffe0a11f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=1511.03745%2C1602.02261%2C1809.04860%2C1811.06521%2C1811.07871%2C1812.03426%2C1901.11528%2C1909.00031%2C1912.03098%2C2003.00431%2C2004.14973%2C2006.16241%2C2007.04428%2C2010.09670%2C2010.12083%2C2012.12350%2C2104.08244%2C2105.11941%2C2106.09422%2C2106.13500%2C2107.05911%2C2108.03298%2C2108.05308%2C2110.10661%2C2201.04100%2C2201.11676%2C2203.10897%2C2204.05239%2C2205.10816%2C2206.08853%2C2206.12839%2C2206.15407%2C2208.06032%2C2208.08080%2C2209.14981%2C2210.01379%2C2210.03735%2C2210.08073%2C2210.10275%2C2210.11435%2C2211.07912%2C2211.09527%2C2212.13326%2C2301.10677%2C2301.13779%2C2302.12173%2C2303.00001%2C2303.11366%2C2303.12027%2C2304.03442&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:11:47Z", "role": "primary", "sha256": "30f2d08b6bf51489cf780365753d9e9e945156951014ae76501b5796eaacf83c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/abs/2201.04100v2"}], "id": "arxiv:2201.04100", "identifiers": {"arxiv": "2201.04100"}, "kind": "paper", "last_verified_at": "2026-10-17T09:11:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2201.04100"}], "name": "Learning to Denoise Raw Mobile UI Layouts for Improving Datasets at Scale", "notes": "Reviewed against the exact-version official arXiv abstract page; its normalized abstract SHA-256 matches the retained bibliographic snapshot. The original API response body was not locally available for this review. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2022-01-11", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces CLAY, a pipeline that repairs noisy mobile UI layouts by removing nodes without valid visual counterparts and assigning semantic element types, supported by human-annotated layouts for evaluation.", "topics": ["screen-understanding", "data-generation"], "updated_at": "2022-01-13", "url": "https://arxiv.org/abs/2201.04100"}
{"aliases": [], "authors": ["Srinivas Sunkara", "Maria Wang", "Lijuan Liu", "Gilles Baechler", "Yu-Chung Hsiao", "Jindong", "Chen", "Abhanshu Sharma", "James Stout"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "24d037f1abeb7708852ac5517ce02cc46622e359d9feab407f188c0657cc72f4", "submitted_at": "2022-10-06T03:48:54Z", "updated_at": "2022-10-06T03:48:54Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2210.02663v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=1802.08802%2C1808.09132%2C2005.03776%2C2101.09465%2C2103.16057%2C2104.08560%2C2105.13231%2C2107.03374%2C2107.13731%2C2108.03353%2C2202.02312%2C2202.08137%2C2204.01691%2C2205.11029%2C2207.01206%2C2209.08199%2C2209.08655%2C2209.14927%2C2209.15099%2C2210.02663%2C2210.03629%2C2210.03945%2C2211.07615%2C2301.10165%2C2301.13280%2C2302.04761%2C2303.07280%2C2303.08774%2C2303.17491%2C2303.17580%2C2304.04370%2C2304.07061%2C2305.04091%2C2305.05662%2C2305.08144%2C2305.10601%2C2305.11854%2C2305.12487%2C2305.14257%2C2305.14325%2C2305.16504%2C2305.16653%2C2305.19308%2C2306.00245%2C2306.06070%2C2306.07863%2C2307.10088%2C2307.12856%2C2307.13854%2C2308.01990&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?id_list=1802.08802%2C1808.09132%2C2005.03776%2C2101.09465%2C2103.16057%2C2104.08560%2C2105.13231%2C2107.03374%2C2107.13731%2C2108.03353%2C2202.02312%2C2202.08137%2C2204.01691%2C2205.11029%2C2207.01206%2C2209.08199%2C2209.08655%2C2209.14927%2C2209.15099%2C2210.02663%2C2210.03629%2C2210.03945%2C2211.07615%2C2301.10165%2C2301.13280%2C2302.04761%2C2303.07280%2C2303.08774%2C2303.17491%2C2303.17580%2C2304.04370%2C2304.07061%2C2305.04091%2C2305.05662%2C2305.08144%2C2305.10601%2C2305.11854%2C2305.12487%2C2305.14257%2C2305.14325%2C2305.16504%2C2305.16653%2C2305.19308%2C2306.00245%2C2306.06070%2C2306.07863%2C2307.10088%2C2307.12856%2C2307.13854%2C2308.01990&max_results=50"}], "id": "arxiv:2210.02663", "identifiers": {"arxiv": "2210.02663"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:20Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2210.02663"}], "name": "Towards Better Semantic Understanding of Mobile Interfaces", "notes": "Reviewed against the retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2022-10-06", "relations": [], "review_status": "curated", "scope": "core", "summary": "Augments RICO screens and view hierarchies with human labels for icon shape and meaning and for links between UI elements and text labels. It compares image-only and multimodal models on semantic understanding of previously unseen apps.", "topics": ["screen-understanding", "grounding", "data-generation"], "updated_at": "2022-10-06", "url": "https://arxiv.org/abs/2210.02663"}
{"aliases": [], "authors": ["Kenton Lee", "Mandar Joshi", "Iulia Turc", "Hexiang Hu", "Fangyu Liu", "Julian Eisenschlos", "Urvashi Khandelwal", "Peter Shaw", "Ming-Wei Chang", "Kristina Toutanova"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "7ebb243c4690aba5f1a92b0d1e050399560998a7f304ea483dae2f451cf51ec0", "submitted_at": "2022-10-07T06:42:06Z", "updated_at": "2023-06-15T21:34:23Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2210.03347v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-18T04:21:52Z", "role": "primary", "sha256": "c194d15b7e422aae48b4e05216a1f1d84471b6666718b9d3dd4c9d250c356e77", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2210.03347%2C2403.04984%2C2405.07090%2C2406.06282%2C2409.00088%2C2409.10811%2C2411.10943%2C2503.09241%2C2503.20279%2C2503.23350%2C2505.05440%2C2505.10593%2C2505.16422%2C2505.21279%2C2505.24878%2C2506.02153%2C2506.02839%2C2506.05079%2C2506.07976%2C2508.02046%2C2511.07885%2C2511.11248%2C2512.21720%2C2603.20351%2C2604.19299&max_results=25"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:21:52Z", "role": "primary", "sha256": "c194d15b7e422aae48b4e05216a1f1d84471b6666718b9d3dd4c9d250c356e77", "supports": ["name", "kind", "scope", "environments", "topics", "summary", "notes"], "url": "https://export.arxiv.org/api/query?id_list=2210.03347%2C2403.04984%2C2405.07090%2C2406.06282%2C2409.00088%2C2409.10811%2C2411.10943%2C2503.09241%2C2503.20279%2C2503.23350%2C2505.05440%2C2505.10593%2C2505.16422%2C2505.21279%2C2505.24878%2C2506.02153%2C2506.02839%2C2506.05079%2C2506.07976%2C2508.02046%2C2511.07885%2C2511.11248%2C2512.21720%2C2603.20351%2C2604.19299&max_results=25"}], "id": "arxiv:2210.03347", "identifiers": {"arxiv": "2210.03347"}, "kind": "paper", "last_verified_at": "2026-10-18T04:21:52Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2210.03347"}], "name": "Pix2Struct: Screenshot Parsing as Pretraining for Visual Language Understanding", "notes": "Reviewed against the complete official arXiv abstract. Experimental results and linked artifact availability have not been independently reproduced. Adjacent visual-language pretraining: webpage screenshot parsing and explicit UI evaluation support interface understanding; the paper does not establish an autonomous GUI policy.", "published_at": "2022-10-07", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "Pix2Struct pretrains an image-to-text model by reconstructing simplified HTML from masked webpage screenshots. Its variable-resolution visual representation is evaluated across several domains, including user-interface understanding.", "topics": ["screen-understanding", "data-generation"], "updated_at": "2023-06-15", "url": "https://arxiv.org/abs/2210.03347"}
{"aliases": [], "authors": ["Izzeddin Gur", "Ofir Nachum", "Yingjie Miao", "Mustafa Safdari", "Austin Huang", "Aakanksha Chowdhery", "Sharan Narang", "Noah Fiedel", "Aleksandra Faust"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "9b9089164b5c784735d8915596fe902840e01c8215f507caed0e10d4361d13ec", "submitted_at": "2022-10-08T07:27:17Z", "updated_at": "2023-05-19T17:42:56Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2210.03945v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=1802.08802%2C1808.09132%2C2005.03776%2C2101.09465%2C2103.16057%2C2104.08560%2C2105.13231%2C2107.03374%2C2107.13731%2C2108.03353%2C2202.02312%2C2202.08137%2C2204.01691%2C2205.11029%2C2207.01206%2C2209.08199%2C2209.08655%2C2209.14927%2C2209.15099%2C2210.02663%2C2210.03629%2C2210.03945%2C2211.07615%2C2301.10165%2C2301.13280%2C2302.04761%2C2303.07280%2C2303.08774%2C2303.17491%2C2303.17580%2C2304.04370%2C2304.07061%2C2305.04091%2C2305.05662%2C2305.08144%2C2305.10601%2C2305.11854%2C2305.12487%2C2305.14257%2C2305.14325%2C2305.16504%2C2305.16653%2C2305.19308%2C2306.00245%2C2306.06070%2C2306.07863%2C2307.10088%2C2307.12856%2C2307.13854%2C2308.01990&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?id_list=1802.08802%2C1808.09132%2C2005.03776%2C2101.09465%2C2103.16057%2C2104.08560%2C2105.13231%2C2107.03374%2C2107.13731%2C2108.03353%2C2202.02312%2C2202.08137%2C2204.01691%2C2205.11029%2C2207.01206%2C2209.08199%2C2209.08655%2C2209.14927%2C2209.15099%2C2210.02663%2C2210.03629%2C2210.03945%2C2211.07615%2C2301.10165%2C2301.13280%2C2302.04761%2C2303.07280%2C2303.08774%2C2303.17491%2C2303.17580%2C2304.04370%2C2304.07061%2C2305.04091%2C2305.05662%2C2305.08144%2C2305.10601%2C2305.11854%2C2305.12487%2C2305.14257%2C2305.14325%2C2305.16504%2C2305.16653%2C2305.19308%2C2306.00245%2C2306.06070%2C2306.07863%2C2307.10088%2C2307.12856%2C2307.13854%2C2308.01990&max_results=50"}], "id": "arxiv:2210.03945", "identifiers": {"arxiv": "2210.03945"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:20Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2210.03945"}], "name": "Understanding HTML with Large Language Models", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2022-10-08", "relations": [], "review_status": "curated", "scope": "core", "summary": "Studies language models for HTML element classification, page description and autonomous web navigation. Fine-tuned models are evaluated on MiniWoB, and a distilled, automatically labeled CommonCrawl HTML corpus supports the training approach.", "topics": ["screen-understanding", "web-navigation", "data-generation"], "updated_at": "2023-05-19", "url": "https://arxiv.org/abs/2210.03945"}
{"aliases": [], "authors": ["Pratyay Banerjee", "Shweti Mahajan", "Kushal Arora", "Chitta Baral", "Oriana Riva"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "bb647a96c7f8263786e85a5061c8c5284cb92ed0b2ebc4c405e1243b2ca3fa2b", "submitted_at": "2023-01-23T09:05:49Z", "updated_at": "2023-01-23T09:05:49Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2301.10165v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=1802.08802%2C1808.09132%2C2005.03776%2C2101.09465%2C2103.16057%2C2104.08560%2C2105.13231%2C2107.03374%2C2107.13731%2C2108.03353%2C2202.02312%2C2202.08137%2C2204.01691%2C2205.11029%2C2207.01206%2C2209.08199%2C2209.08655%2C2209.14927%2C2209.15099%2C2210.02663%2C2210.03629%2C2210.03945%2C2211.07615%2C2301.10165%2C2301.13280%2C2302.04761%2C2303.07280%2C2303.08774%2C2303.17491%2C2303.17580%2C2304.04370%2C2304.07061%2C2305.04091%2C2305.05662%2C2305.08144%2C2305.10601%2C2305.11854%2C2305.12487%2C2305.14257%2C2305.14325%2C2305.16504%2C2305.16653%2C2305.19308%2C2306.00245%2C2306.06070%2C2306.07863%2C2307.10088%2C2307.12856%2C2307.13854%2C2308.01990&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?id_list=1802.08802%2C1808.09132%2C2005.03776%2C2101.09465%2C2103.16057%2C2104.08560%2C2105.13231%2C2107.03374%2C2107.13731%2C2108.03353%2C2202.02312%2C2202.08137%2C2204.01691%2C2205.11029%2C2207.01206%2C2209.08199%2C2209.08655%2C2209.14927%2C2209.15099%2C2210.02663%2C2210.03629%2C2210.03945%2C2211.07615%2C2301.10165%2C2301.13280%2C2302.04761%2C2303.07280%2C2303.08774%2C2303.17491%2C2303.17580%2C2304.04370%2C2304.07061%2C2305.04091%2C2305.05662%2C2305.08144%2C2305.10601%2C2305.11854%2C2305.12487%2C2305.14257%2C2305.14325%2C2305.16504%2C2305.16653%2C2305.19308%2C2306.00245%2C2306.06070%2C2306.07863%2C2307.10088%2C2307.12856%2C2307.13854%2C2308.01990&max_results=50"}], "id": "arxiv:2301.10165", "identifiers": {"arxiv": "2301.10165"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:20Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2301.10165"}], "name": "Lexi: Self-Supervised Learning of the UI Language", "notes": "Reviewed against the retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2023-01-23", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces Lexi and UICaption to learn UI image-text representations from descriptions of interface functionality. Evaluation covers UI action entailment, instruction-based retrieval, referring-expression grounding, and UI entity recognition.", "topics": ["screen-understanding", "grounding", "data-generation"], "updated_at": "2023-01-23", "url": "https://arxiv.org/abs/2301.10165"}
{"aliases": [], "authors": ["Jason Wu", "Siyan Wang", "Siman Shen", "Yi-Hao Peng", "Jeffrey Nichols", "Jeffrey P. Bigham"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "0cc3b91aa85bc8c7d88a878c7b24926943eea0bcc92e25f517915242f2a82a39", "submitted_at": "2023-01-30T20:47:12Z", "updated_at": "2023-01-30T20:47:12Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2301.13280v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=1802.08802%2C1808.09132%2C2005.03776%2C2101.09465%2C2103.16057%2C2104.08560%2C2105.13231%2C2107.03374%2C2107.13731%2C2108.03353%2C2202.02312%2C2202.08137%2C2204.01691%2C2205.11029%2C2207.01206%2C2209.08199%2C2209.08655%2C2209.14927%2C2209.15099%2C2210.02663%2C2210.03629%2C2210.03945%2C2211.07615%2C2301.10165%2C2301.13280%2C2302.04761%2C2303.07280%2C2303.08774%2C2303.17491%2C2303.17580%2C2304.04370%2C2304.07061%2C2305.04091%2C2305.05662%2C2305.08144%2C2305.10601%2C2305.11854%2C2305.12487%2C2305.14257%2C2305.14325%2C2305.16504%2C2305.16653%2C2305.19308%2C2306.00245%2C2306.06070%2C2306.07863%2C2307.10088%2C2307.12856%2C2307.13854%2C2308.01990&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=1802.08802%2C1808.09132%2C2005.03776%2C2101.09465%2C2103.16057%2C2104.08560%2C2105.13231%2C2107.03374%2C2107.13731%2C2108.03353%2C2202.02312%2C2202.08137%2C2204.01691%2C2205.11029%2C2207.01206%2C2209.08199%2C2209.08655%2C2209.14927%2C2209.15099%2C2210.02663%2C2210.03629%2C2210.03945%2C2211.07615%2C2301.10165%2C2301.13280%2C2302.04761%2C2303.07280%2C2303.08774%2C2303.17491%2C2303.17580%2C2304.04370%2C2304.07061%2C2305.04091%2C2305.05662%2C2305.08144%2C2305.10601%2C2305.11854%2C2305.12487%2C2305.14257%2C2305.14325%2C2305.16504%2C2305.16653%2C2305.19308%2C2306.00245%2C2306.06070%2C2306.07863%2C2307.10088%2C2307.12856%2C2307.13854%2C2308.01990&max_results=50"}], "id": "arxiv:2301.13280", "identifiers": {"arxiv": "2301.13280"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:20Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2301.13280"}], "name": "WebUI: A Dataset for Enhancing Visual UI Understanding with Web Semantics", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2023-01-30", "relations": [], "review_status": "curated", "scope": "core", "summary": "Collects rendered webpages and structural metadata for visual UI modeling, studying how web semantics transfer to interface-understanding tasks.", "topics": ["screen-understanding", "data-generation"], "updated_at": "2023-01-30", "url": "https://arxiv.org/abs/2301.13280"}
{"aliases": [], "authors": ["Xiang Deng", "Yu Gu", "Boyuan Zheng", "Shijie Chen", "Samuel Stevens", "Boshi Wang", "Huan Sun", "Yu Su"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "de055dfa452130f3576839c99c007053c62a0c96c9bfbcdd79ecad1a9f74f830", "submitted_at": "2023-06-09T17:44:31Z", "updated_at": "2023-12-09T05:57:46Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2306.06070v3"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=1802.08802%2C1808.09132%2C2005.03776%2C2101.09465%2C2103.16057%2C2104.08560%2C2105.13231%2C2107.03374%2C2107.13731%2C2108.03353%2C2202.02312%2C2202.08137%2C2204.01691%2C2205.11029%2C2207.01206%2C2209.08199%2C2209.08655%2C2209.14927%2C2209.15099%2C2210.02663%2C2210.03629%2C2210.03945%2C2211.07615%2C2301.10165%2C2301.13280%2C2302.04761%2C2303.07280%2C2303.08774%2C2303.17491%2C2303.17580%2C2304.04370%2C2304.07061%2C2305.04091%2C2305.05662%2C2305.08144%2C2305.10601%2C2305.11854%2C2305.12487%2C2305.14257%2C2305.14325%2C2305.16504%2C2305.16653%2C2305.19308%2C2306.00245%2C2306.06070%2C2306.07863%2C2307.10088%2C2307.12856%2C2307.13854%2C2308.01990&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=1802.08802%2C1808.09132%2C2005.03776%2C2101.09465%2C2103.16057%2C2104.08560%2C2105.13231%2C2107.03374%2C2107.13731%2C2108.03353%2C2202.02312%2C2202.08137%2C2204.01691%2C2205.11029%2C2207.01206%2C2209.08199%2C2209.08655%2C2209.14927%2C2209.15099%2C2210.02663%2C2210.03629%2C2210.03945%2C2211.07615%2C2301.10165%2C2301.13280%2C2302.04761%2C2303.07280%2C2303.08774%2C2303.17491%2C2303.17580%2C2304.04370%2C2304.07061%2C2305.04091%2C2305.05662%2C2305.08144%2C2305.10601%2C2305.11854%2C2305.12487%2C2305.14257%2C2305.14325%2C2305.16504%2C2305.16653%2C2305.19308%2C2306.00245%2C2306.06070%2C2306.07863%2C2307.10088%2C2307.12856%2C2307.13854%2C2308.01990&max_results=50"}], "id": "arxiv:2306.06070", "identifiers": {"arxiv": "2306.06070"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:20Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2306.06070"}], "name": "Mind2Web: Towards a Generalist Agent for the Web", "notes": "Editorial summary based on the primary paper abstract; experiments and released artifacts have not been independently reproduced.", "published_at": "2023-06-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces a web-task dataset spanning diverse websites and evaluates generalization to unseen sites and domains.", "topics": ["data-generation", "web-navigation", "evaluation"], "updated_at": "2023-12-09", "url": "https://arxiv.org/abs/2306.06070"}
{"aliases": [], "authors": ["Christopher Rawles", "Alice Li", "Daniel Rodriguez", "Oriana Riva", "Timothy Lillicrap"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "def4616f7cb13fe839e500c4b3171eafbe37b155b5d14301137806d6a5cde939", "submitted_at": "2023-07-19T15:57:24Z", "updated_at": "2023-10-27T14:24:31Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2307.10088v2"}, "publication": {"authors": ["Christopher Rawles", "Alice Li", "Daniel Rodriguez", "Oriana Riva", "Timothy Lillicrap"], "date_basis": "Explicit publisher citation_publication_date: 2023-12-15. Preserve supplied precision; do not substitute conference event dates.", "date_parts": [2023, 12, 15], "doi": "10.52202/075280-2609", "event_dates": null, "evidence_urls": ["https://proceedings.neurips.cc/paper_files/paper/2023/hash/bbbb6308b402fe909c39dd29950c32e0-Abstract-Datasets_and_Benchmarks.html"], "identity_basis": "Same work: normalized title, ordered author correspondence and the specific contribution in the complete publisher abstract were reviewed against the retained arXiv identity.", "issue": null, "notes": "Publisher proceedings edition; retained separately from the arXiv snapshot.", "pages": "59708-59728", "pdf_url": "https://proceedings.neurips.cc/paper_files/paper/2023/file/bbbb6308b402fe909c39dd29950c32e0-Paper-Datasets_and_Benchmarks.pdf", "publisher": null, "series": null, "title": "AndroidInTheWild: A Large-Scale Dataset For Android Device Control", "type": "conference-paper", "url": "https://proceedings.neurips.cc/paper_files/paper/2023/hash/bbbb6308b402fe909c39dd29950c32e0-Abstract-Datasets_and_Benchmarks.html", "venue": "Advances in Neural Information Processing Systems", "volume": "36"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=1802.08802%2C1808.09132%2C2005.03776%2C2101.09465%2C2103.16057%2C2104.08560%2C2105.13231%2C2107.03374%2C2107.13731%2C2108.03353%2C2202.02312%2C2202.08137%2C2204.01691%2C2205.11029%2C2207.01206%2C2209.08199%2C2209.08655%2C2209.14927%2C2209.15099%2C2210.02663%2C2210.03629%2C2210.03945%2C2211.07615%2C2301.10165%2C2301.13280%2C2302.04761%2C2303.07280%2C2303.08774%2C2303.17491%2C2303.17580%2C2304.04370%2C2304.07061%2C2305.04091%2C2305.05662%2C2305.08144%2C2305.10601%2C2305.11854%2C2305.12487%2C2305.14257%2C2305.14325%2C2305.16504%2C2305.16653%2C2305.19308%2C2306.00245%2C2306.06070%2C2306.07863%2C2307.10088%2C2307.12856%2C2307.13854%2C2308.01990&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:10:02Z", "role": "primary", "sha256": "7fa75eae1a17878ef28414f64e1c46cc83539008f2644331b762248ec96efce5", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue", "details.publication.volume"], "url": "https://proceedings.neurips.cc/paper_files/paper/2023/hash/bbbb6308b402fe909c39dd29950c32e0-Abstract-Datasets_and_Benchmarks.html"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=1802.08802%2C1808.09132%2C2005.03776%2C2101.09465%2C2103.16057%2C2104.08560%2C2105.13231%2C2107.03374%2C2107.13731%2C2108.03353%2C2202.02312%2C2202.08137%2C2204.01691%2C2205.11029%2C2207.01206%2C2209.08199%2C2209.08655%2C2209.14927%2C2209.15099%2C2210.02663%2C2210.03629%2C2210.03945%2C2211.07615%2C2301.10165%2C2301.13280%2C2302.04761%2C2303.07280%2C2303.08774%2C2303.17491%2C2303.17580%2C2304.04370%2C2304.07061%2C2305.04091%2C2305.05662%2C2305.08144%2C2305.10601%2C2305.11854%2C2305.12487%2C2305.14257%2C2305.14325%2C2305.16504%2C2305.16653%2C2305.19308%2C2306.00245%2C2306.06070%2C2306.07863%2C2307.10088%2C2307.12856%2C2307.13854%2C2308.01990&max_results=50"}], "id": "arxiv:2307.10088", "identifiers": {"arxiv": "2307.10088"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:20Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2307.10088"}, {"rel": "publication", "url": "https://proceedings.neurips.cc/paper_files/paper/2023/hash/bbbb6308b402fe909c39dd29950c32e0-Abstract-Datasets_and_Benchmarks.html"}, {"rel": "publication_pdf", "url": "https://proceedings.neurips.cc/paper_files/paper/2023/file/bbbb6308b402fe909c39dd29950c32e0-Paper-Datasets_and_Benchmarks.pdf"}], "name": "Android in the Wild: A Large-Scale Dataset for Android Device Control", "notes": "Annotation reviewed against the primary abstract. Original dataset scale and experimental results are not independently reproduced.", "published_at": "2023-07-19", "relations": [], "review_status": "curated", "scope": "core", "summary": "Collects Android interaction demonstrations with screens, actions, and instructions across device and platform variations for robustness evaluation.", "topics": ["data-generation", "mobile-navigation", "evaluation"], "updated_at": "2023-10-27", "url": "https://arxiv.org/abs/2307.10088"}
{"aliases": [], "authors": ["Yue Jiang", "Eldon Schoop", "Amanda Swearngin", "Jeffrey Nichols"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a49e879891017071fce4df0ff4b07bd9e8b5a6e14385af565549aadbde1bc1cf", "submitted_at": "2023-10-07T16:32:34Z", "updated_at": "2023-10-07T16:32:34Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2310.04869v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:24Z", "role": "primary", "sha256": "3aac6e9748769efff3c475ede0da192776daa03431d231b5f67e2aa50f8a3be9", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2308.03688%2C2308.15272%2C2309.08172%2C2309.11436%2C2310.04716%2C2310.04869%2C2310.05126%2C2310.06770%2C2310.08740%2C2310.10634%2C2310.11441%2C2310.16042%2C2310.18331%2C2311.01767%2C2311.07562%2C2311.09278%2C2311.12983%2C2311.18751%2C2312.03003%2C2312.06677%2C2312.08914%2C2312.11671%2C2312.13108%2C2312.13771%2C2312.15820%2C2312.16062%2C2401.01614%2C2401.04124%2C2401.07781%2C2401.10935%2C2401.13178%2C2401.13649%2C2401.13919%2C2401.14268%2C2401.16158%2C2402.01030%2C2402.04476%2C2402.04615%2C2402.05930%2C2402.06596%2C2402.07456%2C2402.07939%2C2402.07945%2C2402.08567%2C2402.11941%2C2402.14073%2C2402.15057%2C2402.16965%2C2402.17553%2C2403.02713&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:24Z", "role": "primary", "sha256": "3aac6e9748769efff3c475ede0da192776daa03431d231b5f67e2aa50f8a3be9", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2308.03688%2C2308.15272%2C2309.08172%2C2309.11436%2C2310.04716%2C2310.04869%2C2310.05126%2C2310.06770%2C2310.08740%2C2310.10634%2C2310.11441%2C2310.16042%2C2310.18331%2C2311.01767%2C2311.07562%2C2311.09278%2C2311.12983%2C2311.18751%2C2312.03003%2C2312.06677%2C2312.08914%2C2312.11671%2C2312.13108%2C2312.13771%2C2312.15820%2C2312.16062%2C2401.01614%2C2401.04124%2C2401.07781%2C2401.10935%2C2401.13178%2C2401.13649%2C2401.13919%2C2401.14268%2C2401.16158%2C2402.01030%2C2402.04476%2C2402.04615%2C2402.05930%2C2402.06596%2C2402.07456%2C2402.07939%2C2402.07945%2C2402.08567%2C2402.11941%2C2402.14073%2C2402.15057%2C2402.16965%2C2402.17553%2C2403.02713&max_results=50"}], "id": "arxiv:2310.04869", "identifiers": {"arxiv": "2310.04869"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:24Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2310.04869"}], "name": "ILuvUI: Instruction-tuned LangUage-Vision modeling of UIs from Machine Conversations", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2023-10-07", "relations": [], "review_status": "curated", "scope": "core", "summary": "ILuvUI generates 335,000 conversational training examples paired with UI screenshots, covering questions, descriptions, and plans. The resulting vision-language model is assessed on element detection, response quality, and examples of multi-step UI navigation.", "topics": ["data-generation", "screen-understanding", "grounding", "planning"], "updated_at": "2023-10-07", "url": "https://arxiv.org/abs/2310.04869"}
{"aliases": [], "authors": ["Qi Chen", "Dileepa Pitawela", "Chongyang Zhao", "Gengze Zhou", "Hsiang-Ting Chen", "Qi Wu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "47dedddced8f06bf436ab5a101d1ec4ecf4380c710e457dc7a35fc86d2e1f230", "submitted_at": "2023-12-25T22:13:26Z", "updated_at": "2023-12-25T22:13:26Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2312.15820v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:24Z", "role": "primary", "sha256": "3aac6e9748769efff3c475ede0da192776daa03431d231b5f67e2aa50f8a3be9", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2308.03688%2C2308.15272%2C2309.08172%2C2309.11436%2C2310.04716%2C2310.04869%2C2310.05126%2C2310.06770%2C2310.08740%2C2310.10634%2C2310.11441%2C2310.16042%2C2310.18331%2C2311.01767%2C2311.07562%2C2311.09278%2C2311.12983%2C2311.18751%2C2312.03003%2C2312.06677%2C2312.08914%2C2312.11671%2C2312.13108%2C2312.13771%2C2312.15820%2C2312.16062%2C2401.01614%2C2401.04124%2C2401.07781%2C2401.10935%2C2401.13178%2C2401.13649%2C2401.13919%2C2401.14268%2C2401.16158%2C2402.01030%2C2402.04476%2C2402.04615%2C2402.05930%2C2402.06596%2C2402.07456%2C2402.07939%2C2402.07945%2C2402.08567%2C2402.11941%2C2402.14073%2C2402.15057%2C2402.16965%2C2402.17553%2C2403.02713&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:24Z", "role": "primary", "sha256": "3aac6e9748769efff3c475ede0da192776daa03431d231b5f67e2aa50f8a3be9", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2308.03688%2C2308.15272%2C2309.08172%2C2309.11436%2C2310.04716%2C2310.04869%2C2310.05126%2C2310.06770%2C2310.08740%2C2310.10634%2C2310.11441%2C2310.16042%2C2310.18331%2C2311.01767%2C2311.07562%2C2311.09278%2C2311.12983%2C2311.18751%2C2312.03003%2C2312.06677%2C2312.08914%2C2312.11671%2C2312.13108%2C2312.13771%2C2312.15820%2C2312.16062%2C2401.01614%2C2401.04124%2C2401.07781%2C2401.10935%2C2401.13178%2C2401.13649%2C2401.13919%2C2401.14268%2C2401.16158%2C2402.01030%2C2402.04476%2C2402.04615%2C2402.05930%2C2402.06596%2C2402.07456%2C2402.07939%2C2402.07945%2C2402.08567%2C2402.11941%2C2402.14073%2C2402.15057%2C2402.16965%2C2402.17553%2C2403.02713&max_results=50"}], "id": "arxiv:2312.15820", "identifiers": {"arxiv": "2312.15820"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:24Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2312.15820"}], "name": "WebVLN: Vision-and-Language Navigation on Websites", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2023-12-25", "relations": [], "review_status": "curated", "scope": "core", "summary": "WebVLN trains website navigation from question-based instructions using both rendered visual information and underlying HTML. It introduces WebVLN-v1 and a website-aware navigation network to study this digital variant of vision-and-language navigation.", "topics": ["web-navigation", "data-generation", "evaluation"], "updated_at": "2023-12-25", "url": "https://arxiv.org/abs/2312.15820"}
{"aliases": [], "authors": ["Kanzhi Cheng", "Qiushi Sun", "Yougang Chu", "Fangzhi Xu", "Yantao Li", "Jianbing Zhang", "Zhiyong Wu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "c92ec9323eec9362aa07984f9fb1cbfbe98ecae426a9f8adc9d436328a87aed3", "submitted_at": "2024-01-17T08:10:35Z", "updated_at": "2024-02-23T04:36:51Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2401.10935v2"}, "publication": {"authors": ["Kanzhi Cheng", "Qiushi Sun", "Yougang Chu", "Fangzhi Xu", "Yantao Li", "Jianbing Zhang", "Zhiyong Wu"], "date_basis": "Explicit publisher metadata: 2024/8; precision=month.", "date_parts": [2024, 8], "doi": "10.18653/v1/2024.acl-long.505", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2024.acl-long.505/", "https://aclanthology.org/2024.acl-long.505.pdf"], "identity_basis": "Same work: exact normalized landing title, corresponding author list, and matching specific contribution. Publisher and arXiv versions remain distinct.", "issue": null, "notes": "The ACL landing metadata spells the fifth author Li YanTao. The formal PDF title page spells Yantao Li; the publication author field uses the PDF. The retained arXiv author list is untouched. Publisher-hosted proceedings record; distinct from the retained arXiv preprint snapshot.", "pages": "9313-9332", "pdf_url": "https://aclanthology.org/2024.acl-long.505.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "SeeClick: Harnessing GUI Grounding for Advanced Visual GUI Agents", "type": "conference-paper", "url": "https://aclanthology.org/2024.acl-long.505/", "venue": "Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)", "volume": null}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:53:48+00:00", "role": "primary", "sha256": "62d56e417487461c8b34da15167eafd5f14f58092e4269222dc5c1c9048731a9", "supports": ["details.publication.date_basis", "details.publication.date_parts", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.publisher", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue"], "url": "https://aclanthology.org/2024.acl-long.505/"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:56:04+00:00", "role": "primary", "sha256": "8e1f6809e35751d762714286c55e79e81bae071608f90ef561d804aa59d7e406", "supports": ["details.publication.authors", "details.publication.identity_basis", "details.publication.notes"], "url": "https://aclanthology.org/2024.acl-long.505.pdf"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2401.10935", "identifiers": {"arxiv": "2401.10935"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:58Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2401.10935"}, {"rel": "publication", "url": "https://aclanthology.org/2024.acl-long.505/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2024.acl-long.505.pdf"}], "name": "SeeClick: Harnessing GUI Grounding for Advanced Visual GUI Agents", "notes": "Summary and topic tags reviewed against the primary abstract. Reported experiments, task counts, and artifact availability are not independently reproduced.", "published_at": "2024-01-17", "relations": [], "review_status": "curated", "scope": "core", "summary": "Trains a screenshot-driven GUI agent with grounding pretraining and introduces ScreenSpot for evaluating interface target localization.", "topics": ["grounding", "data-generation"], "updated_at": "2024-02-23", "url": "https://arxiv.org/abs/2401.10935"}
{"aliases": [], "authors": ["Gilles Baechler", "Srinivas Sunkara", "Maria Wang", "Fedir Zubach", "Hassan Mansoor", "Vincent Etter", "Victor Cărbune", "Jason Lin", "Jindong Chen", "Abhanshu Sharma"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "b293ab33fe0f3607eccc2b99cbf06fb459792444433a4179bb65e081f33e426c", "submitted_at": "2024-02-07T06:42:33Z", "updated_at": "2024-07-04T07:08:15Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2402.04615v3"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:24Z", "role": "primary", "sha256": "3aac6e9748769efff3c475ede0da192776daa03431d231b5f67e2aa50f8a3be9", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2308.03688%2C2308.15272%2C2309.08172%2C2309.11436%2C2310.04716%2C2310.04869%2C2310.05126%2C2310.06770%2C2310.08740%2C2310.10634%2C2310.11441%2C2310.16042%2C2310.18331%2C2311.01767%2C2311.07562%2C2311.09278%2C2311.12983%2C2311.18751%2C2312.03003%2C2312.06677%2C2312.08914%2C2312.11671%2C2312.13108%2C2312.13771%2C2312.15820%2C2312.16062%2C2401.01614%2C2401.04124%2C2401.07781%2C2401.10935%2C2401.13178%2C2401.13649%2C2401.13919%2C2401.14268%2C2401.16158%2C2402.01030%2C2402.04476%2C2402.04615%2C2402.05930%2C2402.06596%2C2402.07456%2C2402.07939%2C2402.07945%2C2402.08567%2C2402.11941%2C2402.14073%2C2402.15057%2C2402.16965%2C2402.17553%2C2403.02713&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:24Z", "role": "primary", "sha256": "3aac6e9748769efff3c475ede0da192776daa03431d231b5f67e2aa50f8a3be9", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2308.03688%2C2308.15272%2C2309.08172%2C2309.11436%2C2310.04716%2C2310.04869%2C2310.05126%2C2310.06770%2C2310.08740%2C2310.10634%2C2310.11441%2C2310.16042%2C2310.18331%2C2311.01767%2C2311.07562%2C2311.09278%2C2311.12983%2C2311.18751%2C2312.03003%2C2312.06677%2C2312.08914%2C2312.11671%2C2312.13108%2C2312.13771%2C2312.15820%2C2312.16062%2C2401.01614%2C2401.04124%2C2401.07781%2C2401.10935%2C2401.13178%2C2401.13649%2C2401.13919%2C2401.14268%2C2401.16158%2C2402.01030%2C2402.04476%2C2402.04615%2C2402.05930%2C2402.06596%2C2402.07456%2C2402.07939%2C2402.07945%2C2402.08567%2C2402.11941%2C2402.14073%2C2402.15057%2C2402.16965%2C2402.17553%2C2403.02713&max_results=50"}], "id": "arxiv:2402.04615", "identifiers": {"arxiv": "2402.04615"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:24Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2402.04615"}], "name": "ScreenAI: A Vision-Language Model for UI and Infographics Understanding", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-02-07", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces ScreenAI and screen-annotation training that supports UI question answering, navigation, and summarization data generation.", "topics": ["screen-understanding", "data-generation"], "updated_at": "2024-07-04", "url": "https://arxiv.org/abs/2402.04615"}
{"aliases": [], "authors": ["Xing Han Lù", "Zdeněk Kasner", "Siva Reddy"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6a7c1a1f260f9287e8999525e1adc5534a70f7b96271d469973fceee1ac23609", "submitted_at": "2024-02-08T18:58:02Z", "updated_at": "2024-09-10T22:22:43Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2402.05930v2"}, "publication": {"authors": ["Xing Han Lù", "Zdeněk Kasner", "Siva Reddy"], "date_basis": "Explicit publisher metadata: 2024/07/08; precision=day.", "date_parts": [2024, 7, 8], "doi": null, "event_dates": "21--27 Jul 2024", "evidence_urls": ["https://proceedings.mlr.press/v235/lu24e.html", "https://raw.githubusercontent.com/mlresearch/v235/main/assets/lu24e/lu24e.pdf"], "identity_basis": "Same work: exact normalized landing title, corresponding author list, and matching specific contribution. Publisher and arXiv versions remain distinct.", "issue": null, "notes": "The PMLR landing renders Xing Han Lu; the paper title page has Xing Han Lù. Publication authors use the PDF. The publisher issue date comes from citation_publication_date; the conference date range is retained separately and does not replace it. Publisher-hosted proceedings record; distinct from the retained arXiv preprint snapshot.", "pages": "33007-33056", "pdf_url": "https://raw.githubusercontent.com/mlresearch/v235/main/assets/lu24e/lu24e.pdf", "publisher": "PMLR", "series": "Proceedings of Machine Learning Research", "title": "WebLINX: Real-World Website Navigation with Multi-Turn Dialogue", "type": "conference-paper", "url": "https://proceedings.mlr.press/v235/lu24e.html", "venue": "Proceedings of the 41st International Conference on Machine Learning", "volume": "235"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:24Z", "role": "primary", "sha256": "3aac6e9748769efff3c475ede0da192776daa03431d231b5f67e2aa50f8a3be9", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2308.03688%2C2308.15272%2C2309.08172%2C2309.11436%2C2310.04716%2C2310.04869%2C2310.05126%2C2310.06770%2C2310.08740%2C2310.10634%2C2310.11441%2C2310.16042%2C2310.18331%2C2311.01767%2C2311.07562%2C2311.09278%2C2311.12983%2C2311.18751%2C2312.03003%2C2312.06677%2C2312.08914%2C2312.11671%2C2312.13108%2C2312.13771%2C2312.15820%2C2312.16062%2C2401.01614%2C2401.04124%2C2401.07781%2C2401.10935%2C2401.13178%2C2401.13649%2C2401.13919%2C2401.14268%2C2401.16158%2C2402.01030%2C2402.04476%2C2402.04615%2C2402.05930%2C2402.06596%2C2402.07456%2C2402.07939%2C2402.07945%2C2402.08567%2C2402.11941%2C2402.14073%2C2402.15057%2C2402.16965%2C2402.17553%2C2403.02713&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:53:47+00:00", "role": "primary", "sha256": "1b30008f833b3b06238d87d0509a10876a08153a78a06c79bc7590c30c471777", "supports": ["details.publication.date_basis", "details.publication.date_parts", "details.publication.event_dates", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.publisher", "details.publication.series", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue", "details.publication.volume"], "url": "https://proceedings.mlr.press/v235/lu24e.html"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:59:46+00:00", "role": "primary", "sha256": "c644b1e3f071879aec8ea0e741e31e84193905dd730382ada54f1f24d04298ab", "supports": ["details.publication.authors", "details.publication.identity_basis", "details.publication.notes"], "url": "https://raw.githubusercontent.com/mlresearch/v235/main/assets/lu24e/lu24e.pdf"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:24Z", "role": "primary", "sha256": "3aac6e9748769efff3c475ede0da192776daa03431d231b5f67e2aa50f8a3be9", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2308.03688%2C2308.15272%2C2309.08172%2C2309.11436%2C2310.04716%2C2310.04869%2C2310.05126%2C2310.06770%2C2310.08740%2C2310.10634%2C2310.11441%2C2310.16042%2C2310.18331%2C2311.01767%2C2311.07562%2C2311.09278%2C2311.12983%2C2311.18751%2C2312.03003%2C2312.06677%2C2312.08914%2C2312.11671%2C2312.13108%2C2312.13771%2C2312.15820%2C2312.16062%2C2401.01614%2C2401.04124%2C2401.07781%2C2401.10935%2C2401.13178%2C2401.13649%2C2401.13919%2C2401.14268%2C2401.16158%2C2402.01030%2C2402.04476%2C2402.04615%2C2402.05930%2C2402.06596%2C2402.07456%2C2402.07939%2C2402.07945%2C2402.08567%2C2402.11941%2C2402.14073%2C2402.15057%2C2402.16965%2C2402.17553%2C2403.02713&max_results=50"}], "id": "arxiv:2402.05930", "identifiers": {"arxiv": "2402.05930"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:24Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2402.05930"}, {"rel": "publication", "url": "https://proceedings.mlr.press/v235/lu24e.html"}, {"rel": "publication_pdf", "url": "https://raw.githubusercontent.com/mlresearch/v235/main/assets/lu24e/lu24e.pdf"}], "name": "WebLINX: Real-World Website Navigation with Multi-Turn Dialogue", "notes": "Summary and topic tags reviewed against the primary abstract. Reported experiments, task counts, and artifact availability are not independently reproduced.", "published_at": "2024-02-08", "relations": [], "review_status": "curated", "scope": "core", "summary": "Collects expert demonstrations of conversational web navigation and studies retrieval-based element selection for processing large pages.", "topics": ["data-generation", "web-navigation", "evaluation"], "updated_at": "2024-09-10", "url": "https://arxiv.org/abs/2402.05930"}
{"aliases": [], "authors": ["Andrés Muñoz", "Daniel Borrajo"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "aacb4d2cac31e478aa56ee25b6b39e5ba8fe164f89102e665f8fb25cf3822e7b", "submitted_at": "2024-03-15T10:26:52Z", "updated_at": "2024-08-27T18:36:12Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2403.10170v2"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:28Z", "role": "primary", "sha256": "be244bfae7f89ef9449b69d85aac44295eb7b5f7e5e1bbedd76df312696676a7", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2403.03186%2C2403.03788%2C2403.07718%2C2403.10170%2C2403.11905%2C2403.16971%2C2403.17918%2C2403.20271%2C2404.01549%2C2404.01744%2C2404.02475%2C2404.03648%2C2404.05427%2C2404.05719%2C2404.05955%2C2404.06474%2C2404.07972%2C2404.08755%2C2404.08860%2C2404.08948%2C2404.09992%2C2404.10887%2C2404.11459%2C2404.16006%2C2404.16048%2C2404.16054%2C2404.16660%2C2404.18074%2C2404.19296%2C2405.00516%2C2405.11120%2C2405.12842%2C2405.14573%2C2405.14751%2C2405.15793%2C2405.16247%2C2405.20309%2C2406.01014%2C2406.01623%2C2406.03679%2C2406.06947%2C2406.07822%2C2406.08184%2C2406.08340%2C2406.08451%2C2406.10227%2C2406.10819%2C2406.11317%2C2406.11896%2C2406.12373&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:28Z", "role": "primary", "sha256": "be244bfae7f89ef9449b69d85aac44295eb7b5f7e5e1bbedd76df312696676a7", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2403.03186%2C2403.03788%2C2403.07718%2C2403.10170%2C2403.11905%2C2403.16971%2C2403.17918%2C2403.20271%2C2404.01549%2C2404.01744%2C2404.02475%2C2404.03648%2C2404.05427%2C2404.05719%2C2404.05955%2C2404.06474%2C2404.07972%2C2404.08755%2C2404.08860%2C2404.08948%2C2404.09992%2C2404.10887%2C2404.11459%2C2404.16006%2C2404.16048%2C2404.16054%2C2404.16660%2C2404.18074%2C2404.19296%2C2405.00516%2C2405.11120%2C2405.12842%2C2405.14573%2C2405.14751%2C2405.15793%2C2405.16247%2C2405.20309%2C2406.01014%2C2406.01623%2C2406.03679%2C2406.06947%2C2406.07822%2C2406.08184%2C2406.08340%2C2406.08451%2C2406.10227%2C2406.10819%2C2406.11317%2C2406.11896%2C2406.12373&max_results=50"}], "id": "arxiv:2403.10170", "identifiers": {"arxiv": "2403.10170"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:28Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2403.10170"}], "name": "Computer User Interface Understanding. A New Dataset and a Learning Framework", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-03-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces desktop interaction videos and a framework combining synthetic examples with hierarchical contrastive learning for interface classification.", "topics": ["screen-understanding", "video", "data-generation"], "updated_at": "2024-08-27", "url": "https://arxiv.org/abs/2403.10170"}
{"aliases": [], "authors": ["Hanyu Lai", "Xiao Liu", "Iat Long Iong", "Shuntian Yao", "Yuxuan Chen", "Pengbo Shen", "Hao Yu", "Hanchen Zhang", "Xiaohan Zhang", "Yuxiao Dong", "Jie Tang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "340206da25df688ff73da617fd66a872586d1e36981c431d18331cee9b9669e0", "submitted_at": "2024-04-04T17:58:40Z", "updated_at": "2024-10-12T05:05:38Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2404.03648v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:28Z", "role": "primary", "sha256": "be244bfae7f89ef9449b69d85aac44295eb7b5f7e5e1bbedd76df312696676a7", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2403.03186%2C2403.03788%2C2403.07718%2C2403.10170%2C2403.11905%2C2403.16971%2C2403.17918%2C2403.20271%2C2404.01549%2C2404.01744%2C2404.02475%2C2404.03648%2C2404.05427%2C2404.05719%2C2404.05955%2C2404.06474%2C2404.07972%2C2404.08755%2C2404.08860%2C2404.08948%2C2404.09992%2C2404.10887%2C2404.11459%2C2404.16006%2C2404.16048%2C2404.16054%2C2404.16660%2C2404.18074%2C2404.19296%2C2405.00516%2C2405.11120%2C2405.12842%2C2405.14573%2C2405.14751%2C2405.15793%2C2405.16247%2C2405.20309%2C2406.01014%2C2406.01623%2C2406.03679%2C2406.06947%2C2406.07822%2C2406.08184%2C2406.08340%2C2406.08451%2C2406.10227%2C2406.10819%2C2406.11317%2C2406.11896%2C2406.12373&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:28Z", "role": "primary", "sha256": "be244bfae7f89ef9449b69d85aac44295eb7b5f7e5e1bbedd76df312696676a7", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2403.03186%2C2403.03788%2C2403.07718%2C2403.10170%2C2403.11905%2C2403.16971%2C2403.17918%2C2403.20271%2C2404.01549%2C2404.01744%2C2404.02475%2C2404.03648%2C2404.05427%2C2404.05719%2C2404.05955%2C2404.06474%2C2404.07972%2C2404.08755%2C2404.08860%2C2404.08948%2C2404.09992%2C2404.10887%2C2404.11459%2C2404.16006%2C2404.16048%2C2404.16054%2C2404.16660%2C2404.18074%2C2404.19296%2C2405.00516%2C2405.11120%2C2405.12842%2C2405.14573%2C2405.14751%2C2405.15793%2C2405.16247%2C2405.20309%2C2406.01014%2C2406.01623%2C2406.03679%2C2406.06947%2C2406.07822%2C2406.08184%2C2406.08340%2C2406.08451%2C2406.10227%2C2406.10819%2C2406.11317%2C2406.11896%2C2406.12373&max_results=50"}], "id": "arxiv:2404.03648", "identifiers": {"arxiv": "2404.03648"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:28Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2404.03648"}], "name": "AutoWebGLM: A Large Language Model-based Web Navigating Agent", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-04-04", "relations": [], "review_status": "curated", "scope": "core", "summary": "Builds a web agent through simplified HTML observations, human-and-AI browsing data, curriculum training, and reinforcement-learning refinement.", "topics": ["web-navigation", "reinforcement-learning", "data-generation"], "updated_at": "2024-10-12", "url": "https://arxiv.org/abs/2404.03648"}
{"aliases": [], "authors": ["Michael Lutz", "Arth Bohra", "Manvel Saroyan", "Artem Harutyunyan", "Giovanni Campagna"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "4c18dbfedcccf6434b3a50a9aa62552372df8b772ccda6a1aa699a8f052476bc", "submitted_at": "2024-04-08T23:10:47Z", "updated_at": "2024-04-08T23:10:47Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2404.05902v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2404.05902", "identifiers": {"arxiv": "2404.05902"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:58Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2404.05902"}], "name": "WILBUR: Adaptive In-Context Learning for Robust and Accurate Web Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-04-08", "relations": [], "review_status": "curated", "scope": "core", "summary": "WILBUR learns to rank prior task demonstrations for a web agent’s prompt and synthesizes instructions from past runs. A generated task curriculum and a backtracking mechanism support adaptation and recovery, evaluated on WebVoyager.", "topics": ["web-navigation", "memory", "robustness", "data-generation"], "updated_at": "2024-04-08", "url": "https://arxiv.org/abs/2404.05902"}
{"aliases": [], "authors": ["Rajat Chawla", "Adarsh Jha", "Muskaan Kumar", "Mukunda NS", "Ishaan Bhola"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e29d14a41d59942c88938b8e83217484a1cb619847c5dff0ec1f516095ee5654", "submitted_at": "2024-04-09T11:59:41Z", "updated_at": "2024-10-27T05:54:50Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2404.16048v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:28Z", "role": "primary", "sha256": "be244bfae7f89ef9449b69d85aac44295eb7b5f7e5e1bbedd76df312696676a7", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2403.03186%2C2403.03788%2C2403.07718%2C2403.10170%2C2403.11905%2C2403.16971%2C2403.17918%2C2403.20271%2C2404.01549%2C2404.01744%2C2404.02475%2C2404.03648%2C2404.05427%2C2404.05719%2C2404.05955%2C2404.06474%2C2404.07972%2C2404.08755%2C2404.08860%2C2404.08948%2C2404.09992%2C2404.10887%2C2404.11459%2C2404.16006%2C2404.16048%2C2404.16054%2C2404.16660%2C2404.18074%2C2404.19296%2C2405.00516%2C2405.11120%2C2405.12842%2C2405.14573%2C2405.14751%2C2405.15793%2C2405.16247%2C2405.20309%2C2406.01014%2C2406.01623%2C2406.03679%2C2406.06947%2C2406.07822%2C2406.08184%2C2406.08340%2C2406.08451%2C2406.10227%2C2406.10819%2C2406.11317%2C2406.11896%2C2406.12373&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:28Z", "role": "primary", "sha256": "be244bfae7f89ef9449b69d85aac44295eb7b5f7e5e1bbedd76df312696676a7", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2403.03186%2C2403.03788%2C2403.07718%2C2403.10170%2C2403.11905%2C2403.16971%2C2403.17918%2C2403.20271%2C2404.01549%2C2404.01744%2C2404.02475%2C2404.03648%2C2404.05427%2C2404.05719%2C2404.05955%2C2404.06474%2C2404.07972%2C2404.08755%2C2404.08860%2C2404.08948%2C2404.09992%2C2404.10887%2C2404.11459%2C2404.16006%2C2404.16048%2C2404.16054%2C2404.16660%2C2404.18074%2C2404.19296%2C2405.00516%2C2405.11120%2C2405.12842%2C2405.14573%2C2405.14751%2C2405.15793%2C2405.16247%2C2405.20309%2C2406.01014%2C2406.01623%2C2406.03679%2C2406.06947%2C2406.07822%2C2406.08184%2C2406.08340%2C2406.08451%2C2406.10227%2C2406.10819%2C2406.11317%2C2406.11896%2C2406.12373&max_results=50"}], "id": "arxiv:2404.16048", "identifiers": {"arxiv": "2404.16048"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:28Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2404.16048"}], "name": "GUIDE: Graphical User Interface Data for Execution", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-04-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces annotated website task trajectories with screenshots, prior actions, reasoning, and grounded next actions for process automation.", "topics": ["data-generation", "grounding"], "updated_at": "2024-10-27", "url": "https://arxiv.org/abs/2404.16048"}
{"aliases": [], "authors": ["Sidong Feng", "Suyu Ma", "Han Wang", "David Kong", "Chunyang Chen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d200a997596ea919d5210b5ca5760241f9dfa3ad581e090a541871cd6e117522", "submitted_at": "2024-05-11T20:32:16Z", "updated_at": "2024-05-11T20:32:16Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2405.07090v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-18T04:21:52Z", "role": "primary", "sha256": "c194d15b7e422aae48b4e05216a1f1d84471b6666718b9d3dd4c9d250c356e77", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2210.03347%2C2403.04984%2C2405.07090%2C2406.06282%2C2409.00088%2C2409.10811%2C2411.10943%2C2503.09241%2C2503.20279%2C2503.23350%2C2505.05440%2C2505.10593%2C2505.16422%2C2505.21279%2C2505.24878%2C2506.02153%2C2506.02839%2C2506.05079%2C2506.07976%2C2508.02046%2C2511.07885%2C2511.11248%2C2512.21720%2C2603.20351%2C2604.19299&max_results=25"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:21:52Z", "role": "primary", "sha256": "c194d15b7e422aae48b4e05216a1f1d84471b6666718b9d3dd4c9d250c356e77", "supports": ["name", "kind", "scope", "environments", "topics", "summary", "notes"], "url": "https://export.arxiv.org/api/query?id_list=2210.03347%2C2403.04984%2C2405.07090%2C2406.06282%2C2409.00088%2C2409.10811%2C2411.10943%2C2503.09241%2C2503.20279%2C2503.23350%2C2505.05440%2C2505.10593%2C2505.16422%2C2505.21279%2C2505.24878%2C2506.02153%2C2506.02839%2C2506.05079%2C2506.07976%2C2508.02046%2C2511.07885%2C2511.11248%2C2512.21720%2C2603.20351%2C2604.19299&max_results=25"}], "id": "arxiv:2405.07090", "identifiers": {"arxiv": "2405.07090"}, "kind": "paper", "last_verified_at": "2026-10-18T04:21:52Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2405.07090"}], "name": "MUD: Towards a Large-Scale and Noise-Filtered UI Dataset for Modern Style UI Modeling", "notes": "Reviewed against the complete official arXiv abstract. Experimental results and linked artifact availability have not been independently reproduced.", "published_at": "2024-05-11", "relations": [], "review_status": "curated", "scope": "core", "summary": "MUD combines LLM-guided Android app exploration, noise filtering and human annotation to collect modern UI screenshots. The study evaluates the resulting data for element detection and UI retrieval.", "topics": ["data-generation", "screen-understanding"], "updated_at": "2024-05-11", "url": "https://arxiv.org/abs/2405.07090"}
{"aliases": [], "authors": ["Ajay Patel", "Markus Hofmarcher", "Claudiu Leoveanu-Condrei", "Marius-Constantin Dinu", "Chris Callison-Burch", "Sepp Hochreiter"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "77b292052a2ab4e8db61606ffe1da8bd5e63b3ce8c84b2e0e4b7a2fdc2f926ca", "submitted_at": "2024-05-30T17:52:36Z", "updated_at": "2024-10-01T21:28:29Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2405.20309v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2405.20309", "identifiers": {"arxiv": "2405.20309"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:58Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2405.20309"}], "name": "Large Language Models Can Self-Improve At Web Agent Tasks", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-05-30", "relations": [], "review_status": "curated", "scope": "core", "summary": "Examines self-improvement by fine-tuning web agents on three mixtures of model-generated training data. WebArena experiments and trajectory-level evaluation measures study task performance, robustness, and behavior beyond aggregate success rates.", "topics": ["data-generation", "web-navigation", "evaluation"], "updated_at": "2024-10-01", "url": "https://arxiv.org/abs/2405.20309"}
{"aliases": [], "authors": ["Wei Li", "William Bishop", "Alice Li", "Chris Rawles", "Folawiyo Campbell-Ajala", "Divya Tyamagundlu", "Oriana Riva"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6eb3209d8e2a431987fe19f01cd0818f9fb490e684d7369e7d3002dc388ff271", "submitted_at": "2024-06-06T01:49:29Z", "updated_at": "2024-11-13T16:42:22Z", "version": 6, "versioned_url": "https://arxiv.org/abs/2406.03679v6"}, "publication": {"authors": ["Wei Li", "William Bishop", "Alice Li", "Chris Rawles", "Folawiyo Campbell-Ajala", "Divya Tyamagundlu", "Oriana Riva"], "date_basis": "Explicit publisher citation_publication_date: 2024-12-16. Preserve supplied precision; do not substitute conference event dates.", "date_parts": [2024, 12, 16], "doi": "10.52202/079017-2925", "event_dates": null, "evidence_urls": ["https://proceedings.neurips.cc/paper_files/paper/2024/hash/a79f3ef3b445fd4659f44648f7ea8ffd-Abstract-Datasets_and_Benchmarks_Track.html"], "identity_basis": "Same work: normalized title, ordered author correspondence and the specific contribution in the complete publisher abstract were reviewed against the retained arXiv identity.", "issue": null, "notes": "Publisher proceedings edition; retained separately from the arXiv snapshot.", "pages": "92130-92154", "pdf_url": "https://proceedings.neurips.cc/paper_files/paper/2024/file/a79f3ef3b445fd4659f44648f7ea8ffd-Paper-Datasets_and_Benchmarks_Track.pdf", "publisher": null, "series": null, "title": "On the Effects of Data Scale on UI Control Agents", "type": "conference-paper", "url": "https://proceedings.neurips.cc/paper_files/paper/2024/hash/a79f3ef3b445fd4659f44648f7ea8ffd-Abstract-Datasets_and_Benchmarks_Track.html", "venue": "Advances in Neural Information Processing Systems", "volume": "37"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:28Z", "role": "primary", "sha256": "be244bfae7f89ef9449b69d85aac44295eb7b5f7e5e1bbedd76df312696676a7", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2403.03186%2C2403.03788%2C2403.07718%2C2403.10170%2C2403.11905%2C2403.16971%2C2403.17918%2C2403.20271%2C2404.01549%2C2404.01744%2C2404.02475%2C2404.03648%2C2404.05427%2C2404.05719%2C2404.05955%2C2404.06474%2C2404.07972%2C2404.08755%2C2404.08860%2C2404.08948%2C2404.09992%2C2404.10887%2C2404.11459%2C2404.16006%2C2404.16048%2C2404.16054%2C2404.16660%2C2404.18074%2C2404.19296%2C2405.00516%2C2405.11120%2C2405.12842%2C2405.14573%2C2405.14751%2C2405.15793%2C2405.16247%2C2405.20309%2C2406.01014%2C2406.01623%2C2406.03679%2C2406.06947%2C2406.07822%2C2406.08184%2C2406.08340%2C2406.08451%2C2406.10227%2C2406.10819%2C2406.11317%2C2406.11896%2C2406.12373&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:10:03Z", "role": "primary", "sha256": "0f499dc380fcd787579aab846e3d1e39b720b672fd9916b4a016688218ca48dd", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue", "details.publication.volume"], "url": "https://proceedings.neurips.cc/paper_files/paper/2024/hash/a79f3ef3b445fd4659f44648f7ea8ffd-Abstract-Datasets_and_Benchmarks_Track.html"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:28Z", "role": "primary", "sha256": "be244bfae7f89ef9449b69d85aac44295eb7b5f7e5e1bbedd76df312696676a7", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2403.03186%2C2403.03788%2C2403.07718%2C2403.10170%2C2403.11905%2C2403.16971%2C2403.17918%2C2403.20271%2C2404.01549%2C2404.01744%2C2404.02475%2C2404.03648%2C2404.05427%2C2404.05719%2C2404.05955%2C2404.06474%2C2404.07972%2C2404.08755%2C2404.08860%2C2404.08948%2C2404.09992%2C2404.10887%2C2404.11459%2C2404.16006%2C2404.16048%2C2404.16054%2C2404.16660%2C2404.18074%2C2404.19296%2C2405.00516%2C2405.11120%2C2405.12842%2C2405.14573%2C2405.14751%2C2405.15793%2C2405.16247%2C2405.20309%2C2406.01014%2C2406.01623%2C2406.03679%2C2406.06947%2C2406.07822%2C2406.08184%2C2406.08340%2C2406.08451%2C2406.10227%2C2406.10819%2C2406.11317%2C2406.11896%2C2406.12373&max_results=50"}], "id": "arxiv:2406.03679", "identifiers": {"arxiv": "2406.03679"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:28Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2406.03679"}, {"rel": "publication", "url": "https://proceedings.neurips.cc/paper_files/paper/2024/hash/a79f3ef3b445fd4659f44648f7ea8ffd-Abstract-Datasets_and_Benchmarks_Track.html"}, {"rel": "publication_pdf", "url": "https://proceedings.neurips.cc/paper_files/paper/2024/file/a79f3ef3b445fd4659f44648f7ea8ffd-Paper-Datasets_and_Benchmarks_Track.pdf"}], "name": "On the Effects of Data Scale on UI Control Agents", "notes": "Annotation reviewed against the primary abstract. Original dataset scale and experimental results are not independently reproduced.", "published_at": "2024-06-06", "relations": [], "review_status": "curated", "scope": "core", "summary": "Studies how Android task performance changes with demonstration scale and releases AndroidControl with high- and low-level instructions.", "topics": ["data-generation", "mobile-navigation", "evaluation"], "updated_at": "2024-11-13", "url": "https://arxiv.org/abs/2406.03679"}
{"aliases": [], "authors": ["Andrea Burns", "Kate Saenko", "Bryan A. Plummer"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "47d0ca0467a104f9b8be037ec73131f12b66a22b0a89fe975bec7680073c8f39", "submitted_at": "2024-06-12T02:43:19Z", "updated_at": "2024-08-08T02:36:04Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2406.07822v2"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:28Z", "role": "primary", "sha256": "be244bfae7f89ef9449b69d85aac44295eb7b5f7e5e1bbedd76df312696676a7", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2403.03186%2C2403.03788%2C2403.07718%2C2403.10170%2C2403.11905%2C2403.16971%2C2403.17918%2C2403.20271%2C2404.01549%2C2404.01744%2C2404.02475%2C2404.03648%2C2404.05427%2C2404.05719%2C2404.05955%2C2404.06474%2C2404.07972%2C2404.08755%2C2404.08860%2C2404.08948%2C2404.09992%2C2404.10887%2C2404.11459%2C2404.16006%2C2404.16048%2C2404.16054%2C2404.16660%2C2404.18074%2C2404.19296%2C2405.00516%2C2405.11120%2C2405.12842%2C2405.14573%2C2405.14751%2C2405.15793%2C2405.16247%2C2405.20309%2C2406.01014%2C2406.01623%2C2406.03679%2C2406.06947%2C2406.07822%2C2406.08184%2C2406.08340%2C2406.08451%2C2406.10227%2C2406.10819%2C2406.11317%2C2406.11896%2C2406.12373&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:28Z", "role": "primary", "sha256": "be244bfae7f89ef9449b69d85aac44295eb7b5f7e5e1bbedd76df312696676a7", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2403.03186%2C2403.03788%2C2403.07718%2C2403.10170%2C2403.11905%2C2403.16971%2C2403.17918%2C2403.20271%2C2404.01549%2C2404.01744%2C2404.02475%2C2404.03648%2C2404.05427%2C2404.05719%2C2404.05955%2C2404.06474%2C2404.07972%2C2404.08755%2C2404.08860%2C2404.08948%2C2404.09992%2C2404.10887%2C2404.11459%2C2404.16006%2C2404.16048%2C2404.16054%2C2404.16660%2C2404.18074%2C2404.19296%2C2405.00516%2C2405.11120%2C2405.12842%2C2405.14573%2C2405.14751%2C2405.15793%2C2405.16247%2C2405.20309%2C2406.01014%2C2406.01623%2C2406.03679%2C2406.06947%2C2406.07822%2C2406.08184%2C2406.08340%2C2406.08451%2C2406.10227%2C2406.10819%2C2406.11317%2C2406.11896%2C2406.12373&max_results=50"}], "id": "arxiv:2406.07822", "identifiers": {"arxiv": "2406.07822"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:28Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2406.07822"}], "name": "Tell Me What's Next: Textual Foresight for Generic UI Representations", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-06-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "Textual Foresight learns UI representations by describing the future screen expected after an action on the current interface. The OpenApp dataset supports pretraining and downstream evaluation of representations that connect individual elements with whole-screen changes.", "topics": ["screen-understanding", "world-models", "data-generation"], "updated_at": "2024-08-08", "url": "https://arxiv.org/abs/2406.07822"}
{"aliases": [], "authors": ["Quanfeng Lu", "Wenqi Shao", "Zitao Liu", "Lingxiao Du", "Fanqing Meng", "Boxuan Li", "Botong Chen", "Siyuan Huang", "Kaipeng Zhang", "Ping Luo"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "8b45039151353cc0fede71e2e9ba109ec53d0c1f495037a82d8fb742c9e97bbb", "submitted_at": "2024-06-12T17:44:26Z", "updated_at": "2025-08-01T03:37:57Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2406.08451v2"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:28Z", "role": "primary", "sha256": "be244bfae7f89ef9449b69d85aac44295eb7b5f7e5e1bbedd76df312696676a7", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2403.03186%2C2403.03788%2C2403.07718%2C2403.10170%2C2403.11905%2C2403.16971%2C2403.17918%2C2403.20271%2C2404.01549%2C2404.01744%2C2404.02475%2C2404.03648%2C2404.05427%2C2404.05719%2C2404.05955%2C2404.06474%2C2404.07972%2C2404.08755%2C2404.08860%2C2404.08948%2C2404.09992%2C2404.10887%2C2404.11459%2C2404.16006%2C2404.16048%2C2404.16054%2C2404.16660%2C2404.18074%2C2404.19296%2C2405.00516%2C2405.11120%2C2405.12842%2C2405.14573%2C2405.14751%2C2405.15793%2C2405.16247%2C2405.20309%2C2406.01014%2C2406.01623%2C2406.03679%2C2406.06947%2C2406.07822%2C2406.08184%2C2406.08340%2C2406.08451%2C2406.10227%2C2406.10819%2C2406.11317%2C2406.11896%2C2406.12373&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:28Z", "role": "primary", "sha256": "be244bfae7f89ef9449b69d85aac44295eb7b5f7e5e1bbedd76df312696676a7", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2403.03186%2C2403.03788%2C2403.07718%2C2403.10170%2C2403.11905%2C2403.16971%2C2403.17918%2C2403.20271%2C2404.01549%2C2404.01744%2C2404.02475%2C2404.03648%2C2404.05427%2C2404.05719%2C2404.05955%2C2404.06474%2C2404.07972%2C2404.08755%2C2404.08860%2C2404.08948%2C2404.09992%2C2404.10887%2C2404.11459%2C2404.16006%2C2404.16048%2C2404.16054%2C2404.16660%2C2404.18074%2C2404.19296%2C2405.00516%2C2405.11120%2C2405.12842%2C2405.14573%2C2405.14751%2C2405.15793%2C2405.16247%2C2405.20309%2C2406.01014%2C2406.01623%2C2406.03679%2C2406.06947%2C2406.07822%2C2406.08184%2C2406.08340%2C2406.08451%2C2406.10227%2C2406.10819%2C2406.11317%2C2406.11896%2C2406.12373&max_results=50"}], "id": "arxiv:2406.08451", "identifiers": {"arxiv": "2406.08451"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:28Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2406.08451"}], "name": "GUIOdyssey: A Comprehensive Dataset for Cross-App GUI Navigation on Mobile Devices", "notes": "Summary and topic tags reviewed against the primary abstract. Reported experiments, task counts, and artifact availability are not independently reproduced.", "published_at": "2024-06-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces annotated cross-application mobile trajectories and studies history-aware navigation over long task sequences.", "topics": ["cross-app", "data-generation"], "updated_at": "2025-08-01", "url": "https://arxiv.org/abs/2406.08451"}
{"aliases": [], "authors": ["Xiao Liu", "Tianjie Zhang", "Yu Gu", "Iat Long Iong", "Yifan Xu", "Xixuan Song", "Shudan Zhang", "Hanyu Lai", "Xinyi Liu", "Hanlin Zhao", "Jiadai Sun", "Xinyue Yang", "Yu Yang", "Zehan Qi", "Shuntian Yao", "Xueqiao Sun", "Siyi Cheng", "Qinkai Zheng", "Hao Yu", "Hanchen Zhang", "Wenyi Hong", "Ming Ding", "Lihang Pan", "Xiaotao Gu", "Aohan Zeng", "Zhengxiao Du", "Chan Hee Song", "Yu Su", "Yuxiao Dong", "Jie Tang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "603bfa4357617bfeb7a718065e77c71d77614f7240258684938dd4e6fd1a20df", "submitted_at": "2024-08-12T17:44:17Z", "updated_at": "2024-08-12T17:44:17Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2408.06327v1"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:32Z", "role": "primary", "sha256": "b9a83994d8ac36ccdb606a5219511f8bb35ad3948ade4f9f3e28d386221aedfb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2406.12814%2C2406.13719%2C2406.14056%2C2406.14250%2C2406.14314%2C2406.18082%2C2406.19263%2C2407.00993%2C2407.01476%2C2407.01489%2C2407.01511%2C2407.01558%2C2407.03037%2C2407.03913%2C2407.04346%2C2407.05291%2C2407.07061%2C2407.08850%2C2407.09018%2C2407.09295%2C2407.10956%2C2407.13032%2C2407.13063%2C2407.15673%2C2407.15711%2C2407.16741%2C2407.17490%2C2407.18901%2C2407.19056%2C2407.20183%2C2408.00203%2C2408.02544%2C2408.06327%2C2408.07199%2C2408.11824%2C2408.15978%2C2409.02453%2C2409.04566%2C2409.07429%2C2409.08264%2C2409.09354%2C2409.10741%2C2409.11295%2C2409.13711%2C2409.14337%2C2409.14818%2C2409.15441%2C2409.15637%2C2409.17140%2C2409.20566&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:32Z", "role": "primary", "sha256": "b9a83994d8ac36ccdb606a5219511f8bb35ad3948ade4f9f3e28d386221aedfb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2406.12814%2C2406.13719%2C2406.14056%2C2406.14250%2C2406.14314%2C2406.18082%2C2406.19263%2C2407.00993%2C2407.01476%2C2407.01489%2C2407.01511%2C2407.01558%2C2407.03037%2C2407.03913%2C2407.04346%2C2407.05291%2C2407.07061%2C2407.08850%2C2407.09018%2C2407.09295%2C2407.10956%2C2407.13032%2C2407.13063%2C2407.15673%2C2407.15711%2C2407.16741%2C2407.17490%2C2407.18901%2C2407.19056%2C2407.20183%2C2408.00203%2C2408.02544%2C2408.06327%2C2408.07199%2C2408.11824%2C2408.15978%2C2409.02453%2C2409.04566%2C2409.07429%2C2409.08264%2C2409.09354%2C2409.10741%2C2409.11295%2C2409.13711%2C2409.14337%2C2409.14818%2C2409.15441%2C2409.15637%2C2409.17140%2C2409.20566&max_results=50"}], "id": "arxiv:2408.06327", "identifiers": {"arxiv": "2408.06327"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2408.06327"}], "name": "VisualAgentBench: Towards Large Multimodal Models as Visual Foundation Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-08-12", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "Benchmarks visual agents across GUI, embodied, and design tasks, with training trajectories collected through solvers, agents, and human demonstrations.", "topics": ["evaluation", "data-generation"], "updated_at": "2024-08-12", "url": "https://arxiv.org/abs/2408.06327"}
{"aliases": [], "authors": ["Longxi Gao", "Li Zhang", "Shihe Wang", "Pengzhi Gao", "Wei Liu", "Jian Luan", "Shangguang Wang", "Yuanchun Li", "Mengwei Xu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "4bb00ad1be49c9bd51881827556b3a168d34c69a45e31053406f780ad2a62af1", "submitted_at": "2024-09-22T06:45:38Z", "updated_at": "2025-11-25T02:43:11Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2409.14337v3"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:32Z", "role": "primary", "sha256": "b9a83994d8ac36ccdb606a5219511f8bb35ad3948ade4f9f3e28d386221aedfb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2406.12814%2C2406.13719%2C2406.14056%2C2406.14250%2C2406.14314%2C2406.18082%2C2406.19263%2C2407.00993%2C2407.01476%2C2407.01489%2C2407.01511%2C2407.01558%2C2407.03037%2C2407.03913%2C2407.04346%2C2407.05291%2C2407.07061%2C2407.08850%2C2407.09018%2C2407.09295%2C2407.10956%2C2407.13032%2C2407.13063%2C2407.15673%2C2407.15711%2C2407.16741%2C2407.17490%2C2407.18901%2C2407.19056%2C2407.20183%2C2408.00203%2C2408.02544%2C2408.06327%2C2408.07199%2C2408.11824%2C2408.15978%2C2409.02453%2C2409.04566%2C2409.07429%2C2409.08264%2C2409.09354%2C2409.10741%2C2409.11295%2C2409.13711%2C2409.14337%2C2409.14818%2C2409.15441%2C2409.15637%2C2409.17140%2C2409.20566&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:32Z", "role": "primary", "sha256": "b9a83994d8ac36ccdb606a5219511f8bb35ad3948ade4f9f3e28d386221aedfb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2406.12814%2C2406.13719%2C2406.14056%2C2406.14250%2C2406.14314%2C2406.18082%2C2406.19263%2C2407.00993%2C2407.01476%2C2407.01489%2C2407.01511%2C2407.01558%2C2407.03037%2C2407.03913%2C2407.04346%2C2407.05291%2C2407.07061%2C2407.08850%2C2407.09018%2C2407.09295%2C2407.10956%2C2407.13032%2C2407.13063%2C2407.15673%2C2407.15711%2C2407.16741%2C2407.17490%2C2407.18901%2C2407.19056%2C2407.20183%2C2408.00203%2C2408.02544%2C2408.06327%2C2408.07199%2C2408.11824%2C2408.15978%2C2409.02453%2C2409.04566%2C2409.07429%2C2409.08264%2C2409.09354%2C2409.10741%2C2409.11295%2C2409.13711%2C2409.14337%2C2409.14818%2C2409.15441%2C2409.15637%2C2409.17140%2C2409.20566&max_results=50"}], "id": "arxiv:2409.14337", "identifiers": {"arxiv": "2409.14337"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2409.14337"}], "name": "MobileViews: A Million-scale and Diverse Mobile GUI Dataset", "notes": "Summary and topic tags reviewed against the primary abstract. Reported experiments, task counts, and artifact availability are not independently reproduced.", "published_at": "2024-09-22", "relations": [], "review_status": "curated", "scope": "core", "summary": "Builds mobile screenshot and view-hierarchy pairs through automated application traversal and evaluates their usefulness for GUI grounding.", "topics": ["screen-understanding", "data-generation"], "updated_at": "2025-11-25", "url": "https://arxiv.org/abs/2409.14337"}
{"aliases": [], "authors": ["Qinzhuo Wu", "Weikai Xu", "Wei Liu", "Tao Tan", "Jianfeng Liu", "Ang Li", "Jian Luan", "Bin Wang", "Shuo Shang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "cc22337ef29946dffcf7149df1008fd6cf02350c68d21ec0386e6485f5a14ebf", "submitted_at": "2024-09-23T08:47:54Z", "updated_at": "2024-10-03T05:23:22Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2409.14818v2"}, "publication": {"authors": ["Qinzhuo Wu", "Weikai Xu", "Wei Liu", "Tao Tan", "Jianfeng Liu", "Ang Li", "Jian Luan", "Bin Wang", "Shuo Shang"], "date_basis": "Explicit publisher citation_publication_date: 2024/11. Preserve supplied precision; do not substitute conference event dates.", "date_parts": [2024, 11], "doi": "10.18653/v1/2024.findings-emnlp.599", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2024.findings-emnlp.599/", "https://aclanthology.org/2024.findings-emnlp.599.pdf"], "identity_basis": "Same work: normalized title, ordered author correspondence and the specific contribution in the complete publisher abstract were reviewed against the retained arXiv identity. The normalized publisher abstract matches the retained arXiv abstract SHA-256. The publisher PDF title and complete first-page author list were visually checked.", "issue": null, "notes": "Publisher proceedings edition; retained separately from the arXiv snapshot. Publisher spelling, initials, name annotations or author lists differ from the retained preprint. Authors follow the publisher PDF, page 1. The landing page and its BibTeX list the fifth author as Liujian Liujianfeng; the PDF names Jianfeng Liu.", "pages": "10231-10251", "pdf_url": "https://aclanthology.org/2024.findings-emnlp.599.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "MobileVLM: A Vision-Language Model for Better Intra- and Inter-UI Understanding", "type": "conference-paper", "url": "https://aclanthology.org/2024.findings-emnlp.599/", "venue": "Findings of the Association for Computational Linguistics: EMNLP 2024", "volume": null}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:32Z", "role": "primary", "sha256": "b9a83994d8ac36ccdb606a5219511f8bb35ad3948ade4f9f3e28d386221aedfb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2406.12814%2C2406.13719%2C2406.14056%2C2406.14250%2C2406.14314%2C2406.18082%2C2406.19263%2C2407.00993%2C2407.01476%2C2407.01489%2C2407.01511%2C2407.01558%2C2407.03037%2C2407.03913%2C2407.04346%2C2407.05291%2C2407.07061%2C2407.08850%2C2407.09018%2C2407.09295%2C2407.10956%2C2407.13032%2C2407.13063%2C2407.15673%2C2407.15711%2C2407.16741%2C2407.17490%2C2407.18901%2C2407.19056%2C2407.20183%2C2408.00203%2C2408.02544%2C2408.06327%2C2408.07199%2C2408.11824%2C2408.15978%2C2409.02453%2C2409.04566%2C2409.07429%2C2409.08264%2C2409.09354%2C2409.10741%2C2409.11295%2C2409.13711%2C2409.14337%2C2409.14818%2C2409.15441%2C2409.15637%2C2409.17140%2C2409.20566&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:09:52Z", "role": "primary", "sha256": "c7f0e00d6102bac259b21a9422fba11f3da72da06535f37581b2670082c82a7b", "supports": ["details.publication.date_basis", "details.publication.date_parts", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.publisher", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue"], "url": "https://aclanthology.org/2024.findings-emnlp.599/"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:14:02.000149+00:00", "role": "primary", "sha256": "131e26bc7168b529a0db7c4bda0747cdf318b7ef16bfaf3b4dede3826cb85c1a", "supports": ["details.publication.authors", "details.publication.identity_basis", "details.publication.notes"], "url": "https://aclanthology.org/2024.findings-emnlp.599.pdf"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:32Z", "role": "primary", "sha256": "b9a83994d8ac36ccdb606a5219511f8bb35ad3948ade4f9f3e28d386221aedfb", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2406.12814%2C2406.13719%2C2406.14056%2C2406.14250%2C2406.14314%2C2406.18082%2C2406.19263%2C2407.00993%2C2407.01476%2C2407.01489%2C2407.01511%2C2407.01558%2C2407.03037%2C2407.03913%2C2407.04346%2C2407.05291%2C2407.07061%2C2407.08850%2C2407.09018%2C2407.09295%2C2407.10956%2C2407.13032%2C2407.13063%2C2407.15673%2C2407.15711%2C2407.16741%2C2407.17490%2C2407.18901%2C2407.19056%2C2407.20183%2C2408.00203%2C2408.02544%2C2408.06327%2C2408.07199%2C2408.11824%2C2408.15978%2C2409.02453%2C2409.04566%2C2409.07429%2C2409.08264%2C2409.09354%2C2409.10741%2C2409.11295%2C2409.13711%2C2409.14337%2C2409.14818%2C2409.15441%2C2409.15637%2C2409.17140%2C2409.20566&max_results=50"}], "id": "arxiv:2409.14818", "identifiers": {"arxiv": "2409.14818"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2409.14818"}, {"rel": "publication", "url": "https://aclanthology.org/2024.findings-emnlp.599/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2024.findings-emnlp.599.pdf"}], "name": "MobileVLM: A Vision-Language Model for Better Intra- and Inter-UI Understanding", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-09-23", "relations": [], "review_status": "curated", "scope": "core", "summary": "MobileVLM adds two pretraining stages and four UI tasks to learn element detail and relationships between successive screens. Its Mobile3M dataset contains three million Chinese mobile UI pages connected by recorded transition actions.", "topics": ["screen-understanding", "data-generation", "action-prediction"], "updated_at": "2024-10-03", "url": "https://arxiv.org/abs/2409.14818"}
{"aliases": [], "authors": ["Tianyue Ou", "Frank F. Xu", "Aman Madaan", "Jiarui Liu", "Robert Lo", "Abishek Sridhar", "Sudipta Sengupta", "Dan Roth", "Graham Neubig", "Shuyan Zhou"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6c19b869b10f0b4c985b88b032f78e0e72f83e7ca8d9f346417054cd2a9002e3", "submitted_at": "2024-09-24T00:51:45Z", "updated_at": "2024-11-27T16:34:52Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2409.15637v2"}, "publication": {"authors": ["Tianyue Ou", "Frank F. Xu", "Aman Madaan", "Jiarui Liu", "Robert Lo", "Abishek Sridhar", "Sudipta Sengupta", "Dan Roth", "Graham Neubig", "Shuyan Zhou"], "date_basis": "Explicit publisher citation_publication_date: 2024-12-16. Preserve supplied precision; do not substitute conference event dates.", "date_parts": [2024, 12, 16], "doi": "10.52202/079017-2908", "event_dates": null, "evidence_urls": ["https://proceedings.neurips.cc/paper_files/paper/2024/hash/a6a6891cf1dfc64d664f086cf5976e93-Abstract-Conference.html"], "identity_basis": "Same work: normalized title, ordered author correspondence and the specific contribution in the complete publisher abstract were reviewed against the retained arXiv identity. The normalized publisher abstract matches the retained arXiv abstract SHA-256.", "issue": null, "notes": "Publisher proceedings edition; retained separately from the arXiv snapshot.", "pages": "91618-91652", "pdf_url": "https://proceedings.neurips.cc/paper_files/paper/2024/file/a6a6891cf1dfc64d664f086cf5976e93-Paper-Conference.pdf", "publisher": null, "series": null, "title": "Synatra: Turning Indirect Knowledge into Direct Demonstrations for Digital Agents at Scale", "type": "conference-paper", "url": "https://proceedings.neurips.cc/paper_files/paper/2024/hash/a6a6891cf1dfc64d664f086cf5976e93-Abstract-Conference.html", "venue": "Advances in Neural Information Processing Systems", "volume": "37"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:32Z", "role": "primary", "sha256": "b9a83994d8ac36ccdb606a5219511f8bb35ad3948ade4f9f3e28d386221aedfb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2406.12814%2C2406.13719%2C2406.14056%2C2406.14250%2C2406.14314%2C2406.18082%2C2406.19263%2C2407.00993%2C2407.01476%2C2407.01489%2C2407.01511%2C2407.01558%2C2407.03037%2C2407.03913%2C2407.04346%2C2407.05291%2C2407.07061%2C2407.08850%2C2407.09018%2C2407.09295%2C2407.10956%2C2407.13032%2C2407.13063%2C2407.15673%2C2407.15711%2C2407.16741%2C2407.17490%2C2407.18901%2C2407.19056%2C2407.20183%2C2408.00203%2C2408.02544%2C2408.06327%2C2408.07199%2C2408.11824%2C2408.15978%2C2409.02453%2C2409.04566%2C2409.07429%2C2409.08264%2C2409.09354%2C2409.10741%2C2409.11295%2C2409.13711%2C2409.14337%2C2409.14818%2C2409.15441%2C2409.15637%2C2409.17140%2C2409.20566&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:10:04Z", "role": "primary", "sha256": "bf42d07f0959d7ad5ed0f4e00c34ef6cb7e8b1f10f17ac1e7a0ee1951523f821", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue", "details.publication.volume"], "url": "https://proceedings.neurips.cc/paper_files/paper/2024/hash/a6a6891cf1dfc64d664f086cf5976e93-Abstract-Conference.html"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:32Z", "role": "primary", "sha256": "b9a83994d8ac36ccdb606a5219511f8bb35ad3948ade4f9f3e28d386221aedfb", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2406.12814%2C2406.13719%2C2406.14056%2C2406.14250%2C2406.14314%2C2406.18082%2C2406.19263%2C2407.00993%2C2407.01476%2C2407.01489%2C2407.01511%2C2407.01558%2C2407.03037%2C2407.03913%2C2407.04346%2C2407.05291%2C2407.07061%2C2407.08850%2C2407.09018%2C2407.09295%2C2407.10956%2C2407.13032%2C2407.13063%2C2407.15673%2C2407.15711%2C2407.16741%2C2407.17490%2C2407.18901%2C2407.19056%2C2407.20183%2C2408.00203%2C2408.02544%2C2408.06327%2C2408.07199%2C2408.11824%2C2408.15978%2C2409.02453%2C2409.04566%2C2409.07429%2C2409.08264%2C2409.09354%2C2409.10741%2C2409.11295%2C2409.13711%2C2409.14337%2C2409.14818%2C2409.15441%2C2409.15637%2C2409.17140%2C2409.20566&max_results=50"}], "id": "arxiv:2409.15637", "identifiers": {"arxiv": "2409.15637"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2409.15637"}, {"rel": "publication", "url": "https://proceedings.neurips.cc/paper_files/paper/2024/hash/a6a6891cf1dfc64d664f086cf5976e93-Abstract-Conference.html"}, {"rel": "publication_pdf", "url": "https://proceedings.neurips.cc/paper_files/paper/2024/file/a6a6891cf1dfc64d664f086cf5976e93-Paper-Conference.pdf"}], "name": "Synatra: Turning Indirect Knowledge into Direct Demonstrations for Digital Agents at Scale", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-09-24", "relations": [], "review_status": "curated", "scope": "core", "summary": "Synatra converts indirect procedural knowledge, including human-oriented tutorials, into direct demonstrations for digital agents. It trains a 7B CodeLlama on 100,000 synthesized demonstrations and evaluates the resulting policy on Mind2Web, MiniWoB++, and WebArena.", "topics": ["data-generation", "imitation-learning", "web-navigation"], "updated_at": "2024-11-27", "url": "https://arxiv.org/abs/2409.15637"}
{"aliases": [], "authors": ["Yi-Hao Peng", "Faria Huq", "Yue Jiang", "Jason Wu", "Amanda Xin Yue Li", "Jeffrey Bigham", "Amy Pavel"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a85dca895f00b442a60123c7c85cd26a90e3d1c2a810c91c32385f877bcd755a", "submitted_at": "2024-09-30T19:55:54Z", "updated_at": "2024-09-30T19:55:54Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2410.00201v1"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:45:04Z", "role": "primary", "sha256": "bab88dfed88ca440c231aa768530374a0f19498777a70921481a904fddd46bb8", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2303.17760%2C2310.03720%2C2312.11190%2C2403.02502%2C2406.12639%2C2409.18060%2C2410.00201%2C2501.01702%2C2501.11425%2C2502.02533%2C2502.02955%2C2502.12110%2C2502.15760%2C2502.17110%2C2502.19328%2C2503.02197%2C2503.02403%2C2503.02682%2C2503.06580%2C2503.07365%2C2503.09572%2C2503.14021%2C2503.24180%2C2504.02357%2C2504.11675%2C2504.13936&max_results=26"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:45:04Z", "role": "primary", "sha256": "bab88dfed88ca440c231aa768530374a0f19498777a70921481a904fddd46bb8", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2303.17760%2C2310.03720%2C2312.11190%2C2403.02502%2C2406.12639%2C2409.18060%2C2410.00201%2C2501.01702%2C2501.11425%2C2502.02533%2C2502.02955%2C2502.12110%2C2502.15760%2C2502.17110%2C2502.19328%2C2503.02197%2C2503.02403%2C2503.02682%2C2503.06580%2C2503.07365%2C2503.09572%2C2503.14021%2C2503.24180%2C2504.02357%2C2504.11675%2C2504.13936&max_results=26"}], "id": "arxiv:2410.00201", "identifiers": {"arxiv": "2410.00201"}, "kind": "paper", "last_verified_at": "2026-10-16T06:45:04Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2410.00201"}], "name": "DreamStruct: Understanding Slides and User Interfaces via Synthetic Data Generation", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-09-30", "relations": [], "review_status": "curated", "scope": "core", "summary": "Generates labeled slides and interface visuals through code to support visual element detection, description, and classification with less manual annotation.", "topics": ["data-generation", "screen-understanding"], "updated_at": "2024-09-30", "url": "https://arxiv.org/abs/2410.00201"}
{"aliases": [], "authors": ["Shikhar Murty", "Hao Zhu", "Dzmitry Bahdanau", "Christopher D. Manning"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "5e227f3686b3234745ca1749cb8fb7c2b19383ac9dfebf68976f61f07d1ddf3b", "submitted_at": "2024-10-03T18:56:51Z", "updated_at": "2025-02-05T18:56:51Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2410.02907v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2410.02907", "identifiers": {"arxiv": "2410.02907"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:58Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2410.02907"}], "name": "NNetNav: Unsupervised Learning of Browser Agents Through Environment Interaction in the Wild", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-10-03", "relations": [], "review_status": "curated", "scope": "core", "summary": "Generates browser-agent demonstrations through autonomous exploration and retrospective task labeling, pruning trajectories that lack meaningful subtasks.", "topics": ["data-generation", "web-navigation"], "updated_at": "2025-02-05", "url": "https://arxiv.org/abs/2410.02907"}
{"aliases": [], "authors": ["Boyu Gou", "Ruohan Wang", "Boyuan Zheng", "Yanan Xie", "Cheng Chang", "Yiheng Shu", "Huan Sun", "Yu Su"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "08ece826d135b9ab9dfed2a3baf9deeab99489120da749cd67517b636649c5a6", "submitted_at": "2024-10-07T17:47:50Z", "updated_at": "2025-06-17T15:06:02Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2410.05243v3"}, "publication": {"authors": ["Boyu Gou", "Demi Ruohan Wang", "Boyuan Zheng", "Yanan Xie", "Cheng Chang", "Yiheng Shu", "Huan Sun", "Yu Su"], "date_basis": "Explicit publisher citation_publication_date: 2025-05-01. Preserve supplied precision; do not substitute conference event dates.", "date_parts": [2025, 5, 1], "doi": null, "event_dates": null, "evidence_urls": ["https://proceedings.iclr.cc/paper_files/paper/2025/hash/4ca0e369689dadb25a5345ba9755ad6f-Abstract-Conference.html"], "identity_basis": "Same work: normalized title, ordered author correspondence and the specific contribution in the complete publisher abstract were reviewed against the retained arXiv identity.", "issue": null, "notes": "Publisher proceedings edition; retained separately from the arXiv snapshot. Publisher spelling, initials, name annotations or author lists differ from the retained preprint.", "pages": "30851-30883", "pdf_url": "https://proceedings.iclr.cc/paper_files/paper/2025/file/4ca0e369689dadb25a5345ba9755ad6f-Paper-Conference.pdf", "publisher": null, "series": null, "title": "Navigating the Digital World as Humans Do: Universal Visual Grounding for GUI Agents", "type": "conference-paper", "url": "https://proceedings.iclr.cc/paper_files/paper/2025/hash/4ca0e369689dadb25a5345ba9755ad6f-Abstract-Conference.html", "venue": "International Conference on Learning Representations", "volume": "2025"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:10:01Z", "role": "primary", "sha256": "4e7384659700d8bc8e1f56de671c0d943f15b8c682bdff0fbfb20f41c707b7ff", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue", "details.publication.volume"], "url": "https://proceedings.iclr.cc/paper_files/paper/2025/hash/4ca0e369689dadb25a5345ba9755ad6f-Abstract-Conference.html"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2410.05243", "identifiers": {"arxiv": "2410.05243"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:58Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2410.05243"}, {"rel": "publication", "url": "https://proceedings.iclr.cc/paper_files/paper/2025/hash/4ca0e369689dadb25a5345ba9755ad6f-Abstract-Conference.html"}, {"rel": "publication_pdf", "url": "https://proceedings.iclr.cc/paper_files/paper/2025/file/4ca0e369689dadb25a5345ba9755ad6f-Paper-Conference.pdf"}], "name": "Navigating the Digital World as Humans Do: Universal Visual Grounding for GUI Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-10-07", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces UGround, a visual target-localization model trained with large-scale synthetic references to elements in GUI screenshots.", "topics": ["grounding", "data-generation"], "updated_at": "2025-06-17", "url": "https://arxiv.org/abs/2410.05243"}
{"aliases": [], "authors": ["Junpeng Liu", "Tianyue Ou", "Yifan Song", "Yuxiao Qu", "Wai Lam", "Chenyan Xiong", "Wenhu Chen", "Graham Neubig", "Xiang Yue"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2a3bcd26dff4f1d21a8a2772929c0c97cd1cc180a7606b0744e630dabae94609", "submitted_at": "2024-10-17T17:48:54Z", "updated_at": "2024-11-06T08:29:22Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2410.13824v3"}, "publication": {"authors": ["Junpeng Liu", "Tianyue Ou", "Yifan Song", "Yuxiao Qu", "Wai Lam", "Chenyan Xiong", "Wenhu Chen", "Graham Neubig", "Xiang Yue"], "date_basis": "Explicit publisher citation_publication_date: 2025-05-01. Preserve supplied precision; do not substitute conference event dates.", "date_parts": [2025, 5, 1], "doi": null, "event_dates": null, "evidence_urls": ["https://proceedings.iclr.cc/paper_files/paper/2025/hash/2da53cd1abdae59150e35f4693834f32-Abstract-Conference.html"], "identity_basis": "Same work: normalized title, ordered author correspondence and the specific contribution in the complete publisher abstract were reviewed against the retained arXiv identity.", "issue": null, "notes": "Publisher proceedings edition; retained separately from the arXiv snapshot.", "pages": "17971-18005", "pdf_url": "https://proceedings.iclr.cc/paper_files/paper/2025/file/2da53cd1abdae59150e35f4693834f32-Paper-Conference.pdf", "publisher": null, "series": null, "title": "Harnessing Webpage UIs for Text-Rich Visual Understanding", "type": "conference-paper", "url": "https://proceedings.iclr.cc/paper_files/paper/2025/hash/2da53cd1abdae59150e35f4693834f32-Abstract-Conference.html", "venue": "International Conference on Learning Representations", "volume": "2025"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:35Z", "role": "primary", "sha256": "b3e67984f59a1b05ccd946eb54e2d37b0bb78ab0ca796fbb6d544793dafea3e0", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2410.00467%2C2410.00689%2C2410.02810%2C2410.02907%2C2410.05243%2C2410.06703%2C2410.08164%2C2410.09006%2C2410.09024%2C2410.11843%2C2410.11871%2C2410.11872%2C2410.13232%2C2410.13757%2C2410.13824%2C2410.13825%2C2410.13886%2C2410.14803%2C2410.15164%2C2410.16464%2C2410.17236%2C2410.17401%2C2410.17520%2C2410.17883%2C2410.18603%2C2410.18963%2C2410.18967%2C2410.19054%2C2410.19100%2C2410.19461%2C2410.19609%2C2410.20745%2C2410.22552%2C2410.23218%2C2410.24024%2C2411.00820%2C2411.02337%2C2411.02391%2C2411.04468%2C2411.04890%2C2411.06559%2C2411.07618%2C2411.10323%2C2411.13451%2C2411.13591%2C2411.15004%2C2411.17465%2C2411.18279%2C2412.00402%2C2412.01268&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:10:02Z", "role": "primary", "sha256": "899100e1d0cad8d949c9075f62ab3506995587b01aab3981ac4e26eaf6d88339", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue", "details.publication.volume"], "url": "https://proceedings.iclr.cc/paper_files/paper/2025/hash/2da53cd1abdae59150e35f4693834f32-Abstract-Conference.html"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:35Z", "role": "primary", "sha256": "b3e67984f59a1b05ccd946eb54e2d37b0bb78ab0ca796fbb6d544793dafea3e0", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?id_list=2410.00467%2C2410.00689%2C2410.02810%2C2410.02907%2C2410.05243%2C2410.06703%2C2410.08164%2C2410.09006%2C2410.09024%2C2410.11843%2C2410.11871%2C2410.11872%2C2410.13232%2C2410.13757%2C2410.13824%2C2410.13825%2C2410.13886%2C2410.14803%2C2410.15164%2C2410.16464%2C2410.17236%2C2410.17401%2C2410.17520%2C2410.17883%2C2410.18603%2C2410.18963%2C2410.18967%2C2410.19054%2C2410.19100%2C2410.19461%2C2410.19609%2C2410.20745%2C2410.22552%2C2410.23218%2C2410.24024%2C2411.00820%2C2411.02337%2C2411.02391%2C2411.04468%2C2411.04890%2C2411.06559%2C2411.07618%2C2411.10323%2C2411.13451%2C2411.13591%2C2411.15004%2C2411.17465%2C2411.18279%2C2412.00402%2C2412.01268&max_results=50"}], "id": "arxiv:2410.13824", "identifiers": {"arxiv": "2410.13824"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:35Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2410.13824"}, {"rel": "publication", "url": "https://proceedings.iclr.cc/paper_files/paper/2025/hash/2da53cd1abdae59150e35f4693834f32-Abstract-Conference.html"}, {"rel": "publication_pdf", "url": "https://proceedings.iclr.cc/paper_files/paper/2025/file/2da53cd1abdae59150e35f4693834f32-Paper-Conference.pdf"}], "name": "Harnessing Webpage UIs for Text-Rich Visual Understanding", "notes": "Reviewed against the retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2024-10-17", "relations": [], "review_status": "curated", "scope": "core", "summary": "Builds MultiUI by generating instructions from webpage accessibility trees and pairing them with screenshots for multimodal training. Evaluation examines web-interface understanding, including VisualWebBench and Mind2Web, and transfer to other text-rich visual tasks.", "topics": ["data-generation", "screen-understanding", "grounding"], "updated_at": "2024-11-06", "url": "https://arxiv.org/abs/2410.13824"}
{"aliases": [], "authors": ["Xuetian Chen", "Hangcheng Li", "Jiaqing Liang", "Sihang Jiang", "Deqing Yang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "dce2e22787c6e2cce95ecc4f5583bf94d7670b4b87ab75d2b8c7efb66fbcd1a8", "submitted_at": "2024-10-25T10:46:17Z", "updated_at": "2024-11-02T08:54:21Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2410.19461v2"}}, "environments": ["web", "cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:35Z", "role": "primary", "sha256": "b3e67984f59a1b05ccd946eb54e2d37b0bb78ab0ca796fbb6d544793dafea3e0", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2410.00467%2C2410.00689%2C2410.02810%2C2410.02907%2C2410.05243%2C2410.06703%2C2410.08164%2C2410.09006%2C2410.09024%2C2410.11843%2C2410.11871%2C2410.11872%2C2410.13232%2C2410.13757%2C2410.13824%2C2410.13825%2C2410.13886%2C2410.14803%2C2410.15164%2C2410.16464%2C2410.17236%2C2410.17401%2C2410.17520%2C2410.17883%2C2410.18603%2C2410.18963%2C2410.18967%2C2410.19054%2C2410.19100%2C2410.19461%2C2410.19609%2C2410.20745%2C2410.22552%2C2410.23218%2C2410.24024%2C2411.00820%2C2411.02337%2C2411.02391%2C2411.04468%2C2411.04890%2C2411.06559%2C2411.07618%2C2411.10323%2C2411.13451%2C2411.13591%2C2411.15004%2C2411.17465%2C2411.18279%2C2412.00402%2C2412.01268&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:35Z", "role": "primary", "sha256": "b3e67984f59a1b05ccd946eb54e2d37b0bb78ab0ca796fbb6d544793dafea3e0", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2410.00467%2C2410.00689%2C2410.02810%2C2410.02907%2C2410.05243%2C2410.06703%2C2410.08164%2C2410.09006%2C2410.09024%2C2410.11843%2C2410.11871%2C2410.11872%2C2410.13232%2C2410.13757%2C2410.13824%2C2410.13825%2C2410.13886%2C2410.14803%2C2410.15164%2C2410.16464%2C2410.17236%2C2410.17401%2C2410.17520%2C2410.17883%2C2410.18603%2C2410.18963%2C2410.18967%2C2410.19054%2C2410.19100%2C2410.19461%2C2410.19609%2C2410.20745%2C2410.22552%2C2410.23218%2C2410.24024%2C2411.00820%2C2411.02337%2C2411.02391%2C2411.04468%2C2411.04890%2C2411.06559%2C2411.07618%2C2411.10323%2C2411.13451%2C2411.13591%2C2411.15004%2C2411.17465%2C2411.18279%2C2412.00402%2C2412.01268&max_results=50"}], "id": "arxiv:2410.19461", "identifiers": {"arxiv": "2410.19461"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:35Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2410.19461"}], "name": "EDGE: Enhanced Grounded GUI Understanding with Enriched Multi-Granularity Synthetic Data", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-10-25", "relations": [], "review_status": "curated", "scope": "core", "summary": "Synthesizes GUI training examples at multiple levels of detail from webpages and studies transfer of learned understanding to other interface environments.", "topics": ["data-generation", "grounding", "screen-understanding"], "updated_at": "2024-11-02", "url": "https://arxiv.org/abs/2410.19461"}
{"aliases": [], "authors": ["Hongliang He", "Wenlin Yao", "Kaixin Ma", "Wenhao Yu", "Hongming Zhang", "Tianqing Fang", "Zhenzhong Lan", "Dong Yu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "db2a115281f831ea9333a03fc2a9ebb2f09528eb082060f00a33b125f9ef7762", "submitted_at": "2024-10-25T15:01:27Z", "updated_at": "2024-10-25T15:01:27Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2410.19609v1"}, "publication": {"authors": ["Hongliang He", "Wenlin Yao", "Kaixin Ma", "Wenhao Yu", "Hongming Zhang", "Tianqing Fang", "Zhenzhong Lan", "Dong Yu"], "date_basis": "Explicit publisher citation_publication_date: 2025/7. Preserve supplied precision; do not substitute conference event dates.", "date_parts": [2025, 7], "doi": "10.18653/v1/2025.acl-long.1336", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2025.acl-long.1336/"], "identity_basis": "Same work: normalized title, ordered author correspondence and the specific contribution in the complete publisher abstract were reviewed against the retained arXiv identity.", "issue": null, "notes": "Publisher proceedings edition; retained separately from the arXiv snapshot.", "pages": "27545-27564", "pdf_url": "https://aclanthology.org/2025.acl-long.1336.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "OpenWebVoyager: Building Multimodal Web Agents via Iterative Real-World Exploration, Feedback and Optimization", "type": "conference-paper", "url": "https://aclanthology.org/2025.acl-long.1336/", "venue": "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)", "volume": null}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:09:55Z", "role": "primary", "sha256": "efd9fe9277a980d42a991abe428ccc3983299c6a8a4d91e18d19cc07a37ed22c", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.publisher", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue"], "url": "https://aclanthology.org/2025.acl-long.1336/"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2410.19609", "identifiers": {"arxiv": "2410.19609"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:58Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2410.19609"}, {"rel": "publication", "url": "https://aclanthology.org/2025.acl-long.1336/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2025.acl-long.1336.pdf"}], "name": "OpenWebVoyager: Building Multimodal Web Agents via Iterative Real-World Exploration, Feedback and Optimization", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-10-25", "relations": [], "review_status": "curated", "scope": "core", "summary": "Develops multimodal browser agents through repeated cycles of real-web exploration, trajectory feedback, and policy optimization.", "topics": ["web-navigation", "data-generation"], "updated_at": "2024-10-25", "url": "https://arxiv.org/abs/2410.19609"}
{"aliases": [], "authors": ["Zhiyong Wu", "Zhenyu Wu", "Fangzhi Xu", "Yian Wang", "Qiushi Sun", "Chengyou Jia", "Kanzhi Cheng", "Zichen Ding", "Liheng Chen", "Paul Pu Liang", "Yu Qiao"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e0dbd8ef8807e7886c02ed6296c9d188503fe7b676062fcff1890fd0652f10f3", "submitted_at": "2024-10-30T17:10:19Z", "updated_at": "2024-10-30T17:10:19Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2410.23218v1"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2410.23218", "identifiers": {"arxiv": "2410.23218"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:58Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2410.23218"}], "name": "OS-ATLAS: A Foundation Action Model for Generalist GUI Agents", "notes": "Summary and topic tags reviewed against the primary abstract. Reported experiments, task counts, and artifact availability are not independently reproduced.", "published_at": "2024-10-30", "relations": [], "review_status": "curated", "scope": "core", "summary": "Combines cross-platform GUI grounding data synthesis with action-model training to study generalization to unfamiliar interfaces.", "topics": ["grounding", "data-generation"], "updated_at": "2024-10-30", "url": "https://arxiv.org/abs/2410.23218"}
{"aliases": [], "authors": ["Yifan Xu", "Xiao Liu", "Xueqiao Sun", "Siyi Cheng", "Hao Yu", "Hanyu Lai", "Shudan Zhang", "Dan Zhang", "Jie Tang", "Yuxiao Dong"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d29fe5262adf8677bbd5fead628c5fc28707af263aa7e800d1dee1f7e499784c", "submitted_at": "2024-10-31T15:25:20Z", "updated_at": "2024-11-04T05:57:31Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2410.24024v2"}, "publication": {"authors": ["Yifan Xu", "Xiao Liu", "Xueqiao Sun", "Siyi Cheng", "Hao Yu (余浩)", "Hanyu Lai", "Shudan Zhang", "Dan Zhang", "Jie Tang", "Yuxiao Dong"], "date_basis": "Explicit publisher citation_publication_date: 2025/7. Preserve supplied precision; do not substitute conference event dates.", "date_parts": [2025, 7], "doi": "10.18653/v1/2025.acl-long.107", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2025.acl-long.107/"], "identity_basis": "Same work: normalized title, ordered author correspondence and the specific contribution in the complete publisher abstract were reviewed against the retained arXiv identity.", "issue": null, "notes": "Publisher proceedings edition; retained separately from the arXiv snapshot. Publisher spelling, initials, name annotations or author lists differ from the retained preprint.", "pages": "2144-2166", "pdf_url": "https://aclanthology.org/2025.acl-long.107.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "AndroidLab: Training and Systematic Benchmarking of Android Autonomous Agents", "type": "conference-paper", "url": "https://aclanthology.org/2025.acl-long.107/", "venue": "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)", "volume": null}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:35Z", "role": "primary", "sha256": "b3e67984f59a1b05ccd946eb54e2d37b0bb78ab0ca796fbb6d544793dafea3e0", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2410.00467%2C2410.00689%2C2410.02810%2C2410.02907%2C2410.05243%2C2410.06703%2C2410.08164%2C2410.09006%2C2410.09024%2C2410.11843%2C2410.11871%2C2410.11872%2C2410.13232%2C2410.13757%2C2410.13824%2C2410.13825%2C2410.13886%2C2410.14803%2C2410.15164%2C2410.16464%2C2410.17236%2C2410.17401%2C2410.17520%2C2410.17883%2C2410.18603%2C2410.18963%2C2410.18967%2C2410.19054%2C2410.19100%2C2410.19461%2C2410.19609%2C2410.20745%2C2410.22552%2C2410.23218%2C2410.24024%2C2411.00820%2C2411.02337%2C2411.02391%2C2411.04468%2C2411.04890%2C2411.06559%2C2411.07618%2C2411.10323%2C2411.13451%2C2411.13591%2C2411.15004%2C2411.17465%2C2411.18279%2C2412.00402%2C2412.01268&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:09:56Z", "role": "primary", "sha256": "52b82854b5f4703e6e7f3cd3e9fc49adcd011ca3ba0a8fd1aced77214c452043", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.publisher", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue"], "url": "https://aclanthology.org/2025.acl-long.107/"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:35Z", "role": "primary", "sha256": "b3e67984f59a1b05ccd946eb54e2d37b0bb78ab0ca796fbb6d544793dafea3e0", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2410.00467%2C2410.00689%2C2410.02810%2C2410.02907%2C2410.05243%2C2410.06703%2C2410.08164%2C2410.09006%2C2410.09024%2C2410.11843%2C2410.11871%2C2410.11872%2C2410.13232%2C2410.13757%2C2410.13824%2C2410.13825%2C2410.13886%2C2410.14803%2C2410.15164%2C2410.16464%2C2410.17236%2C2410.17401%2C2410.17520%2C2410.17883%2C2410.18603%2C2410.18963%2C2410.18967%2C2410.19054%2C2410.19100%2C2410.19461%2C2410.19609%2C2410.20745%2C2410.22552%2C2410.23218%2C2410.24024%2C2411.00820%2C2411.02337%2C2411.02391%2C2411.04468%2C2411.04890%2C2411.06559%2C2411.07618%2C2411.10323%2C2411.13451%2C2411.13591%2C2411.15004%2C2411.17465%2C2411.18279%2C2412.00402%2C2412.01268&max_results=50"}], "id": "arxiv:2410.24024", "identifiers": {"arxiv": "2410.24024"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:35Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2410.24024"}, {"rel": "publication", "url": "https://aclanthology.org/2025.acl-long.107/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2025.acl-long.107.pdf"}], "name": "AndroidLab: Training and Systematic Benchmarking of Android Autonomous Agents", "notes": "Summary and topic tags reviewed against the primary abstract. Reported experiments, task counts, and artifact availability are not independently reproduced.", "published_at": "2024-10-31", "relations": [], "review_status": "curated", "scope": "core", "summary": "Provides a shared Android action environment, reproducible tasks, and instruction data for training and comparing language and multimodal agents.", "topics": ["evaluation", "data-generation", "mobile-navigation"], "updated_at": "2024-11-04", "url": "https://arxiv.org/abs/2410.24024"}
{"aliases": [], "authors": ["Gaurav Verma", "Rachneet Kaur", "Nishan Srishankar", "Zhen Zeng", "Tucker Balch", "Manuela Veloso"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "94d908dd89ff27434805fe7915c5a93d5c2b1d873449babc1c566c5582cca734", "submitted_at": "2024-11-20T16:54:15Z", "updated_at": "2024-11-20T16:54:15Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2411.13451v1"}, "publication": {"authors": ["Gaurav Verma", "Rachneet Kaur", "Nishan Srishankar", "Zhen Zeng", "Tucker Balch", "Manuela Veloso"], "date_basis": "Explicit publisher citation_publication_date: 2025/7. Preserve supplied precision; do not substitute conference event dates.", "date_parts": [2025, 7], "doi": "10.18653/v1/2025.acl-long.1008", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2025.acl-long.1008/"], "identity_basis": "Same work: normalized title, ordered author correspondence and the specific contribution in the complete publisher abstract were reviewed against the retained arXiv identity.", "issue": null, "notes": "Publisher proceedings edition; retained separately from the arXiv snapshot.", "pages": "20635-20651", "pdf_url": "https://aclanthology.org/2025.acl-long.1008.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "AdaptAgent: Adapting Multimodal Web Agents with Few-Shot Learning from Human Demonstrations", "type": "conference-paper", "url": "https://aclanthology.org/2025.acl-long.1008/", "venue": "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)", "volume": null}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:09:56Z", "role": "primary", "sha256": "e8930e90d2d78f0d6ef35acc487ba355d3798cdafbcc1d2637258bad81407d41", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.publisher", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue"], "url": "https://aclanthology.org/2025.acl-long.1008/"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2411.13451", "identifiers": {"arxiv": "2411.13451"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:58Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2411.13451"}, {"rel": "publication", "url": "https://aclanthology.org/2025.acl-long.1008/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2025.acl-long.1008.pdf"}], "name": "AdaptAgent: Adapting Multimodal Web Agents with Few-Shot Learning from Human Demonstrations", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-11-20", "relations": [], "review_status": "curated", "scope": "core", "summary": "Adapts multimodal web agents to unfamiliar websites using a small number of human demonstrations through prompting or meta-learning.", "topics": ["web-navigation", "data-generation"], "updated_at": "2024-11-20", "url": "https://arxiv.org/abs/2411.13451"}
{"aliases": [], "authors": ["Weikai Xie", "Li Zhang", "Shihe Wang", "Rongjie Yi", "Mengwei Xu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "9c22d6a9aafb89e5628198fb4f6ea50fe40b1e955a2fa367571459753cea61df", "submitted_at": "2024-11-30T08:55:39Z", "updated_at": "2024-11-30T08:55:39Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2412.00402v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:35Z", "role": "primary", "sha256": "b3e67984f59a1b05ccd946eb54e2d37b0bb78ab0ca796fbb6d544793dafea3e0", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2410.00467%2C2410.00689%2C2410.02810%2C2410.02907%2C2410.05243%2C2410.06703%2C2410.08164%2C2410.09006%2C2410.09024%2C2410.11843%2C2410.11871%2C2410.11872%2C2410.13232%2C2410.13757%2C2410.13824%2C2410.13825%2C2410.13886%2C2410.14803%2C2410.15164%2C2410.16464%2C2410.17236%2C2410.17401%2C2410.17520%2C2410.17883%2C2410.18603%2C2410.18963%2C2410.18967%2C2410.19054%2C2410.19100%2C2410.19461%2C2410.19609%2C2410.20745%2C2410.22552%2C2410.23218%2C2410.24024%2C2411.00820%2C2411.02337%2C2411.02391%2C2411.04468%2C2411.04890%2C2411.06559%2C2411.07618%2C2411.10323%2C2411.13451%2C2411.13591%2C2411.15004%2C2411.17465%2C2411.18279%2C2412.00402%2C2412.01268&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:35Z", "role": "primary", "sha256": "b3e67984f59a1b05ccd946eb54e2d37b0bb78ab0ca796fbb6d544793dafea3e0", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2410.00467%2C2410.00689%2C2410.02810%2C2410.02907%2C2410.05243%2C2410.06703%2C2410.08164%2C2410.09006%2C2410.09024%2C2410.11843%2C2410.11871%2C2410.11872%2C2410.13232%2C2410.13757%2C2410.13824%2C2410.13825%2C2410.13886%2C2410.14803%2C2410.15164%2C2410.16464%2C2410.17236%2C2410.17401%2C2410.17520%2C2410.17883%2C2410.18603%2C2410.18963%2C2410.18967%2C2410.19054%2C2410.19100%2C2410.19461%2C2410.19609%2C2410.20745%2C2410.22552%2C2410.23218%2C2410.24024%2C2411.00820%2C2411.02337%2C2411.02391%2C2411.04468%2C2411.04890%2C2411.06559%2C2411.07618%2C2411.10323%2C2411.13451%2C2411.13591%2C2411.15004%2C2411.17465%2C2411.18279%2C2412.00402%2C2412.01268&max_results=50"}], "id": "arxiv:2412.00402", "identifiers": {"arxiv": "2412.00402"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:35Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2412.00402"}], "name": "DroidCall: A Dataset for LLM-powered Android Intent Invocation", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-11-30", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "Introduces data and a generation pipeline for translating natural-language requests into Android intent invocations using compact language models.", "topics": ["tool-integration", "data-generation"], "updated_at": "2024-11-30", "url": "https://arxiv.org/abs/2412.00402"}
{"aliases": [], "authors": ["Yiheng Xu", "Zekun Wang", "Junli Wang", "Dunjie Lu", "Tianbao Xie", "Amrita Saha", "Doyen Sahoo", "Tao Yu", "Caiming Xiong"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "89dbdf874ace87db12fa321e67ecf818d3964622b045ea7d95e109101820d407", "submitted_at": "2024-12-05T18:58:26Z", "updated_at": "2025-05-05T16:17:20Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2412.04454v2"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2412.04454%2C2412.04626%2C2412.05467%2C2412.09362%2C2412.09605%2C2412.10342%2C2412.10742%2C2412.10840%2C2412.13103%2C2412.13194%2C2412.13501%2C2412.14161%2C2412.14470%2C2412.16256%2C2412.17589%2C2412.18116%2C2412.18426%2C2412.19723%2C2501.01149%2C2501.04575%2C2501.07572%2C2501.11733%2C2501.12326%2C2501.13896%2C2501.16150%2C2502.01600%2C2502.01977%2C2502.02982%2C2502.06395%2C2502.07949%2C2502.08047%2C2502.08226%2C2502.11357%2C2502.13053%2C2502.13923%2C2502.14282%2C2502.18906%2C2503.00401%2C2503.02268%2C2503.03196%2C2503.04730%2C2503.05143%2C2503.06470%2C2503.09263%2C2503.11069%2C2503.11170%2C2503.12687%2C2503.15661%2C2503.15937%2C2503.16465&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2412.04454%2C2412.04626%2C2412.05467%2C2412.09362%2C2412.09605%2C2412.10342%2C2412.10742%2C2412.10840%2C2412.13103%2C2412.13194%2C2412.13501%2C2412.14161%2C2412.14470%2C2412.16256%2C2412.17589%2C2412.18116%2C2412.18426%2C2412.19723%2C2501.01149%2C2501.04575%2C2501.07572%2C2501.11733%2C2501.12326%2C2501.13896%2C2501.16150%2C2502.01600%2C2502.01977%2C2502.02982%2C2502.06395%2C2502.07949%2C2502.08047%2C2502.08226%2C2502.11357%2C2502.13053%2C2502.13923%2C2502.14282%2C2502.18906%2C2503.00401%2C2503.02268%2C2503.03196%2C2503.04730%2C2503.05143%2C2503.06470%2C2503.09263%2C2503.11069%2C2503.11170%2C2503.12687%2C2503.15661%2C2503.15937%2C2503.16465&max_results=50"}], "id": "arxiv:2412.04454", "identifiers": {"arxiv": "2412.04454"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:38Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2412.04454"}], "name": "Aguvis: Unified Pure Vision Agents for Autonomous GUI Interaction", "notes": "Summary and topic tags reviewed against the primary abstract. Reported experiments, task counts, and artifact availability are not independently reproduced.", "published_at": "2024-12-05", "relations": [], "review_status": "curated", "scope": "core", "summary": "Uses screenshots, a unified action space, and staged grounding and reasoning training for cross-platform GUI interaction.", "topics": ["grounding", "planning", "data-generation"], "updated_at": "2025-05-05", "url": "https://arxiv.org/abs/2412.04454"}
{"aliases": [], "authors": ["Huawen Shen", "Chang Liu", "Gengluo Li", "Xinlong Wang", "Yu Zhou", "Can Ma", "Xiangyang Ji"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "ee5840bbd1404674f0e9509b305e995c4dece9257e3f2a79348dde9aee99928b", "submitted_at": "2024-12-12T15:29:36Z", "updated_at": "2024-12-12T15:29:36Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2412.09362v1"}}, "environments": ["web", "android", "cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2412.04454%2C2412.04626%2C2412.05467%2C2412.09362%2C2412.09605%2C2412.10342%2C2412.10742%2C2412.10840%2C2412.13103%2C2412.13194%2C2412.13501%2C2412.14161%2C2412.14470%2C2412.16256%2C2412.17589%2C2412.18116%2C2412.18426%2C2412.19723%2C2501.01149%2C2501.04575%2C2501.07572%2C2501.11733%2C2501.12326%2C2501.13896%2C2501.16150%2C2502.01600%2C2502.01977%2C2502.02982%2C2502.06395%2C2502.07949%2C2502.08047%2C2502.08226%2C2502.11357%2C2502.13053%2C2502.13923%2C2502.14282%2C2502.18906%2C2503.00401%2C2503.02268%2C2503.03196%2C2503.04730%2C2503.05143%2C2503.06470%2C2503.09263%2C2503.11069%2C2503.11170%2C2503.12687%2C2503.15661%2C2503.15937%2C2503.16465&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2412.04454%2C2412.04626%2C2412.05467%2C2412.09362%2C2412.09605%2C2412.10342%2C2412.10742%2C2412.10840%2C2412.13103%2C2412.13194%2C2412.13501%2C2412.14161%2C2412.14470%2C2412.16256%2C2412.17589%2C2412.18116%2C2412.18426%2C2412.19723%2C2501.01149%2C2501.04575%2C2501.07572%2C2501.11733%2C2501.12326%2C2501.13896%2C2501.16150%2C2502.01600%2C2502.01977%2C2502.02982%2C2502.06395%2C2502.07949%2C2502.08047%2C2502.08226%2C2502.11357%2C2502.13053%2C2502.13923%2C2502.14282%2C2502.18906%2C2503.00401%2C2503.02268%2C2503.03196%2C2503.04730%2C2503.05143%2C2503.06470%2C2503.09263%2C2503.11069%2C2503.11170%2C2503.12687%2C2503.15661%2C2503.15937%2C2503.16465&max_results=50"}], "id": "arxiv:2412.09362", "identifiers": {"arxiv": "2412.09362"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:38Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2412.09362"}], "name": "Falcon-UI: Understanding GUI Before Following User Instructions", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-12-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "Pretrains GUI-context understanding with instruction-free web interactions before fine-tuning Falcon-UI on instructed GUI tasks.", "topics": ["screen-understanding", "data-generation"], "updated_at": "2024-12-12", "url": "https://arxiv.org/abs/2412.09362"}
{"aliases": [], "authors": ["Yiheng Xu", "Dunjie Lu", "Zhennan Shen", "Junli Wang", "Zekun Wang", "Yuchen Mao", "Caiming Xiong", "Tao Yu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "8ee630d434cef63f63c368752aeae14b0a60d49c3fde33b3a79ecbfc8fc4d06e", "submitted_at": "2024-12-12T18:59:27Z", "updated_at": "2025-03-03T18:59:36Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2412.09605v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2412.04454%2C2412.04626%2C2412.05467%2C2412.09362%2C2412.09605%2C2412.10342%2C2412.10742%2C2412.10840%2C2412.13103%2C2412.13194%2C2412.13501%2C2412.14161%2C2412.14470%2C2412.16256%2C2412.17589%2C2412.18116%2C2412.18426%2C2412.19723%2C2501.01149%2C2501.04575%2C2501.07572%2C2501.11733%2C2501.12326%2C2501.13896%2C2501.16150%2C2502.01600%2C2502.01977%2C2502.02982%2C2502.06395%2C2502.07949%2C2502.08047%2C2502.08226%2C2502.11357%2C2502.13053%2C2502.13923%2C2502.14282%2C2502.18906%2C2503.00401%2C2503.02268%2C2503.03196%2C2503.04730%2C2503.05143%2C2503.06470%2C2503.09263%2C2503.11069%2C2503.11170%2C2503.12687%2C2503.15661%2C2503.15937%2C2503.16465&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2412.04454%2C2412.04626%2C2412.05467%2C2412.09362%2C2412.09605%2C2412.10342%2C2412.10742%2C2412.10840%2C2412.13103%2C2412.13194%2C2412.13501%2C2412.14161%2C2412.14470%2C2412.16256%2C2412.17589%2C2412.18116%2C2412.18426%2C2412.19723%2C2501.01149%2C2501.04575%2C2501.07572%2C2501.11733%2C2501.12326%2C2501.13896%2C2501.16150%2C2502.01600%2C2502.01977%2C2502.02982%2C2502.06395%2C2502.07949%2C2502.08047%2C2502.08226%2C2502.11357%2C2502.13053%2C2502.13923%2C2502.14282%2C2502.18906%2C2503.00401%2C2503.02268%2C2503.03196%2C2503.04730%2C2503.05143%2C2503.06470%2C2503.09263%2C2503.11069%2C2503.11170%2C2503.12687%2C2503.15661%2C2503.15937%2C2503.16465&max_results=50"}], "id": "arxiv:2412.09605", "identifiers": {"arxiv": "2412.09605"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:38Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2412.09605"}], "name": "AgentTrek: Agent Trajectory Synthesis via Guiding Replay with Web Tutorials", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-12-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "Converts public tutorials into task specifications and executes them with an agent to synthesize multimodal interaction trajectories.", "topics": ["data-generation", "web-navigation"], "updated_at": "2025-03-03", "url": "https://arxiv.org/abs/2412.09605"}
{"aliases": [], "authors": ["Lu Wang", "Fangkai Yang", "Chaoyun Zhang", "Junting Lu", "Jiaxu Qian", "Shilin He", "Pu Zhao", "Bo Qiao", "Ray Huang", "Si Qin", "Qisheng Su", "Jiayi Ye", "Yudi Zhang", "Jian-Guang Lou", "Qingwei Lin", "Saravan Rajmohan", "Dongmei Zhang", "Qi Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "54051d974e388ed2a26f1421b8c1fc79472bb941810519039ef48e90bf4761b6", "submitted_at": "2024-12-13T11:19:56Z", "updated_at": "2025-01-13T06:47:21Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2412.10047v2"}}, "environments": ["desktop", "windows"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T07:57:41Z", "role": "primary", "sha256": "0ba893415123ecc1763614ef5fc97e5c8bb7835f10b26b43f1b921ea860484cf", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2410.06212%2C2410.06215%2C2410.07283%2C2410.07331%2C2410.07869%2C2410.08209%2C2410.08368%2C2410.10570%2C2410.12481%2C2410.12669%2C2410.15549%2C2410.16950%2C2410.21909%2C2410.22584%2C2410.22770%2C2410.23022%2C2410.23555%2C2411.02006%2C2411.02393%2C2411.02479%2C2411.03312%2C2411.03817%2C2411.05451%2C2411.11145%2C2411.11904%2C2411.11916%2C2411.12990%2C2411.14228%2C2411.15397%2C2412.01857%2C2412.04144%2C2412.04383%2C2412.05819%2C2412.06224%2C2412.06771%2C2412.09530%2C2412.09613%2C2412.10047%2C2412.10380%2C2412.10673%2C2412.12004%2C2412.13178%2C2412.13810%2C2412.14203%2C2412.14222%2C2412.16359%2C2412.19638%2C2412.20206%2C2501.03895%2C2501.05179&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:13:28Z", "role": "primary", "sha256": "aaddc3cae2fcadc9780390fceb663c0200ecb042639e2c69bccea19607cd78ea", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/abs/2412.10047v2"}], "id": "arxiv:2412.10047", "identifiers": {"arxiv": "2412.10047"}, "kind": "paper", "last_verified_at": "2026-10-17T09:13:28Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2412.10047"}], "name": "Large Action Models: From Inception to Implementation", "notes": "Reviewed against the exact-version official arXiv abstract page; its normalized abstract SHA-256 matches the retained bibliographic snapshot. The original API response body was not locally available for this review. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2024-12-13", "relations": [], "review_status": "curated", "scope": "core", "summary": "Uses a Windows-agent case study to describe a development workflow for large action models, covering data collection, training, grounding, environment integration, and evaluation.", "topics": ["automation", "data-generation", "evaluation"], "updated_at": "2025-01-13", "url": "https://arxiv.org/abs/2412.10047"}
{"aliases": [], "authors": ["Yifei Zhou", "Qianlan Yang", "Kaixiang Lin", "Min Bai", "Xiong Zhou", "Yu-Xiong Wang", "Sergey Levine", "Erran Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "362d2a470603bece43d2205e56dd633e7310cfad2ba2d91b0b765abcbcc41b2f", "submitted_at": "2024-12-17T18:59:50Z", "updated_at": "2024-12-17T18:59:50Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2412.13194v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2412.04454%2C2412.04626%2C2412.05467%2C2412.09362%2C2412.09605%2C2412.10342%2C2412.10742%2C2412.10840%2C2412.13103%2C2412.13194%2C2412.13501%2C2412.14161%2C2412.14470%2C2412.16256%2C2412.17589%2C2412.18116%2C2412.18426%2C2412.19723%2C2501.01149%2C2501.04575%2C2501.07572%2C2501.11733%2C2501.12326%2C2501.13896%2C2501.16150%2C2502.01600%2C2502.01977%2C2502.02982%2C2502.06395%2C2502.07949%2C2502.08047%2C2502.08226%2C2502.11357%2C2502.13053%2C2502.13923%2C2502.14282%2C2502.18906%2C2503.00401%2C2503.02268%2C2503.03196%2C2503.04730%2C2503.05143%2C2503.06470%2C2503.09263%2C2503.11069%2C2503.11170%2C2503.12687%2C2503.15661%2C2503.15937%2C2503.16465&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2412.04454%2C2412.04626%2C2412.05467%2C2412.09362%2C2412.09605%2C2412.10342%2C2412.10742%2C2412.10840%2C2412.13103%2C2412.13194%2C2412.13501%2C2412.14161%2C2412.14470%2C2412.16256%2C2412.17589%2C2412.18116%2C2412.18426%2C2412.19723%2C2501.01149%2C2501.04575%2C2501.07572%2C2501.11733%2C2501.12326%2C2501.13896%2C2501.16150%2C2502.01600%2C2502.01977%2C2502.02982%2C2502.06395%2C2502.07949%2C2502.08047%2C2502.08226%2C2502.11357%2C2502.13053%2C2502.13923%2C2502.14282%2C2502.18906%2C2503.00401%2C2503.02268%2C2503.03196%2C2503.04730%2C2503.05143%2C2503.06470%2C2503.09263%2C2503.11069%2C2503.11170%2C2503.12687%2C2503.15661%2C2503.15937%2C2503.16465&max_results=50"}], "id": "arxiv:2412.13194", "identifiers": {"arxiv": "2412.13194"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:38Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2412.13194"}], "name": "Proposer-Agent-Evaluator(PAE): Autonomous Skill Discovery For Foundation Model Internet Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-12-17", "relations": [], "review_status": "curated", "scope": "core", "summary": "Combines autonomous task proposal, web interaction, and visual success evaluation to discover skills and improve browser-agent policies.", "topics": ["data-generation", "reinforcement-learning", "web-navigation"], "updated_at": "2024-12-17", "url": "https://arxiv.org/abs/2412.13194"}
{"aliases": [], "authors": ["Yanheng He", "Jiahe Jin", "Shijie Xia", "Jiadi Su", "Runze Fan", "Haoyang Zou", "Xiangkun Hu", "Pengfei Liu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "db567d35887fe5367e456b35bc347400b1bc2d147730e9d34c158e1fd7c8dd6c", "submitted_at": "2024-12-23T14:02:12Z", "updated_at": "2024-12-23T14:02:12Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2412.17589v1"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2412.04454%2C2412.04626%2C2412.05467%2C2412.09362%2C2412.09605%2C2412.10342%2C2412.10742%2C2412.10840%2C2412.13103%2C2412.13194%2C2412.13501%2C2412.14161%2C2412.14470%2C2412.16256%2C2412.17589%2C2412.18116%2C2412.18426%2C2412.19723%2C2501.01149%2C2501.04575%2C2501.07572%2C2501.11733%2C2501.12326%2C2501.13896%2C2501.16150%2C2502.01600%2C2502.01977%2C2502.02982%2C2502.06395%2C2502.07949%2C2502.08047%2C2502.08226%2C2502.11357%2C2502.13053%2C2502.13923%2C2502.14282%2C2502.18906%2C2503.00401%2C2503.02268%2C2503.03196%2C2503.04730%2C2503.05143%2C2503.06470%2C2503.09263%2C2503.11069%2C2503.11170%2C2503.12687%2C2503.15661%2C2503.15937%2C2503.16465&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2412.04454%2C2412.04626%2C2412.05467%2C2412.09362%2C2412.09605%2C2412.10342%2C2412.10742%2C2412.10840%2C2412.13103%2C2412.13194%2C2412.13501%2C2412.14161%2C2412.14470%2C2412.16256%2C2412.17589%2C2412.18116%2C2412.18426%2C2412.19723%2C2501.01149%2C2501.04575%2C2501.07572%2C2501.11733%2C2501.12326%2C2501.13896%2C2501.16150%2C2502.01600%2C2502.01977%2C2502.02982%2C2502.06395%2C2502.07949%2C2502.08047%2C2502.08226%2C2502.11357%2C2502.13053%2C2502.13923%2C2502.14282%2C2502.18906%2C2503.00401%2C2503.02268%2C2503.03196%2C2503.04730%2C2503.05143%2C2503.06470%2C2503.09263%2C2503.11069%2C2503.11170%2C2503.12687%2C2503.15661%2C2503.15937%2C2503.16465&max_results=50"}], "id": "arxiv:2412.17589", "identifiers": {"arxiv": "2412.17589"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:38Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2412.17589"}], "name": "PC Agent: While You Sleep, AI Works -- A Cognitive Journey into Digital World", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-12-23", "relations": [], "review_status": "curated", "scope": "core", "summary": "Combines PC Tracker interaction capture, completion of action and reasoning annotations, and planning and grounding agents for desktop workflows.", "topics": ["recording", "data-generation", "planning"], "updated_at": "2024-12-23", "url": "https://arxiv.org/abs/2412.17589"}
{"aliases": [], "authors": ["Qiushi Sun", "Kanzhi Cheng", "Zichen Ding", "Chuanyang Jin", "Yian Wang", "Fangzhi Xu", "Zhenyu Wu", "Chengyou Jia", "Liheng Chen", "Zhoumianze Liu", "Ben Kao", "Guohao Li", "Junxian He", "Yu Qiao", "Zhiyong Wu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "40b5e3090c25f80d8d786f623812459b329acc2772f945419a585ffb27c54db3", "submitted_at": "2024-12-27T16:21:58Z", "updated_at": "2025-06-27T08:25:48Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2412.19723v3"}, "publication": {"authors": ["Qiushi Sun", "Kanzhi Cheng", "Zichen Ding", "Chuanyang Jin", "Yian Wang", "Fangzhi Xu", "Zhenyu Wu", "Chengyou Jia", "Liheng Chen", "Zhoumianze Liu", "Ben Kao", "Guohao Li", "Junxian He", "Yu Qiao", "Zhiyong Wu"], "date_basis": "Explicit publisher citation_publication_date: 2025/7. Preserve supplied precision; do not substitute conference event dates.", "date_parts": [2025, 7], "doi": "10.18653/v1/2025.acl-long.277", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2025.acl-long.277/"], "identity_basis": "Same work: normalized title, ordered author correspondence and the specific contribution in the complete publisher abstract were reviewed against the retained arXiv identity.", "issue": null, "notes": "Publisher proceedings edition; retained separately from the arXiv snapshot.", "pages": "5555-5579", "pdf_url": "https://aclanthology.org/2025.acl-long.277.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "OS-Genesis: Automating GUI Agent Trajectory Construction via Reverse Task Synthesis", "type": "conference-paper", "url": "https://aclanthology.org/2025.acl-long.277/", "venue": "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)", "volume": null}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:09:57Z", "role": "primary", "sha256": "ea538c35ce448bce6e03932b4a3e20815503c461907234b3107b3f8ccbc5481d", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.publisher", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue"], "url": "https://aclanthology.org/2025.acl-long.277/"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2412.19723", "identifiers": {"arxiv": "2412.19723"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:58Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2412.19723"}, {"rel": "publication", "url": "https://aclanthology.org/2025.acl-long.277/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2025.acl-long.277.pdf"}], "name": "OS-Genesis: Automating GUI Agent Trajectory Construction via Reverse Task Synthesis", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-12-27", "relations": [], "review_status": "curated", "scope": "core", "summary": "Builds GUI trajectories by exploring environments first and deriving tasks retrospectively, with reward-model filtering of generated trajectories.", "topics": ["data-generation"], "updated_at": "2025-06-27", "url": "https://arxiv.org/abs/2412.19723"}
{"aliases": [], "authors": ["Siyu Yuan", "Zehui Chen", "Zhiheng Xi", "Junjie Ye", "Zhengyin Du", "Jiecao Chen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "692a6984b464367c99beb9c5fbd4bbd8378e79f4a4b749503c8fbb72dd80b16f", "submitted_at": "2025-01-20T11:46:04Z", "updated_at": "2025-03-24T10:18:56Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2501.11425v3"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:45:04Z", "role": "primary", "sha256": "bab88dfed88ca440c231aa768530374a0f19498777a70921481a904fddd46bb8", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2303.17760%2C2310.03720%2C2312.11190%2C2403.02502%2C2406.12639%2C2409.18060%2C2410.00201%2C2501.01702%2C2501.11425%2C2502.02533%2C2502.02955%2C2502.12110%2C2502.15760%2C2502.17110%2C2502.19328%2C2503.02197%2C2503.02403%2C2503.02682%2C2503.06580%2C2503.07365%2C2503.09572%2C2503.14021%2C2503.24180%2C2504.02357%2C2504.11675%2C2504.13936&max_results=26"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:45:04Z", "role": "primary", "sha256": "bab88dfed88ca440c231aa768530374a0f19498777a70921481a904fddd46bb8", "supports": ["name", "kind", "summary", "topics"], "url": "https://export.arxiv.org/api/query?id_list=2303.17760%2C2310.03720%2C2312.11190%2C2403.02502%2C2406.12639%2C2409.18060%2C2410.00201%2C2501.01702%2C2501.11425%2C2502.02533%2C2502.02955%2C2502.12110%2C2502.15760%2C2502.17110%2C2502.19328%2C2503.02197%2C2503.02403%2C2503.02682%2C2503.06580%2C2503.07365%2C2503.09572%2C2503.14021%2C2503.24180%2C2504.02357%2C2504.11675%2C2504.13936&max_results=26"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T13:06:43Z", "role": "primary", "sha256": "c35562ab132e13d69155842ec63a2f4189fe8703822e5620df2592a154d2d7ed", "supports": ["environments", "scope", "notes"], "url": "https://arxiv.org/html/2501.11425"}], "id": "arxiv:2501.11425", "identifiers": {"arxiv": "2501.11425"}, "kind": "paper", "last_verified_at": "2026-10-16T06:45:04Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2501.11425"}], "name": "Agent-R: Training Language Model Agents to Reflect via Iterative Self-Training", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced. The full paper was also inspected to confirm WebShop evaluation; the broader agent-learning method is classified as adjacent.", "published_at": "2025-01-20", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "Builds self-critique training examples from searched recovery paths to improve agents' ability to detect and correct errors during interaction.", "topics": ["data-generation", "planning", "robustness"], "updated_at": "2025-03-24", "url": "https://arxiv.org/abs/2501.11425"}
{"aliases": [], "authors": ["Yujia Qin", "Yining Ye", "Junjie Fang", "Haoming Wang", "Shihao Liang", "Shizuo Tian", "Junda Zhang", "Jiahao Li", "Yunxin Li", "Shijue Huang", "Wanjun Zhong", "Kuanye Li", "Jiale Yang", "Yu Miao", "Woyu Lin", "Longxiang Liu", "Xu Jiang", "Qianli Ma", "Jingyu Li", "Xiaojun Xiao", "Kai Cai", "Chuang Li", "Yaowei Zheng", "Chaolin Jin", "Chen Li", "Xiao Zhou", "Minchao Wang", "Haoli Chen", "Zhaojian Li", "Haihua Yang", "Haifeng Liu", "Feng Lin", "Tao Peng", "Xin Liu", "Guang Shi"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d19aa758e66411a31dd1fea6f07c83d1c828bd4aad86f1564a9d7f3670c3f79f", "submitted_at": "2025-01-21T17:48:10Z", "updated_at": "2025-01-21T17:48:10Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2501.12326v1"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2412.04454%2C2412.04626%2C2412.05467%2C2412.09362%2C2412.09605%2C2412.10342%2C2412.10742%2C2412.10840%2C2412.13103%2C2412.13194%2C2412.13501%2C2412.14161%2C2412.14470%2C2412.16256%2C2412.17589%2C2412.18116%2C2412.18426%2C2412.19723%2C2501.01149%2C2501.04575%2C2501.07572%2C2501.11733%2C2501.12326%2C2501.13896%2C2501.16150%2C2502.01600%2C2502.01977%2C2502.02982%2C2502.06395%2C2502.07949%2C2502.08047%2C2502.08226%2C2502.11357%2C2502.13053%2C2502.13923%2C2502.14282%2C2502.18906%2C2503.00401%2C2503.02268%2C2503.03196%2C2503.04730%2C2503.05143%2C2503.06470%2C2503.09263%2C2503.11069%2C2503.11170%2C2503.12687%2C2503.15661%2C2503.15937%2C2503.16465&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2412.04454%2C2412.04626%2C2412.05467%2C2412.09362%2C2412.09605%2C2412.10342%2C2412.10742%2C2412.10840%2C2412.13103%2C2412.13194%2C2412.13501%2C2412.14161%2C2412.14470%2C2412.16256%2C2412.17589%2C2412.18116%2C2412.18426%2C2412.19723%2C2501.01149%2C2501.04575%2C2501.07572%2C2501.11733%2C2501.12326%2C2501.13896%2C2501.16150%2C2502.01600%2C2502.01977%2C2502.02982%2C2502.06395%2C2502.07949%2C2502.08047%2C2502.08226%2C2502.11357%2C2502.13053%2C2502.13923%2C2502.14282%2C2502.18906%2C2503.00401%2C2503.02268%2C2503.03196%2C2503.04730%2C2503.05143%2C2503.06470%2C2503.09263%2C2503.11069%2C2503.11170%2C2503.12687%2C2503.15661%2C2503.15937%2C2503.16465&max_results=50"}], "id": "arxiv:2501.12326", "identifiers": {"arxiv": "2501.12326"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:38Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2501.12326"}], "name": "UI-TARS: Pioneering Automated GUI Interaction with Native Agents", "notes": "Editorial summary based on the primary paper abstract; experiments and released artifacts have not been independently reproduced.", "published_at": "2025-01-21", "relations": [], "review_status": "curated", "scope": "core", "summary": "Studies a screenshot-driven native GUI model with unified actions, deliberate reasoning, and iterative training on interaction traces.", "topics": ["grounding", "planning", "data-generation"], "updated_at": "2025-01-21", "url": "https://arxiv.org/abs/2501.12326"}
{"aliases": [], "authors": ["Yue Fan", "Handong Zhao", "Ruiyi Zhang", "Yu Shen", "Xin Eric Wang", "Gang Wu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "0804055deb6c3513990ed8d596486400f1cb791338b7866d9e949186a34eb9c5", "submitted_at": "2025-01-23T18:16:21Z", "updated_at": "2025-01-27T18:58:42Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2501.13896v2"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2412.04454%2C2412.04626%2C2412.05467%2C2412.09362%2C2412.09605%2C2412.10342%2C2412.10742%2C2412.10840%2C2412.13103%2C2412.13194%2C2412.13501%2C2412.14161%2C2412.14470%2C2412.16256%2C2412.17589%2C2412.18116%2C2412.18426%2C2412.19723%2C2501.01149%2C2501.04575%2C2501.07572%2C2501.11733%2C2501.12326%2C2501.13896%2C2501.16150%2C2502.01600%2C2502.01977%2C2502.02982%2C2502.06395%2C2502.07949%2C2502.08047%2C2502.08226%2C2502.11357%2C2502.13053%2C2502.13923%2C2502.14282%2C2502.18906%2C2503.00401%2C2503.02268%2C2503.03196%2C2503.04730%2C2503.05143%2C2503.06470%2C2503.09263%2C2503.11069%2C2503.11170%2C2503.12687%2C2503.15661%2C2503.15937%2C2503.16465&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2412.04454%2C2412.04626%2C2412.05467%2C2412.09362%2C2412.09605%2C2412.10342%2C2412.10742%2C2412.10840%2C2412.13103%2C2412.13194%2C2412.13501%2C2412.14161%2C2412.14470%2C2412.16256%2C2412.17589%2C2412.18116%2C2412.18426%2C2412.19723%2C2501.01149%2C2501.04575%2C2501.07572%2C2501.11733%2C2501.12326%2C2501.13896%2C2501.16150%2C2502.01600%2C2502.01977%2C2502.02982%2C2502.06395%2C2502.07949%2C2502.08047%2C2502.08226%2C2502.11357%2C2502.13053%2C2502.13923%2C2502.14282%2C2502.18906%2C2503.00401%2C2503.02268%2C2503.03196%2C2503.04730%2C2503.05143%2C2503.06470%2C2503.09263%2C2503.11069%2C2503.11170%2C2503.12687%2C2503.15661%2C2503.15937%2C2503.16465&max_results=50"}], "id": "arxiv:2501.13896", "identifiers": {"arxiv": "2501.13896"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:38Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2501.13896"}], "name": "GUI-Bee: Align GUI Action Grounding to Novel Environments via Autonomous Exploration", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-01-23", "relations": [], "review_status": "curated", "scope": "core", "summary": "Collects environment-specific interaction data through autonomous exploration to adapt grounding models to previously unseen GUIs.", "topics": ["grounding", "data-generation"], "updated_at": "2025-01-27", "url": "https://arxiv.org/abs/2501.13896"}
{"aliases": [], "authors": ["Hongxin Li", "Jingfan Chen", "Jingran Su", "Yuntao Chen", "Qing Li", "Zhaoxiang Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2b76ab1c6c44db9f86b2bfc11097c54e5c3115c0c498a58d896bef043d2113c8", "submitted_at": "2025-02-04T03:39:59Z", "updated_at": "2025-06-07T05:23:58Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2502.01977v2"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2502.01977", "identifiers": {"arxiv": "2502.01977"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2502.01977"}], "name": "AutoGUI: Scaling GUI Grounding with Automatic Functionality Annotations from LLMs", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-02-04", "relations": [], "review_status": "curated", "scope": "core", "summary": "Infers element functionality from UI changes before and after interaction to produce verified annotations for grounding-model training.", "topics": ["grounding", "data-generation"], "updated_at": "2025-06-07", "url": "https://arxiv.org/abs/2502.01977"}
{"aliases": [], "authors": ["Qinzhuo Wu", "Wei Liu", "Jian Luan", "Bin Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d3e0b19f259bcdb57b0dea92578d0ee1c7c229ea11e622c56e808f17ca2568cc", "submitted_at": "2025-02-05T07:35:23Z", "updated_at": "2025-02-05T07:35:23Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2502.02955v1"}, "publication": {"authors": ["Qinzhuo Wu", "Wei Liu", "Jian Luan", "Bin Wang"], "date_basis": "Explicit publisher citation_publication_date: 2025/4. Preserve supplied precision; do not substitute conference event dates.", "date_parts": [2025, 4], "doi": "10.18653/v1/2025.naacl-long.244", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2025.naacl-long.244/"], "identity_basis": "Same work: normalized title, ordered author correspondence and the specific contribution in the complete publisher abstract were reviewed against the retained arXiv identity.", "issue": null, "notes": "Publisher proceedings edition; retained separately from the arXiv snapshot.", "pages": "4760-4775", "pdf_url": "https://aclanthology.org/2025.naacl-long.244.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "ReachAgent: Enhancing Mobile Agent via Page Reaching and Operation", "type": "conference-paper", "url": "https://aclanthology.org/2025.naacl-long.244/", "venue": "Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)", "volume": null}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:45:04Z", "role": "primary", "sha256": "bab88dfed88ca440c231aa768530374a0f19498777a70921481a904fddd46bb8", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2303.17760%2C2310.03720%2C2312.11190%2C2403.02502%2C2406.12639%2C2409.18060%2C2410.00201%2C2501.01702%2C2501.11425%2C2502.02533%2C2502.02955%2C2502.12110%2C2502.15760%2C2502.17110%2C2502.19328%2C2503.02197%2C2503.02403%2C2503.02682%2C2503.06580%2C2503.07365%2C2503.09572%2C2503.14021%2C2503.24180%2C2504.02357%2C2504.11675%2C2504.13936&max_results=26"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:09:58Z", "role": "primary", "sha256": "bb33865bafca385d3d1ce446d5cd85b6c1bf41d567509ac5ea539d4cc9c64799", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.publisher", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue"], "url": "https://aclanthology.org/2025.naacl-long.244/"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:45:04Z", "role": "primary", "sha256": "bab88dfed88ca440c231aa768530374a0f19498777a70921481a904fddd46bb8", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2303.17760%2C2310.03720%2C2312.11190%2C2403.02502%2C2406.12639%2C2409.18060%2C2410.00201%2C2501.01702%2C2501.11425%2C2502.02533%2C2502.02955%2C2502.12110%2C2502.15760%2C2502.17110%2C2502.19328%2C2503.02197%2C2503.02403%2C2503.02682%2C2503.06580%2C2503.07365%2C2503.09572%2C2503.14021%2C2503.24180%2C2504.02357%2C2504.11675%2C2504.13936&max_results=26"}], "id": "arxiv:2502.02955", "identifiers": {"arxiv": "2502.02955"}, "kind": "paper", "last_verified_at": "2026-10-16T06:45:04Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2502.02955"}, {"rel": "publication", "url": "https://aclanthology.org/2025.naacl-long.244/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2025.naacl-long.244.pdf"}], "name": "ReachAgent: Enhancing Mobile Agent via Page Reaching and Operation", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-02-05", "relations": [], "review_status": "curated", "scope": "core", "summary": "Splits mobile tasks into reaching the required page and operating on it, using subtask data and preferred GUI trajectories for training.", "topics": ["mobile-navigation", "planning", "data-generation"], "updated_at": "2025-02-05", "url": "https://arxiv.org/abs/2502.02955"}
{"aliases": [], "authors": ["Wenhao Wang", "Mengying Yuan", "Zijie Yu", "Guangyi Liu", "Rui Ye", "Tian Jin", "Siheng Chen", "Yanfeng Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "ecab608111104ac044bb7d28996604e9b87bdad4543c250c57bf432f883b96d5", "submitted_at": "2025-02-05T08:26:17Z", "updated_at": "2025-05-20T07:03:02Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2502.02982v2"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2502.02982", "identifiers": {"arxiv": "2502.02982"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2502.02982"}], "name": "MobileA3gent: Training Mobile GUI Agents Using Decentralized Self-Sourced Data from Diverse Users", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-02-05", "relations": [], "review_status": "curated", "scope": "core", "summary": "Studies automatic annotation and federated training of mobile GUI agents from decentralized user interaction data with heterogeneous distributions.", "topics": ["mobile-navigation", "data-generation"], "updated_at": "2025-05-20", "url": "https://arxiv.org/abs/2502.02982"}
{"aliases": [], "authors": ["Georgios Papoudakis", "Thomas Coste", "Zhihao Wu", "Jianye Hao", "Jun Wang", "Kun Shao"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "44ec9907c8d536488e59dea307aa05bd8adcadf96cb4f3ba25a355ffe4cf8ff7", "submitted_at": "2025-02-10T12:32:21Z", "updated_at": "2025-02-10T12:32:21Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2502.06395v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2412.04454%2C2412.04626%2C2412.05467%2C2412.09362%2C2412.09605%2C2412.10342%2C2412.10742%2C2412.10840%2C2412.13103%2C2412.13194%2C2412.13501%2C2412.14161%2C2412.14470%2C2412.16256%2C2412.17589%2C2412.18116%2C2412.18426%2C2412.19723%2C2501.01149%2C2501.04575%2C2501.07572%2C2501.11733%2C2501.12326%2C2501.13896%2C2501.16150%2C2502.01600%2C2502.01977%2C2502.02982%2C2502.06395%2C2502.07949%2C2502.08047%2C2502.08226%2C2502.11357%2C2502.13053%2C2502.13923%2C2502.14282%2C2502.18906%2C2503.00401%2C2503.02268%2C2503.03196%2C2503.04730%2C2503.05143%2C2503.06470%2C2503.09263%2C2503.11069%2C2503.11170%2C2503.12687%2C2503.15661%2C2503.15937%2C2503.16465&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2412.04454%2C2412.04626%2C2412.05467%2C2412.09362%2C2412.09605%2C2412.10342%2C2412.10742%2C2412.10840%2C2412.13103%2C2412.13194%2C2412.13501%2C2412.14161%2C2412.14470%2C2412.16256%2C2412.17589%2C2412.18116%2C2412.18426%2C2412.19723%2C2501.01149%2C2501.04575%2C2501.07572%2C2501.11733%2C2501.12326%2C2501.13896%2C2501.16150%2C2502.01600%2C2502.01977%2C2502.02982%2C2502.06395%2C2502.07949%2C2502.08047%2C2502.08226%2C2502.11357%2C2502.13053%2C2502.13923%2C2502.14282%2C2502.18906%2C2503.00401%2C2503.02268%2C2503.03196%2C2503.04730%2C2503.05143%2C2503.06470%2C2503.09263%2C2503.11069%2C2503.11170%2C2503.12687%2C2503.15661%2C2503.15937%2C2503.16465&max_results=50"}], "id": "arxiv:2502.06395", "identifiers": {"arxiv": "2502.06395"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:38Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2502.06395"}], "name": "AppVLM: A Lightweight Vision Language Model for Online App Control", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-02-10", "relations": [], "review_status": "curated", "scope": "core", "summary": "AppVLM first learns app actions from AndroidControl and then refines its policy through data collected in AndroidWorld. The study compares offline action prediction and online task completion for a lightweight vision-language smartphone agent.", "topics": ["mobile-navigation", "imitation-learning", "data-generation", "efficiency"], "updated_at": "2025-02-10", "url": "https://arxiv.org/abs/2502.06395"}
{"aliases": [], "authors": ["Ruichen Zhang", "Mufan Qiu", "Zhen Tan", "Mohan Zhang", "Vincent Lu", "Jie Peng", "Kaidi Xu", "Leandro Z. Agudelo", "Peter Qian", "Tianlong Chen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "0535de4778b6a31adb64f833733ebf94acf54ba872d98afd40b3e1012fb737fd", "submitted_at": "2025-02-11T20:41:49Z", "updated_at": "2025-03-06T19:40:57Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2502.07942v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2502.07942", "identifiers": {"arxiv": "2502.07942"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2502.07942"}], "name": "Symbiotic Cooperation for Web Agents: Harnessing Complementary Strengths of Large and Small LLMs", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-02-11", "relations": [], "review_status": "curated", "scope": "core", "summary": "Couples large- and small-model web-agent exploration with trajectory synthesis and distillation to improve both model types iteratively.", "topics": ["data-generation", "web-navigation", "efficiency"], "updated_at": "2025-03-06", "url": "https://arxiv.org/abs/2502.07942"}
{"aliases": [], "authors": ["Vardaan Pahuja", "Yadong Lu", "Corby Rosset", "Boyu Gou", "Arindam Mitra", "Spencer Whitehead", "Yu Su", "Ahmed Awadallah"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d9f459a02b7a464d9e1e49579943d1b93d46308c1b00e795c903d2fa1502a2b2", "submitted_at": "2025-02-17T02:13:48Z", "updated_at": "2025-05-30T02:30:57Z", "version": 4, "versioned_url": "https://arxiv.org/abs/2502.11357v4"}, "publication": {"authors": ["Vardaan Pahuja", "Yadong Lu", "Corby Rosset", "Boyu Gou", "Arindam Mitra", "Spencer Whitehead", "Yu Su", "Ahmed Awadallah"], "date_basis": "Explicit publisher citation_publication_date: 2025/7. Preserve supplied precision; do not substitute conference event dates.", "date_parts": [2025, 7], "doi": "10.18653/v1/2025.findings-acl.326", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2025.findings-acl.326/", "https://aclanthology.org/2025.findings-acl.326.pdf"], "identity_basis": "Same work: normalized title, ordered author correspondence and the specific contribution in the complete publisher abstract were reviewed against the retained arXiv identity. The normalized publisher abstract matches the retained arXiv abstract SHA-256. The publisher PDF title and complete first-page author list were visually checked.", "issue": null, "notes": "Publisher proceedings edition; retained separately from the arXiv snapshot. Publisher spelling, initials, name annotations or author lists differ from the retained preprint. Authors follow the publisher PDF, page 1. Landing-page citation metadata gives Ahmed Hassan for the final author, while the supplied BibTeX expands the name to Ahmed Hassan Awadallah; the PDF uses Ahmed Awadallah.", "pages": "6300-6323", "pdf_url": "https://aclanthology.org/2025.findings-acl.326.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "Explorer: Scaling Exploration-driven Web Trajectory Synthesis for Multimodal Web Agents", "type": "conference-paper", "url": "https://aclanthology.org/2025.findings-acl.326/", "venue": "Findings of the Association for Computational Linguistics: ACL 2025", "volume": null}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:09:58Z", "role": "primary", "sha256": "8ec40afe322b7c72c9f6d6db617fa2d4fae559cf6cd4e2fcc0ff5a9cc9485e05", "supports": ["details.publication.date_basis", "details.publication.date_parts", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.publisher", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue"], "url": "https://aclanthology.org/2025.findings-acl.326/"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:14:02.000475+00:00", "role": "primary", "sha256": "c1b232e161839514713789c408ae64f66684b49770033c7f3b04bf1d1a232157", "supports": ["details.publication.authors", "details.publication.identity_basis", "details.publication.notes"], "url": "https://aclanthology.org/2025.findings-acl.326.pdf"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2502.11357", "identifiers": {"arxiv": "2502.11357"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2502.11357"}, {"rel": "publication", "url": "https://aclanthology.org/2025.findings-acl.326/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2025.findings-acl.326.pdf"}], "name": "Explorer: Scaling Exploration-driven Web Trajectory Synthesis for Multimodal Web Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-02-17", "relations": [], "review_status": "curated", "scope": "core", "summary": "Synthesizes multimodal web trajectories through exploration and refinement, then trains a browser agent to study the effect of scaling interaction data.", "topics": ["data-generation", "web-navigation"], "updated_at": "2025-05-30", "url": "https://arxiv.org/abs/2502.11357"}
{"aliases": [], "authors": ["Zhixun Chen", "Ming Li", "Yuxuan Huang", "Yali Du", "Meng Fang", "Tianyi Zhou"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2a58e33cd2b60a8c734968dd92f031b9aa4811932b1a666d947b20ec4c6035a7", "submitted_at": "2025-03-04T02:14:55Z", "updated_at": "2025-06-05T01:42:08Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2503.02197v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:45:04Z", "role": "primary", "sha256": "bab88dfed88ca440c231aa768530374a0f19498777a70921481a904fddd46bb8", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2303.17760%2C2310.03720%2C2312.11190%2C2403.02502%2C2406.12639%2C2409.18060%2C2410.00201%2C2501.01702%2C2501.11425%2C2502.02533%2C2502.02955%2C2502.12110%2C2502.15760%2C2502.17110%2C2502.19328%2C2503.02197%2C2503.02403%2C2503.02682%2C2503.06580%2C2503.07365%2C2503.09572%2C2503.14021%2C2503.24180%2C2504.02357%2C2504.11675%2C2504.13936&max_results=26"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:45:04Z", "role": "primary", "sha256": "bab88dfed88ca440c231aa768530374a0f19498777a70921481a904fddd46bb8", "supports": ["name", "kind", "summary", "topics"], "url": "https://export.arxiv.org/api/query?id_list=2303.17760%2C2310.03720%2C2312.11190%2C2403.02502%2C2406.12639%2C2409.18060%2C2410.00201%2C2501.01702%2C2501.11425%2C2502.02533%2C2502.02955%2C2502.12110%2C2502.15760%2C2502.17110%2C2502.19328%2C2503.02197%2C2503.02403%2C2503.02682%2C2503.06580%2C2503.07365%2C2503.09572%2C2503.14021%2C2503.24180%2C2504.02357%2C2504.11675%2C2504.13936&max_results=26"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T13:06:43Z", "role": "primary", "sha256": "9048f54fbe64ee62117e37e3c9ad5b157fc70e83304d9ccb782f8782e487dd4a", "supports": ["environments", "scope", "notes"], "url": "https://arxiv.org/html/2503.02197"}], "id": "arxiv:2503.02197", "identifiers": {"arxiv": "2503.02197"}, "kind": "paper", "last_verified_at": "2026-10-16T06:45:04Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2503.02197"}], "name": "ATLaS: Agent Tuning via Learning Critical Steps", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced. The full paper was also inspected to confirm WebShop evaluation; the broader agent-learning method is classified as adjacent.", "published_at": "2025-03-04", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "Selects critical steps from expert interaction trajectories for targeted agent fine-tuning, studying efficiency and generalization across tasks.", "topics": ["data-generation", "planning", "efficiency"], "updated_at": "2025-06-05", "url": "https://arxiv.org/abs/2503.02197"}
{"aliases": [], "authors": ["Zhiyuan Huang", "Ziming Cheng", "Junting Pan", "Zhaohui Hou", "Mingjie Zhan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d45cacb33df18242eb3f47f6c89c56e41a146dd110e4b4dc5340062b9b981250", "submitted_at": "2025-03-05T05:30:22Z", "updated_at": "2025-04-16T15:34:51Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2503.03196v2"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2503.03196", "identifiers": {"arxiv": "2503.03196"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2503.03196"}], "name": "SpiritSight Agent: Advanced GUI Agent with One Look", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-03-05", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces SpiritSight and GUI-Lasagne data, combining multilevel GUI annotations with visual block parsing for high-resolution interface navigation.", "topics": ["grounding", "screen-understanding", "data-generation"], "updated_at": "2025-04-16", "url": "https://arxiv.org/abs/2503.03196"}
{"aliases": [], "authors": ["Lutfi Eren Erdogan", "Nicholas Lee", "Sehoon Kim", "Suhong Moon", "Hiroki Furuta", "Gopala Anumanchipalli", "Kurt Keutzer", "Amir Gholami"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6b03d84f29bb524d7950039624e0d460d8381f2ba63d70c69f5f958013730db9", "submitted_at": "2025-03-12T17:40:52Z", "updated_at": "2025-04-22T17:56:22Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2503.09572v3"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:45:04Z", "role": "primary", "sha256": "bab88dfed88ca440c231aa768530374a0f19498777a70921481a904fddd46bb8", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2303.17760%2C2310.03720%2C2312.11190%2C2403.02502%2C2406.12639%2C2409.18060%2C2410.00201%2C2501.01702%2C2501.11425%2C2502.02533%2C2502.02955%2C2502.12110%2C2502.15760%2C2502.17110%2C2502.19328%2C2503.02197%2C2503.02403%2C2503.02682%2C2503.06580%2C2503.07365%2C2503.09572%2C2503.14021%2C2503.24180%2C2504.02357%2C2504.11675%2C2504.13936&max_results=26"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:45:04Z", "role": "primary", "sha256": "bab88dfed88ca440c231aa768530374a0f19498777a70921481a904fddd46bb8", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2303.17760%2C2310.03720%2C2312.11190%2C2403.02502%2C2406.12639%2C2409.18060%2C2410.00201%2C2501.01702%2C2501.11425%2C2502.02533%2C2502.02955%2C2502.12110%2C2502.15760%2C2502.17110%2C2502.19328%2C2503.02197%2C2503.02403%2C2503.02682%2C2503.06580%2C2503.07365%2C2503.09572%2C2503.14021%2C2503.24180%2C2504.02357%2C2504.11675%2C2504.13936&max_results=26"}], "id": "arxiv:2503.09572", "identifiers": {"arxiv": "2503.09572"}, "kind": "paper", "last_verified_at": "2026-10-16T06:45:04Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2503.09572"}], "name": "Plan-and-Act: Improving Planning of Agents for Long-Horizon Tasks", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-03-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "Separates planning from execution and creates synthetic plan annotations for training agents on extended web-navigation tasks.", "topics": ["planning", "long-horizon", "data-generation"], "updated_at": "2025-04-22", "url": "https://arxiv.org/abs/2503.09572"}
{"aliases": [], "authors": ["Yibin Xu", "Liang Yang", "Hao Chen", "Hua Wang", "Zhi Chen", "Yaohua Tang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d2e6250e87ea9340914b457c4b0d471450f8cf49d2e2d53abd8e9e76105e908a", "submitted_at": "2025-03-14T08:16:02Z", "updated_at": "2025-03-14T08:16:02Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2503.11170v1"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2503.11170", "identifiers": {"arxiv": "2503.11170"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2503.11170"}], "name": "DeskVision: Large Scale Desktop Region Captioning for Advanced GUI Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-03-14", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces AutoCaptioner, DeskVision, and GUIExplorer for generating desktop region descriptions and training interface-understanding models.", "topics": ["data-generation", "grounding", "screen-understanding"], "updated_at": "2025-03-14", "url": "https://arxiv.org/abs/2503.11170"}
{"aliases": [], "authors": ["Fanbin Lu", "Zhisheng Zhong", "Ziqin Wei", "Shu Liu", "Chi-Wing Fu", "Jiaya Jia"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e883289bafee01759a0f4f3f7051cbbebdcd8059e7520d68c25ba313b8da69a6", "submitted_at": "2025-03-16T14:53:43Z", "updated_at": "2025-03-24T16:33:28Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2503.12532v2"}}, "environments": ["desktop", "windows"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2503.12532", "identifiers": {"arxiv": "2503.12532"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2503.12532"}], "name": "STEVE: A Step Verification Pipeline for Computer-use Agent Training", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-03-16", "relations": [], "review_status": "curated", "scope": "core", "summary": "Labels individual computer-use actions by comparing surrounding screenshots, then trains agents from both successful and unsuccessful steps.", "topics": ["data-generation", "evaluation", "reinforcement-learning"], "updated_at": "2025-03-24", "url": "https://arxiv.org/abs/2503.12532"}
{"aliases": [], "authors": ["Ziwei Wang", "Weizhi Chen", "Leyang Yang", "Sheng Zhou", "Shengchu Zhao", "Hanbei Zhan", "Jiongchao Jin", "Liangcheng Li", "Zirui Shao", "Jiajun Bu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "b260fd1d7e28c8263710f116638e7aa7b51f46aac15199c8a29092161a13f37d", "submitted_at": "2025-03-18T08:32:22Z", "updated_at": "2025-03-18T08:32:22Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2503.14021v1"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:45:04Z", "role": "primary", "sha256": "bab88dfed88ca440c231aa768530374a0f19498777a70921481a904fddd46bb8", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2303.17760%2C2310.03720%2C2312.11190%2C2403.02502%2C2406.12639%2C2409.18060%2C2410.00201%2C2501.01702%2C2501.11425%2C2502.02533%2C2502.02955%2C2502.12110%2C2502.15760%2C2502.17110%2C2502.19328%2C2503.02197%2C2503.02403%2C2503.02682%2C2503.06580%2C2503.07365%2C2503.09572%2C2503.14021%2C2503.24180%2C2504.02357%2C2504.11675%2C2504.13936&max_results=26"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:45:04Z", "role": "primary", "sha256": "bab88dfed88ca440c231aa768530374a0f19498777a70921481a904fddd46bb8", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2303.17760%2C2310.03720%2C2312.11190%2C2403.02502%2C2406.12639%2C2409.18060%2C2410.00201%2C2501.01702%2C2501.11425%2C2502.02533%2C2502.02955%2C2502.12110%2C2502.15760%2C2502.17110%2C2502.19328%2C2503.02197%2C2503.02403%2C2503.02682%2C2503.06580%2C2503.07365%2C2503.09572%2C2503.14021%2C2503.24180%2C2504.02357%2C2504.11675%2C2504.13936&max_results=26"}], "id": "arxiv:2503.14021", "identifiers": {"arxiv": "2503.14021"}, "kind": "paper", "last_verified_at": "2026-10-16T06:45:04Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2503.14021"}], "name": "MP-GUI: Modality Perception with MLLMs for GUI Understanding", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-03-18", "relations": [], "review_status": "curated", "scope": "core", "summary": "Uses specialized graphical, textual, and spatial perception components with a data-collection pipeline to improve GUI understanding.", "topics": ["screen-understanding", "grounding", "data-generation"], "updated_at": "2025-03-18", "url": "https://arxiv.org/abs/2503.14021"}
{"aliases": [], "authors": ["Yuchen Sun", "Shanhui Zhao", "Tao Yu", "Hao Wen", "Samith Va", "Mengwei Xu", "Yuanchun Li", "Chongyang Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "80bb507bf3cdccf6f023d0c334899d9d7099c362e9f37c314b15ff9388bae65c", "submitted_at": "2025-03-22T09:30:37Z", "updated_at": "2025-03-22T09:30:37Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2503.17709v1"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2503.17709", "identifiers": {"arxiv": "2503.17709"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2503.17709"}], "name": "GUI-Xplore: Empowering Generalizable GUI Agents with One Exploration", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-03-22", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces GUI-Xplore exploration videos and downstream tasks, with an agent that combines action-aware modeling and graph-based environment reasoning.", "topics": ["data-generation", "planning", "video"], "updated_at": "2025-03-22", "url": "https://arxiv.org/abs/2503.17709"}
{"aliases": [], "authors": ["Iason Chaimalas", "Arnas Vyšniauskas", "Gabriel Brostow"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "bf23700bf7626733b0016b773852455c503617adc81025fc9e097879b5534cca", "submitted_at": "2025-04-12T22:02:29Z", "updated_at": "2025-04-12T22:02:29Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2504.09352v1"}}, "environments": ["android", "web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:42:42Z", "role": "primary", "sha256": "d1c1bc3fee56d124427abd2aabbcf876418633a8adf7a03f4ef0317cba6a42ae", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2008.05132%2C2010.04295%2C2101.04893%2C2112.05692%2C2306.01242%2C2311.03079%2C2401.05459%2C2402.03746%2C2402.15506%2C2403.20329%2C2404.06395%2C2404.10179%2C2405.10292%2C2406.09187%2C2406.14596%2C2408.03326%2C2409.12089%2C2501.10893%2C2504.09352&max_results=19"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:42:42Z", "role": "primary", "sha256": "d1c1bc3fee56d124427abd2aabbcf876418633a8adf7a03f4ef0317cba6a42ae", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2008.05132%2C2010.04295%2C2101.04893%2C2112.05692%2C2306.01242%2C2311.03079%2C2401.05459%2C2402.03746%2C2402.15506%2C2403.20329%2C2404.06395%2C2404.10179%2C2405.10292%2C2406.09187%2C2406.14596%2C2408.03326%2C2409.12089%2C2501.10893%2C2504.09352&max_results=19"}], "id": "arxiv:2504.09352", "identifiers": {"arxiv": "2504.09352"}, "kind": "paper", "last_verified_at": "2026-10-16T06:42:42Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2504.09352"}], "name": "Explorer: Robust Collection of Interactable GUI Elements", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-04-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "Explorer collects interactable GUI elements from a live target application and records user sessions for application-specific model training. It links repeated interface states into a navigation map that supports path planning and voice-directed interaction.", "topics": ["data-generation", "recording", "grounding", "planning"], "updated_at": "2025-04-12", "url": "https://arxiv.org/abs/2504.09352"}
{"aliases": [], "authors": ["Junlei Zhang", "Zichen Ding", "Chang Ma", "Zijie Chen", "Qiushi Sun", "Zhenzhong Lan", "Junxian He"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "171517999164a3581f71cf363ec6eed357b28ca86908cb97e8498c6958762a5f", "submitted_at": "2025-04-14T11:35:02Z", "updated_at": "2025-04-15T17:13:46Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2504.10127v2"}}, "environments": ["web", "android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2504.10127", "identifiers": {"arxiv": "2504.10127"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2504.10127"}], "name": "Breaking the Data Barrier -- Building GUI Agents Through Task Generalization", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-04-14", "relations": [], "review_status": "curated", "scope": "core", "summary": "Studies whether mid-training on GUI perception, multimodal reasoning, and text reasoning transfers to GUI-agent planning. Experiments compare 11 task sources and curated mixtures using WebArena and AndroidWorld to investigate training-data choices when GUI trajectories are scarce.", "topics": ["data-generation", "planning", "evaluation"], "updated_at": "2025-04-15", "url": "https://arxiv.org/abs/2504.10127"}
{"aliases": [], "authors": ["Xinyi Liu", "Xiaoyi Zhang", "Ziyun Zhang", "Yan Lu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "363e3c3e882fa93a19df74bbf77aa98525cf4639c8f62100c3ddc3e0a438997f", "submitted_at": "2025-04-15T14:56:21Z", "updated_at": "2025-07-30T13:57:49Z", "version": 4, "versioned_url": "https://arxiv.org/abs/2504.11257v4"}, "publication": {"authors": ["Xinyi Liu", "Xiaoyi Zhang", "Ziyun Zhang", "Yan Lu"], "date_basis": "Explicit publisher citation_publication_date: 2025/7. Preserve supplied precision; do not substitute conference event dates.", "date_parts": [2025, 7], "doi": "10.18653/v1/2025.findings-acl.809", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2025.findings-acl.809/"], "identity_basis": "Same work: normalized title, ordered author correspondence and the specific contribution in the complete publisher abstract were reviewed against the retained arXiv identity.", "issue": null, "notes": "Publisher proceedings edition; retained separately from the arXiv snapshot.", "pages": "15668-15684", "pdf_url": "https://aclanthology.org/2025.findings-acl.809.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "UI-E2I-Synth: Advancing GUI Grounding with Large-Scale Instruction Synthesis", "type": "conference-paper", "url": "https://aclanthology.org/2025.findings-acl.809/", "venue": "Findings of the Association for Computational Linguistics: ACL 2025", "volume": null}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:09:59Z", "role": "primary", "sha256": "f7bcea2d3656ebc3ac305d65271000e411b735eb10556d23ee0d6157593a109f", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.publisher", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue"], "url": "https://aclanthology.org/2025.findings-acl.809/"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2504.11257", "identifiers": {"arxiv": "2504.11257"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2504.11257"}, {"rel": "publication", "url": "https://aclanthology.org/2025.findings-acl.809/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2025.findings-acl.809.pdf"}], "name": "UI-E2I-Synth: Advancing GUI Grounding with Large-Scale Instruction Synthesis", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-04-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "UI-E2I-Synth uses GPT-4o to synthesize GUI grounding instructions with varied complexity, addressing small elements, imbalanced element types, and implicit requests. The accompanying UI-I2E-Bench evaluates instruction-to-element localization across these annotation challenges.", "topics": ["data-generation", "grounding", "evaluation"], "updated_at": "2025-07-30", "url": "https://arxiv.org/abs/2504.11257"}
{"aliases": [], "authors": ["Bofei Zhang", "Zirui Shang", "Zhi Gao", "Wang Zhang", "Rui Xie", "Xiaojian Ma", "Tao Yuan", "Xinxiao Wu", "Song-Chun Zhu", "Qing Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "346fc4db7f8fbf2dc66c32e36f5cc5ace22228d60bdd4c61b6b7ac65487a4a8c", "submitted_at": "2025-04-17T06:15:56Z", "updated_at": "2025-12-04T15:01:09Z", "version": 4, "versioned_url": "https://arxiv.org/abs/2504.12679v4"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2504.12679", "identifiers": {"arxiv": "2504.12679"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2504.12679"}], "name": "TongUI: Internet-Scale Trajectories from Multimodal Web Tutorials for Generalized GUI Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-04-17", "relations": [], "review_status": "curated", "scope": "core", "summary": "TongUI converts multimodal online tutorials, including videos and articles, into GUI-agent trajectories. Its GUI-Net dataset contains 143,000 trajectories spanning five operating systems and more than 200 applications, and supports fine-tuning for grounding and navigation.", "topics": ["data-generation", "grounding", "video"], "updated_at": "2025-12-04", "url": "https://arxiv.org/abs/2504.12679"}
{"aliases": [], "authors": ["Qi Yang", "Weichen Bi", "Haiyang Shen", "Yaoqi Guo", "Yun Ma"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e1ec3311094166baac4de66ad0f083b66585c933f19cb66fc580bf128ad809ca", "submitted_at": "2025-04-23T05:01:25Z", "updated_at": "2025-04-27T06:52:23Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2504.16419v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T07:57:47Z", "role": "primary", "sha256": "27144160e8981bb9f36b60c391689774913ff3f026eeeffc2923154c404a6529", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2504.03206%2C2504.03601%2C2504.06821%2C2504.09723%2C2504.11301%2C2504.11543%2C2504.12609%2C2504.14233%2C2504.14348%2C2504.14603%2C2504.16073%2C2504.16419%2C2504.17892%2C2504.18580%2C2504.20965%2C2505.00019%2C2505.00693%2C2505.01578%2C2505.02094%2C2505.03570%2C2505.03574%2C2505.05026%2C2505.05849%2C2505.06507%2C2505.07003%2C2505.07473%2C2505.10468%2C2505.10887%2C2505.12053%2C2505.12490%2C2505.12728%2C2505.13291%2C2505.14030%2C2505.14454%2C2505.14646%2C2505.15801%2C2505.15816%2C2505.17020%2C2505.18746%2C2505.19095%2C2505.19713%2C2505.20023%2C2505.20732%2C2505.20795%2C2505.21055%2C2505.22304%2C2505.22429%2C2505.23450%2C2505.23559%2C2506.00641&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:20:13Z", "role": "primary", "sha256": "496c0419e2c8c752fc2c88d015ad2cf13c182aa4501a8305ff65ad4178113843", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/abs/2504.16419v2"}], "id": "arxiv:2504.16419", "identifiers": {"arxiv": "2504.16419"}, "kind": "paper", "last_verified_at": "2026-10-17T11:20:13Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2504.16419"}], "name": "PixelWeb: The First Web GUI Dataset with Pixel-Wise Labels", "notes": "Reviewed against the exact-version official arXiv abstract page; its complete title, ordered author list, visible version, and normalized abstract SHA match the retained snapshot. The original API response body was not locally available. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2025-04-23", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces PixelWeb, a webpage GUI dataset with element images, contours, and pixel masks. Its annotation pipeline combines bitmap channel information with DOM-based visibility and stacking analysis to address occlusion and overlapping elements.", "topics": ["data-generation", "grounding", "screen-understanding"], "updated_at": "2025-04-27", "url": "https://arxiv.org/abs/2504.16419"}
{"aliases": [], "authors": ["Tianqing Fang", "Hongming Zhang", "Zhisong Zhang", "Kaixin Ma", "Wenhao Yu", "Haitao Mi", "Dong Yu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "86ca58c92ffc2cbbcb0297d8c7618dfab966e36a07b141c068998b0f2934d11e", "submitted_at": "2025-04-23T02:54:31Z", "updated_at": "2025-08-21T05:55:43Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2504.21024v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2504.21024", "identifiers": {"arxiv": "2504.21024"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2504.21024"}], "name": "WebEvolver: Enhancing Web Agent Self-Improvement with Coevolving World Model", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-04-23", "relations": [], "review_status": "curated", "scope": "core", "summary": "WebEvolver trains a co-evolving world model to predict web observations and generate additional agent-training trajectories. The same model supports look-ahead action selection at inference, with evaluation on live web-task benchmarks.", "topics": ["world-models", "data-generation", "web-navigation", "planning"], "updated_at": "2025-08-21", "url": "https://arxiv.org/abs/2504.21024"}
{"aliases": [], "authors": ["Jing Huang", "Zhixiong Zeng", "Wenkang Han", "Yufeng Zhong", "Liming Zheng", "Shuai Fu", "Jingyuan Chen", "Lin Ma"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e582c92521ec41a0a3daa0ca72b8c00250621289352b28400ee6d311c4a106eb", "submitted_at": "2025-05-01T09:27:13Z", "updated_at": "2025-05-01T09:27:13Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2505.00416v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2505.00416", "identifiers": {"arxiv": "2505.00416"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2505.00416"}], "name": "ScaleTrack: Scaling and back-tracking Automated GUI Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-05-01", "relations": [], "review_status": "curated", "scope": "core", "summary": "ScaleTrack combines a unified collection of GUI grounding samples with a planning objective that predicts the next action while reconstructing prior actions. This backward prediction links the current screenshot to interaction history during agent training.", "topics": ["grounding", "planning", "data-generation", "interaction-history"], "updated_at": "2025-05-01", "url": "https://arxiv.org/abs/2505.00416"}
{"aliases": [], "authors": ["Kun Huang", "Weikai Xu", "Yuxuan Liu", "Quandong Wang", "Pengzhi Gao", "Wei Liu", "Jian Luan", "Bin Wang", "Bo An"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "9bc92029e16b5afb4b54837d2ab2304c7b8176548061c9deafdca101754128dc", "submitted_at": "2025-05-18T08:28:05Z", "updated_at": "2026-03-21T09:25:27Z", "version": 4, "versioned_url": "https://arxiv.org/abs/2505.12299v4"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:42Z", "role": "primary", "sha256": "afe4e75ffb8a5c44f4c088a90ffa5f53553036f514aa768f5aee0c5aa38d731e", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2503.16788%2C2503.17709%2C2503.18492%2C2503.19537%2C2503.20978%2C2503.21620%2C2503.23434%2C2504.01382%2C2504.07981%2C2504.10445%2C2504.10458%2C2504.11257%2C2504.11281%2C2504.12679%2C2504.13805%2C2504.13865%2C2504.14239%2C2504.18575%2C2504.19838%2C2504.20464%2C2505.10924%2C2505.11891%2C2505.12299%2C2505.12632%2C2505.13227%2C2505.13887%2C2505.13909%2C2505.14289%2C2505.15810%2C2505.16282%2C2505.16827%2C2505.18829%2C2505.19897%2C2505.21496%2C2505.21499%2C2505.21936%2C2505.22648%2C2505.23762%2C2505.24838%2C2506.00618%2C2506.01391%2C2506.01968%2C2506.02456%2C2506.03143%2C2506.03569%2C2506.04135%2C2506.04614%2C2506.07672%2C2506.08012%2C2506.08972&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:42Z", "role": "primary", "sha256": "afe4e75ffb8a5c44f4c088a90ffa5f53553036f514aa768f5aee0c5aa38d731e", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2503.16788%2C2503.17709%2C2503.18492%2C2503.19537%2C2503.20978%2C2503.21620%2C2503.23434%2C2504.01382%2C2504.07981%2C2504.10445%2C2504.10458%2C2504.11257%2C2504.11281%2C2504.12679%2C2504.13805%2C2504.13865%2C2504.14239%2C2504.18575%2C2504.19838%2C2504.20464%2C2505.10924%2C2505.11891%2C2505.12299%2C2505.12632%2C2505.13227%2C2505.13887%2C2505.13909%2C2505.14289%2C2505.15810%2C2505.16282%2C2505.16827%2C2505.18829%2C2505.19897%2C2505.21496%2C2505.21499%2C2505.21936%2C2505.22648%2C2505.23762%2C2505.24838%2C2506.00618%2C2506.01391%2C2506.01968%2C2506.02456%2C2506.03143%2C2506.03569%2C2506.04135%2C2506.04614%2C2506.07672%2C2506.08012%2C2506.08972&max_results=50"}], "id": "arxiv:2505.12299", "identifiers": {"arxiv": "2505.12299"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:42Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2505.12299"}], "name": "MobileIPL: Enhancing Mobile Agents Thinking Process via Iterative Preference Learning", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-05-18", "relations": [], "review_status": "curated", "scope": "core", "summary": "MobileIPL samples trees of action-planning thoughts, propagates rule-based leaf rewards backward, and forms preference pairs for reasoning-level optimization. It also expands screenshot-based instruction data through staged evolution before evaluation on mobile GUI-agent benchmarks.", "topics": ["mobile-navigation", "reinforcement-learning", "planning", "data-generation"], "updated_at": "2026-03-21", "url": "https://arxiv.org/abs/2505.12299"}
{"aliases": [], "authors": ["Longxi Gao", "Li Zhang", "Pengzhi Gao", "Wei Liu", "Jian Luan", "Mengwei Xu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "10d826ea7715ab7054522a9dcba622d5dae3e6c3faee421e4b168e16104d184b", "submitted_at": "2025-05-18T16:34:30Z", "updated_at": "2025-10-10T09:28:53Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2505.12493v3"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2505.12493", "identifiers": {"arxiv": "2505.12493"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2505.12493"}], "name": "GUI-Shift: Enhancing VLM-Based GUI Agents through Self-supervised Reinforcement Learning", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-05-18", "relations": [], "review_status": "curated", "scope": "core", "summary": "GUI-Shift trains an inverse-dynamics task that predicts the initial action connecting two GUI states, using trajectories without natural-language instructions. It combines rule-based reinforcement learning with data filtering and evaluates transfer to Android automation and screenshot grounding.", "topics": ["reinforcement-learning", "data-generation", "grounding", "mobile-navigation"], "updated_at": "2025-10-10", "url": "https://arxiv.org/abs/2505.12493"}
{"aliases": [], "authors": ["Yunseok Jang", "Yeda Song", "Sungryull Sohn", "Lajanugen Logeswaran", "Tiange Luo", "Dong-Ki Kim", "Kyunghoon Bae", "Honglak Lee"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "64672e6a3450256a8b04d9db231e885e2dd4e8688235f6c7853d70d6276610ff", "submitted_at": "2025-05-19T02:39:03Z", "updated_at": "2025-05-19T02:39:03Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2505.12632v1"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:42Z", "role": "primary", "sha256": "afe4e75ffb8a5c44f4c088a90ffa5f53553036f514aa768f5aee0c5aa38d731e", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2503.16788%2C2503.17709%2C2503.18492%2C2503.19537%2C2503.20978%2C2503.21620%2C2503.23434%2C2504.01382%2C2504.07981%2C2504.10445%2C2504.10458%2C2504.11257%2C2504.11281%2C2504.12679%2C2504.13805%2C2504.13865%2C2504.14239%2C2504.18575%2C2504.19838%2C2504.20464%2C2505.10924%2C2505.11891%2C2505.12299%2C2505.12632%2C2505.13227%2C2505.13887%2C2505.13909%2C2505.14289%2C2505.15810%2C2505.16282%2C2505.16827%2C2505.18829%2C2505.19897%2C2505.21496%2C2505.21499%2C2505.21936%2C2505.22648%2C2505.23762%2C2505.24838%2C2506.00618%2C2506.01391%2C2506.01968%2C2506.02456%2C2506.03143%2C2506.03569%2C2506.04135%2C2506.04614%2C2506.07672%2C2506.08012%2C2506.08972&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:42Z", "role": "primary", "sha256": "afe4e75ffb8a5c44f4c088a90ffa5f53553036f514aa768f5aee0c5aa38d731e", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2503.16788%2C2503.17709%2C2503.18492%2C2503.19537%2C2503.20978%2C2503.21620%2C2503.23434%2C2504.01382%2C2504.07981%2C2504.10445%2C2504.10458%2C2504.11257%2C2504.11281%2C2504.12679%2C2504.13805%2C2504.13865%2C2504.14239%2C2504.18575%2C2504.19838%2C2504.20464%2C2505.10924%2C2505.11891%2C2505.12299%2C2505.12632%2C2505.13227%2C2505.13887%2C2505.13909%2C2505.14289%2C2505.15810%2C2505.16282%2C2505.16827%2C2505.18829%2C2505.19897%2C2505.21496%2C2505.21499%2C2505.21936%2C2505.22648%2C2505.23762%2C2505.24838%2C2506.00618%2C2506.01391%2C2506.01968%2C2506.02456%2C2506.03143%2C2506.03569%2C2506.04135%2C2506.04614%2C2506.07672%2C2506.08012%2C2506.08972&max_results=50"}], "id": "arxiv:2505.12632", "identifiers": {"arxiv": "2505.12632"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:42Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2505.12632"}], "name": "Scalable Video-to-Dataset Generation for Cross-Platform Mobile Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-05-19", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces MONDAY and an automated pipeline that extracts mobile-navigation training data from instructional videos across operating-system platforms.", "topics": ["video", "data-generation", "mobile-navigation"], "updated_at": "2025-05-19", "url": "https://arxiv.org/abs/2505.12632"}
{"aliases": [], "authors": ["Tianbao Xie", "Jiaqi Deng", "Xiaochuan Li", "Junlin Yang", "Haoyuan Wu", "Jixuan Chen", "Wenjing Hu", "Xinyuan Wang", "Yuhui Xu", "Zekun Wang", "Yiheng Xu", "Junli Wang", "Doyen Sahoo", "Tao Yu", "Caiming Xiong"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "1bec4ad578c01ebdfb58bd269562d3ca7a2bd8d2687cde07cb867db0b78cb105", "submitted_at": "2025-05-19T15:09:23Z", "updated_at": "2025-10-24T19:08:03Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2505.13227v3"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2505.13227", "identifiers": {"arxiv": "2505.13227"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2505.13227"}], "name": "Scaling Computer-Use Grounding via User Interface Decomposition and Synthesis", "notes": "Summary and topic tags reviewed against the primary abstract. Reported experiments, task counts, and artifact availability are not independently reproduced.", "published_at": "2025-05-19", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces OSWorld-G grounding evaluation and the Jedi synthesis pipeline for examples spanning layout reasoning and fine-grained interaction.", "topics": ["grounding", "data-generation", "evaluation"], "updated_at": "2025-10-24", "url": "https://arxiv.org/abs/2505.13227"}
{"aliases": [], "authors": ["Yanheng He", "Jiahe Jin", "Pengfei Liu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "026a64046926912a54e6a7c6cf62a17740a6f89a94713615f9750f8bf8302c78", "submitted_at": "2025-05-20T04:20:18Z", "updated_at": "2026-03-03T03:55:55Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2505.13909v2"}}, "environments": ["windows", "desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2505.13909", "identifiers": {"arxiv": "2505.13909"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2505.13909"}], "name": "Efficient Agent Training for Computer Use", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-05-20", "relations": [], "review_status": "curated", "scope": "core", "summary": "PC Agent-E augments 312 human computer-use trajectories with alternative action decisions synthesized by Claude 3.7 Sonnet. It studies training with this combination and introduces WindowsAgentArena-V2 as an updated evaluation benchmark.", "topics": ["data-generation", "evaluation"], "updated_at": "2026-03-03", "url": "https://arxiv.org/abs/2505.13909"}
{"aliases": [], "authors": ["Hyungjoo Chae", "Sunghwan Kim", "Junhee Cho", "Seungone Kim", "Seungjun Moon", "Gyeom Hwangbo", "Dongha Lim", "Minjin Kim", "Yeonjun Hwang", "Minju Gwak", "Dongwook Choi", "Minseok Kang", "Gwanhoon Im", "ByeongUng Cho", "Hyojun Kim", "Jun Hee Han", "Taeyoon Kwon", "Minju Kim", "Beong-woo Kwak", "Dongjin Kang", "Jinyoung Yeo"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "440abdc75c0a51478cfedc6ee0a9196ccff641ec5a23849ba56be43448bab15f", "submitted_at": "2025-05-21T08:56:55Z", "updated_at": "2025-11-25T21:34:57Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2505.15277v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2505.15277", "identifiers": {"arxiv": "2505.15277"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2505.15277"}], "name": "Web-Shepherd: Advancing PRMs for Reinforcing Web Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-05-21", "relations": [], "review_status": "curated", "scope": "core", "summary": "Web-Shepherd is a process reward model for assessing individual steps of web-navigation trajectories. It introduces a collection of 40,000 step-level preference pairs with checklists and a separate benchmark for comparing reward models, then evaluates verification during WebArena-lite navigation.", "topics": ["reinforcement-learning", "evaluation", "data-generation", "web-navigation"], "updated_at": "2025-11-25", "url": "https://arxiv.org/abs/2505.15277"}
{"aliases": [], "authors": ["Bin Xie", "Rui Shao", "Gongwei Chen", "Kaiwen Zhou", "Yinchuan Li", "Jie Liu", "Min Zhang", "Liqiang Nie"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "641001c49fccac1e1c0659fe765eba5a30cefd2a7ff5d00c821f79092c45a1df", "submitted_at": "2025-05-22T16:01:06Z", "updated_at": "2025-05-22T16:01:06Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2505.16827v1"}, "publication": {"authors": ["Bin Xie", "Rui Shao", "Gongwei Chen", "Kaiwen Zhou", "Yinchuan Li", "Jie Liu", "Min Zhang", "Liqiang Nie"], "date_basis": "Explicit publisher citation_publication_date: 2025/7. Preserve supplied precision; do not substitute conference event dates.", "date_parts": [2025, 7], "doi": "10.18653/v1/2025.acl-long.282", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2025.acl-long.282/"], "identity_basis": "Same work: normalized title, ordered author correspondence and the specific contribution in the complete publisher abstract were reviewed against the retained arXiv identity.", "issue": null, "notes": "Publisher proceedings edition; retained separately from the arXiv snapshot.", "pages": "5650-5667", "pdf_url": "https://aclanthology.org/2025.acl-long.282.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "GUI-explorer: Autonomous Exploration and Mining of Transition-aware Knowledge for GUI Agent", "type": "conference-paper", "url": "https://aclanthology.org/2025.acl-long.282/", "venue": "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)", "volume": null}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:09:59Z", "role": "primary", "sha256": "40babce7d310777fa65a498c170320124d4d4941f2cb46ad0ff04aae3583e02f", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.publisher", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue"], "url": "https://aclanthology.org/2025.acl-long.282/"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2505.16827", "identifiers": {"arxiv": "2505.16827"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2505.16827"}, {"rel": "publication", "url": "https://aclanthology.org/2025.acl-long.282/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2025.acl-long.282.pdf"}], "name": "GUI-explorer: Autonomous Exploration and Mining of Transition-aware Knowledge for GUI Agent", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-05-22", "relations": [], "review_status": "curated", "scope": "core", "summary": "GUI-explorer generates exploration goals from screenshots and interface structure, then mines observation-action-outcome triples for app-specific transition knowledge. The resulting training-free agent uses this knowledge during mobile task execution, evaluated on SPA-Bench and AndroidWorld.", "topics": ["mobile-navigation", "data-generation", "memory", "planning"], "updated_at": "2025-05-22", "url": "https://arxiv.org/abs/2505.16827"}
{"aliases": [], "authors": ["Danyang Zhang", "Situo Zhang", "Ziyue Yang", "Zichen Zhu", "Zihan Zhao", "Ruisheng Cao", "Lu Chen", "Kai Yu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "40cbdc37216c58acd1b7523c48ad735df9bde485405a4e7b439aec85730c5bc0", "submitted_at": "2025-05-23T17:23:11Z", "updated_at": "2025-05-23T17:23:11Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2505.18121v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2505.18121", "identifiers": {"arxiv": "2505.18121"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2505.18121"}], "name": "ProgRM: Build Better GUI Agents with Progress Rewards", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-05-23", "relations": [], "review_status": "curated", "scope": "core", "summary": "ProgRM predicts per-step task completion progress to provide intermediate rewards during online GUI-agent training. It derives progress labels by using longest common subsequences to identify key trajectory steps, allowing useful partial progress to contribute even when a trajectory ultimately fails.", "topics": ["reinforcement-learning", "data-generation"], "updated_at": "2025-05-23", "url": "https://arxiv.org/abs/2505.18121"}
{"aliases": [], "authors": ["Runliang Niu", "Jinglong Ji", "Yi Chang", "Qi Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "bb782466eccc7a271c89a4161c7b142f93f8f0d585cad39a1ccad048cf514426", "submitted_at": "2025-05-25T11:13:03Z", "updated_at": "2025-05-25T11:13:03Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2505.19095v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T07:57:47Z", "role": "primary", "sha256": "27144160e8981bb9f36b60c391689774913ff3f026eeeffc2923154c404a6529", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2504.03206%2C2504.03601%2C2504.06821%2C2504.09723%2C2504.11301%2C2504.11543%2C2504.12609%2C2504.14233%2C2504.14348%2C2504.14603%2C2504.16073%2C2504.16419%2C2504.17892%2C2504.18580%2C2504.20965%2C2505.00019%2C2505.00693%2C2505.01578%2C2505.02094%2C2505.03570%2C2505.03574%2C2505.05026%2C2505.05849%2C2505.06507%2C2505.07003%2C2505.07473%2C2505.10468%2C2505.10887%2C2505.12053%2C2505.12490%2C2505.12728%2C2505.13291%2C2505.14030%2C2505.14454%2C2505.14646%2C2505.15801%2C2505.15816%2C2505.17020%2C2505.18746%2C2505.19095%2C2505.19713%2C2505.20023%2C2505.20732%2C2505.20795%2C2505.21055%2C2505.22304%2C2505.22429%2C2505.23450%2C2505.23559%2C2506.00641&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:20:24Z", "role": "primary", "sha256": "645bd2ecbc013dab3cbc601d5a42011d8f7306915ea2ae06baf0767b3962e9ed", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/abs/2505.19095v1"}], "id": "arxiv:2505.19095", "identifiers": {"arxiv": "2505.19095"}, "kind": "paper", "last_verified_at": "2026-10-17T11:20:24Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2505.19095"}], "name": "ScreenExplorer: Training a Vision-Language Model for Diverse Exploration in Open GUI World", "notes": "Reviewed against the exact-version official arXiv abstract page; its complete title, ordered author list, visible version, and normalized abstract SHA match the retained snapshot. The original API response body was not locally available. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2025-05-25", "relations": [], "review_status": "curated", "scope": "core", "summary": "Trains ScreenExplorer with reinforcement learning in dynamic GUI environments, using a world-model curiosity reward and experience-stream distillation to support sustained exploration.", "topics": ["reinforcement-learning", "world-models", "data-generation"], "updated_at": "2025-05-25", "url": "https://arxiv.org/abs/2505.19095"}
{"aliases": [], "authors": ["Minda Hu", "Tianqing Fang", "Jianshu Zhang", "Junyu Ma", "Zhisong Zhang", "Jingyan Zhou", "Hongming Zhang", "Haitao Mi", "Dong Yu", "Irwin King"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "171071f26a6dc8b9c8851d2f11d60b113983191c905631e431f352711e0d0e26", "submitted_at": "2025-05-26T14:03:37Z", "updated_at": "2025-09-18T11:32:15Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2505.20013v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2505.20013", "identifiers": {"arxiv": "2505.20013"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2505.20013"}], "name": "WebCoT: Enhancing Web Agent Reasoning by Reconstructing Chain-of-Thought in Reflection, Branching, and Rollback", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-05-26", "relations": [], "review_status": "curated", "scope": "core", "summary": "WebCoT turns reflection, look-ahead, branching and rollback procedures into chain-of-thought training rationales. Fine-tuning on these reconstructed trajectories is evaluated for web-agent self-improvement and transfer across navigation tasks.", "topics": ["data-generation", "planning", "self-reflection", "web-navigation"], "updated_at": "2025-09-18", "url": "https://arxiv.org/abs/2505.20013"}
{"aliases": [], "authors": ["Qinzhuo Wu", "Pengzhi Gao", "Wei Liu", "Jian Luan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d2d51388193e19501b580740969a6e08661c6e0a2cf09e2489c46fe4f19209cd", "submitted_at": "2025-05-27T03:09:06Z", "updated_at": "2025-05-27T03:09:06Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2505.20660v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2505.20660", "identifiers": {"arxiv": "2505.20660"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2505.20660"}], "name": "BacktrackAgent: Enhancing GUI Agent with Error Detection and Backtracking Mechanism", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-05-27", "relations": [], "review_status": "curated", "scope": "core", "summary": "BacktrackAgent adds verification, judgment, and reflection modules to detect erroneous GUI actions and guide recovery. It develops backtracking-oriented training data that includes post-action pages and evaluates task completion and step accuracy on Mobile3M and Auto-UI.", "topics": ["robustness", "planning", "data-generation", "mobile-navigation"], "updated_at": "2025-05-27", "url": "https://arxiv.org/abs/2505.20660"}
{"aliases": [], "authors": ["Han Xiao", "Guozhi Wang", "Yuxiang Chai", "Zimu Lu", "Weifeng Lin", "Hao He", "Lue Fan", "Liuyang Bian", "Rui Hu", "Liang Liu", "Shuai Ren", "Yafei Wen", "Xiaoxin Chen", "Aojun Zhou", "Hongsheng Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6f19671a527de76b702e68311656e42feea88249716f68911a41ec80be205e4c", "submitted_at": "2025-05-27T17:58:06Z", "updated_at": "2025-05-27T17:58:06Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2505.21496v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2505.21496", "identifiers": {"arxiv": "2505.21496"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2505.21496"}], "name": "UI-Genie: A Self-Improving Approach for Iteratively Boosting MLLM-based Mobile GUI Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-05-27", "relations": [], "review_status": "curated", "scope": "core", "summary": "UI-Genie jointly improves a mobile GUI agent and a reward model that processes screenshots with interaction history and supplies action- and task-level feedback. Its data pipeline uses rule-based verification, corrupted trajectories, and hard negatives to support reward-guided exploration and iterative training.", "topics": ["data-generation", "reinforcement-learning", "mobile-navigation", "interaction-history"], "updated_at": "2025-05-27", "url": "https://arxiv.org/abs/2505.21496"}
{"aliases": [], "authors": ["Chenyu Yang", "Shiqian Su", "Shi Liu", "Xuan Dong", "Yue Yu", "Weijie Su", "Xuehui Wang", "Zhaoyang Liu", "Jinguo Zhu", "Hao Li", "Wenhai Wang", "Yu Qiao", "Xizhou Zhu", "Jifeng Dai"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "799e7b387f7d281f0aa9b182f37b215e3baca2041a0a0f6d414ddf1cec0de003", "submitted_at": "2025-05-29T17:59:51Z", "updated_at": "2025-05-29T17:59:51Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2505.23762v1"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:42Z", "role": "primary", "sha256": "afe4e75ffb8a5c44f4c088a90ffa5f53553036f514aa768f5aee0c5aa38d731e", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2503.16788%2C2503.17709%2C2503.18492%2C2503.19537%2C2503.20978%2C2503.21620%2C2503.23434%2C2504.01382%2C2504.07981%2C2504.10445%2C2504.10458%2C2504.11257%2C2504.11281%2C2504.12679%2C2504.13805%2C2504.13865%2C2504.14239%2C2504.18575%2C2504.19838%2C2504.20464%2C2505.10924%2C2505.11891%2C2505.12299%2C2505.12632%2C2505.13227%2C2505.13887%2C2505.13909%2C2505.14289%2C2505.15810%2C2505.16282%2C2505.16827%2C2505.18829%2C2505.19897%2C2505.21496%2C2505.21499%2C2505.21936%2C2505.22648%2C2505.23762%2C2505.24838%2C2506.00618%2C2506.01391%2C2506.01968%2C2506.02456%2C2506.03143%2C2506.03569%2C2506.04135%2C2506.04614%2C2506.07672%2C2506.08012%2C2506.08972&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:42Z", "role": "primary", "sha256": "afe4e75ffb8a5c44f4c088a90ffa5f53553036f514aa768f5aee0c5aa38d731e", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2503.16788%2C2503.17709%2C2503.18492%2C2503.19537%2C2503.20978%2C2503.21620%2C2503.23434%2C2504.01382%2C2504.07981%2C2504.10445%2C2504.10458%2C2504.11257%2C2504.11281%2C2504.12679%2C2504.13805%2C2504.13865%2C2504.14239%2C2504.18575%2C2504.19838%2C2504.20464%2C2505.10924%2C2505.11891%2C2505.12299%2C2505.12632%2C2505.13227%2C2505.13887%2C2505.13909%2C2505.14289%2C2505.15810%2C2505.16282%2C2505.16827%2C2505.18829%2C2505.19897%2C2505.21496%2C2505.21499%2C2505.21936%2C2505.22648%2C2505.23762%2C2505.24838%2C2506.00618%2C2506.01391%2C2506.01968%2C2506.02456%2C2506.03143%2C2506.03569%2C2506.04135%2C2506.04614%2C2506.07672%2C2506.08012%2C2506.08972&max_results=50"}], "id": "arxiv:2505.23762", "identifiers": {"arxiv": "2505.23762"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:42Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2505.23762"}], "name": "ZeroGUI: Automating Online GUI Learning at Zero Human Cost", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-05-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "Combines generated tasks, estimated success rewards, and online reinforcement learning to automate GUI-agent interaction training.", "topics": ["reinforcement-learning", "data-generation"], "updated_at": "2025-05-29", "url": "https://arxiv.org/abs/2505.23762"}
{"aliases": [], "authors": ["Brandon Man", "Ghadi Nehme", "Md Ferdous Alam", "Faez Ahmed"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "f913b87f07fdad0e4c45264c58c8d9b9d13ca26695306d9e0644eb856f18ced7", "submitted_at": "2025-05-30T17:39:52Z", "updated_at": "2025-11-08T18:27:03Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2505.24838v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:42Z", "role": "primary", "sha256": "afe4e75ffb8a5c44f4c088a90ffa5f53553036f514aa768f5aee0c5aa38d731e", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2503.16788%2C2503.17709%2C2503.18492%2C2503.19537%2C2503.20978%2C2503.21620%2C2503.23434%2C2504.01382%2C2504.07981%2C2504.10445%2C2504.10458%2C2504.11257%2C2504.11281%2C2504.12679%2C2504.13805%2C2504.13865%2C2504.14239%2C2504.18575%2C2504.19838%2C2504.20464%2C2505.10924%2C2505.11891%2C2505.12299%2C2505.12632%2C2505.13227%2C2505.13887%2C2505.13909%2C2505.14289%2C2505.15810%2C2505.16282%2C2505.16827%2C2505.18829%2C2505.19897%2C2505.21496%2C2505.21499%2C2505.21936%2C2505.22648%2C2505.23762%2C2505.24838%2C2506.00618%2C2506.01391%2C2506.01968%2C2506.02456%2C2506.03143%2C2506.03569%2C2506.04135%2C2506.04614%2C2506.07672%2C2506.08012%2C2506.08972&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:42Z", "role": "primary", "sha256": "afe4e75ffb8a5c44f4c088a90ffa5f53553036f514aa768f5aee0c5aa38d731e", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2503.16788%2C2503.17709%2C2503.18492%2C2503.19537%2C2503.20978%2C2503.21620%2C2503.23434%2C2504.01382%2C2504.07981%2C2504.10445%2C2504.10458%2C2504.11257%2C2504.11281%2C2504.12679%2C2504.13805%2C2504.13865%2C2504.14239%2C2504.18575%2C2504.19838%2C2504.20464%2C2505.10924%2C2505.11891%2C2505.12299%2C2505.12632%2C2505.13227%2C2505.13887%2C2505.13909%2C2505.14289%2C2505.15810%2C2505.16282%2C2505.16827%2C2505.18829%2C2505.19897%2C2505.21496%2C2505.21499%2C2505.21936%2C2505.22648%2C2505.23762%2C2505.24838%2C2506.00618%2C2506.01391%2C2506.01968%2C2506.02456%2C2506.03143%2C2506.03569%2C2506.04135%2C2506.04614%2C2506.07672%2C2506.08012%2C2506.08972&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T12:02:34.311949+00:00", "role": "primary", "sha256": "a303d95a811331cd6ee23b0340f1b9ef2dd996835bfbc66656a871ff94eead50", "supports": ["environments", "notes"], "url": "https://arxiv.org/html/2505.24838v2"}], "id": "arxiv:2505.24838", "identifiers": {"arxiv": "2505.24838"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:42Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2505.24838"}], "name": "VideoCAD: A Dataset and Model for Learning Long-Horizon 3D CAD UI Interactions from Video", "notes": "The contribution was reviewed from the retained abstract. Sections 1 and 3 of the exact-version v2 paper identify Onshape as the browser-based interaction environment; the use of pixel-level input does not make the operated surface a desktop application. No experimental results were independently reproduced.", "published_at": "2025-05-30", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces synthetic CAD interaction videos, a model for learning interface actions from video, and evaluation tasks for spatial and long-horizon reasoning.", "topics": ["video", "data-generation", "science", "long-horizon"], "updated_at": "2025-11-08", "url": "https://arxiv.org/abs/2505.24838"}
{"aliases": [], "authors": ["Mathieu Andreux", "Breno Baldas Skuk", "Hamza Benchekroun", "Emilien Biré", "Antoine Bonnet", "Riaz Bordie", "Nathan Bout", "Matthias Brunel", "Pierre-Louis Cedoz", "Antoine Chassang", "Mickaël Chen", "Alexandra D. Constantinou", "Antoine d'Andigné", "Hubert de La Jonquière", "Aurélien Delfosse", "Ludovic Denoyer", "Alexis Deprez", "Augustin Derupti", "Michael Eickenberg", "Mathïs Federico", "Charles Kantor", "Xavier Koegler", "Yann Labbé", "Matthew C. H. Lee", "Erwan Le Jumeau de Kergaradec", "Amir Mahla", "Avshalom Manevich", "Adrien Maret", "Charles Masson", "Rafaël Maurin", "Arturo Mena", "Philippe Modard", "Axel Moyal", "Axel Nguyen Kerbel", "Julien Revelle", "Mats L. Richter", "María Santos", "Laurent Sifre", "Maxime Theillard", "Marc Thibault", "Louis Thiry", "Léo Tronchon", "Nicolas Usunier", "Tony Wu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "89c2d658e03204b56f6d273da9572820f9a23ca359a441ea87baa451a93c5b58", "submitted_at": "2025-06-03T13:29:03Z", "updated_at": "2025-06-11T09:33:38Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2506.02865v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2506.02865", "identifiers": {"arxiv": "2506.02865"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2506.02865"}], "name": "Surfer-H Meets Holo1: Cost-Efficient Web Agent Powered by Open Weights", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-06-03", "relations": [], "review_status": "curated", "scope": "core", "summary": "Describes Surfer-H, a web agent paired with the Holo1 vision-language model family trained on web content, synthetic examples, and agent-generated data. The work introduces WebClick for web-interface localization and evaluates browser task execution on WebVoyager.", "topics": ["web-navigation", "grounding", "data-generation", "evaluation"], "updated_at": "2025-06-11", "url": "https://arxiv.org/abs/2506.02865"}
{"aliases": [], "authors": ["Man Luo", "David Cobbley", "Xin Su", "Shachar Rosenman", "Vasudev Lal", "Shao-Yen Tseng", "Phillip Howard"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "068eae2cab9899967850f21616a701f3eeb258b9654c83d34f68be266e48854e", "submitted_at": "2025-06-03T17:27:04Z", "updated_at": "2025-06-03T17:27:04Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2506.03095v1"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2506.03095", "identifiers": {"arxiv": "2506.03095"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2506.03095"}], "name": "DPO Learning with LLMs-Judge Signal for Computer Use Agents", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-06-03", "relations": [], "review_status": "curated", "scope": "core", "summary": "Uses an LLM judge to evaluate and filter synthetic GUI interaction trajectories for training a compact vision-language agent. OSWorld experiments assess the policy designed for local execution.", "topics": ["data-generation", "evaluation", "efficiency"], "updated_at": "2025-06-03", "url": "https://arxiv.org/abs/2506.03095"}
{"aliases": [], "authors": ["Apurva Gandhi", "Graham Neubig"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2bccee69e390b5170ff264e689c209d13317e7d3be9b37e6d3fbed0e0d089943", "submitted_at": "2025-06-04T03:27:56Z", "updated_at": "2026-03-02T22:31:00Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2506.03533v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2506.03533", "identifiers": {"arxiv": "2506.03533"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2506.03533"}], "name": "Go-Browse: Training Web Agents with Structured Exploration", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-06-04", "relations": [], "review_status": "curated", "scope": "core", "summary": "Go-Browse frames web-environment exploration as graph search so that information can be reused across data-collection episodes. It collects 10,000 successful trajectories and 40,000 interaction steps on WebArena and uses them to fine-tune a web agent.", "topics": ["data-generation", "web-navigation"], "updated_at": "2026-03-02", "url": "https://arxiv.org/abs/2506.03533"}
{"aliases": [], "authors": ["Penghao Wu", "Shengnan Ma", "Bo Wang", "Jiaheng Yu", "Lewei Lu", "Ziwei Liu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "ce3b7d2c25b5a99f110f00be7389d4230d625ac1f7f33d5ab81f3d6f441f3610", "submitted_at": "2025-06-09T17:59:57Z", "updated_at": "2025-06-09T17:59:57Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2506.08012v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:42Z", "role": "primary", "sha256": "afe4e75ffb8a5c44f4c088a90ffa5f53553036f514aa768f5aee0c5aa38d731e", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2503.16788%2C2503.17709%2C2503.18492%2C2503.19537%2C2503.20978%2C2503.21620%2C2503.23434%2C2504.01382%2C2504.07981%2C2504.10445%2C2504.10458%2C2504.11257%2C2504.11281%2C2504.12679%2C2504.13805%2C2504.13865%2C2504.14239%2C2504.18575%2C2504.19838%2C2504.20464%2C2505.10924%2C2505.11891%2C2505.12299%2C2505.12632%2C2505.13227%2C2505.13887%2C2505.13909%2C2505.14289%2C2505.15810%2C2505.16282%2C2505.16827%2C2505.18829%2C2505.19897%2C2505.21496%2C2505.21499%2C2505.21936%2C2505.22648%2C2505.23762%2C2505.24838%2C2506.00618%2C2506.01391%2C2506.01968%2C2506.02456%2C2506.03143%2C2506.03569%2C2506.04135%2C2506.04614%2C2506.07672%2C2506.08012%2C2506.08972&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:42Z", "role": "primary", "sha256": "afe4e75ffb8a5c44f4c088a90ffa5f53553036f514aa768f5aee0c5aa38d731e", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2503.16788%2C2503.17709%2C2503.18492%2C2503.19537%2C2503.20978%2C2503.21620%2C2503.23434%2C2504.01382%2C2504.07981%2C2504.10445%2C2504.10458%2C2504.11257%2C2504.11281%2C2504.12679%2C2504.13805%2C2504.13865%2C2504.14239%2C2504.18575%2C2504.19838%2C2504.20464%2C2505.10924%2C2505.11891%2C2505.12299%2C2505.12632%2C2505.13227%2C2505.13887%2C2505.13909%2C2505.14289%2C2505.15810%2C2505.16282%2C2505.16827%2C2505.18829%2C2505.19897%2C2505.21496%2C2505.21499%2C2505.21936%2C2505.22648%2C2505.23762%2C2505.24838%2C2506.00618%2C2506.01391%2C2506.01968%2C2506.02456%2C2506.03143%2C2506.03569%2C2506.04135%2C2506.04614%2C2506.07672%2C2506.08012%2C2506.08972&max_results=50"}], "id": "arxiv:2506.08012", "identifiers": {"arxiv": "2506.08012"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:42Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2506.08012"}], "name": "GUI-Reflection: Empowering Multimodal GUI Models with Self-Reflection Behavior", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-06-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "GUI-Reflection synthesizes reflection and error-correction examples from successful trajectories and adds dedicated learning stages for these behaviors. It introduces a reflection task suite, a mobile environment for online data collection, and iterative reflection tuning.", "topics": ["robustness", "data-generation", "evaluation", "mobile-navigation"], "updated_at": "2025-06-09", "url": "https://arxiv.org/abs/2506.08012"}
{"aliases": [], "authors": ["Wenkang Han", "Zhixiong Zeng", "Jing Huang", "Shu Jiang", "Liming Zheng", "Longrong Yang", "Haibo Qiu", "Chang Yao", "Jingyuan Chen", "Lin Ma"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "3cce7c2e1623ec841e8afed03c2e8e0c6f969f3dd05d53ddc4ef11e74310c21c", "submitted_at": "2025-06-10T12:16:27Z", "updated_at": "2025-11-26T07:48:14Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2506.11127v3"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2506.11127", "identifiers": {"arxiv": "2506.11127"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2506.11127"}], "name": "UITron-Speech: Towards Automated GUI Agents Based on Speech Instructions", "notes": "Reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2025-06-10", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces UITron-Speech, which predicts GUI actions from spoken instructions and screenshots. It combines synthetic speech data and mixed-modality training with a training-free grounding refinement step.", "topics": ["grounding", "action-prediction", "data-generation"], "updated_at": "2025-11-26", "url": "https://arxiv.org/abs/2506.11127"}
{"aliases": [], "authors": ["Jingxu Xie", "Dylan Xu", "Xuandong Zhao", "Dawn Song"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "ae36b271f9b4799093831cabee772845edfaeca97107414e424e170ccf89904d", "submitted_at": "2025-06-17T05:46:52Z", "updated_at": "2026-03-02T00:51:36Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2506.14205v2"}}, "environments": ["desktop", "web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:14:13+00:00", "role": "primary", "sha256": "42e8b8dead7caea3f8b21e93a075e8c895dfc30a25e1470bf97a90940bbf0106", "supports": ["scope", "environments"], "url": "https://raw.githubusercontent.com/sunblaze-ucb/AgentSynth/HEAD/README.md"}], "id": "arxiv:2506.14205", "identifiers": {"arxiv": "2506.14205"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2506.14205"}], "name": "AgentSynth: Scalable Task Generation for Generalist Computer-Use Agents", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-06-17", "relations": [], "review_status": "curated", "scope": "core", "summary": "AgentSynth composes easily generated subtasks into longer computer-use tasks and trajectories with adjustable difficulty. Its evaluation uses the number of subtasks to probe how agent success changes as task complexity grows.", "topics": ["data-generation", "evaluation", "long-horizon"], "updated_at": "2026-03-02", "url": "https://arxiv.org/abs/2506.14205"}
{"aliases": [], "authors": ["Jingqi Yang", "Zhilong Song", "Jiawei Chen", "Mingli Song", "Sheng Zhou", "linjun sun", "Xiaogang Ouyang", "Chun Chen", "Can Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "f9e64118f03c2a9f3f55e35f79c38835f1eb823265c83c933fdceca2610b9904", "submitted_at": "2025-06-17T12:50:35Z", "updated_at": "2025-06-17T12:50:35Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2506.14477v1"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2506.14477", "identifiers": {"arxiv": "2506.14477"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2506.14477"}], "name": "GUI-Robust: A Comprehensive Dataset for Testing GUI Agent Robustness in Real-World Anomalies", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-06-17", "relations": [], "review_status": "curated", "scope": "core", "summary": "Builds GUI task data with real-world interaction anomalies and evaluates how agent behavior changes under those conditions.", "topics": ["robustness", "evaluation", "data-generation"], "updated_at": "2025-06-17", "url": "https://arxiv.org/abs/2506.14477"}
{"aliases": [], "authors": ["Xinzge Gao", "Chuanrui Hu", "Bin Chen", "Teng Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "aaa62e24a7a269fea49471f17028a3d5d8068d934e450f53bc002ae80fb7cb03", "submitted_at": "2025-06-22T20:17:46Z", "updated_at": "2025-06-22T20:17:46Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2506.18158v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2506.18158", "identifiers": {"arxiv": "2506.18158"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2506.18158"}], "name": "Chain-of-Memory: Enhancing GUI Agents for Cross-Application Navigation", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-06-22", "relations": [], "review_status": "curated", "scope": "core", "summary": "Chain-of-Memory explicitly records action descriptions and task-relevant screen information in short- and long-term memory for GUI navigation. GUI Odyssey-CoM provides 111,000 screen-action pairs with memory annotations for learning and evaluating information retention across applications.", "topics": ["memory", "state-tracking", "cross-app", "data-generation"], "updated_at": "2025-06-22", "url": "https://arxiv.org/abs/2506.18158"}
{"aliases": [], "authors": ["ZongHan Hsieh", "Tzer-Jen Wei", "ShengJing Yang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "bb3678b0d0e645a87ce9392b606c1d67dcbc40656880c57d1cbb8ddd8ede7ea1", "submitted_at": "2025-06-30T03:33:02Z", "updated_at": "2025-07-18T06:03:26Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2506.23491v3"}}, "environments": ["web", "desktop", "cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2506.23491", "identifiers": {"arxiv": "2506.23491"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2506.23491"}], "name": "ZonUI-3B: A Lightweight Vision-Language Model for Cross-Resolution GUI Grounding", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-06-30", "relations": [], "review_status": "curated", "scope": "core", "summary": "ZonUI-3B trains GUI grounding with a curated 24,000-example collection of mobile, desktop, and web screenshots. Its two-stage procedure first learns across platforms and then specializes on high-resolution images, with ablations on data diversity and redundant examples.", "topics": ["grounding", "data-generation", "efficiency"], "updated_at": "2025-07-18", "url": "https://arxiv.org/abs/2506.23491"}
{"aliases": [], "authors": ["Yucheng Shi", "Wenhao Yu", "Zaitang Li", "Yonglin Wang", "Hongming Zhang", "Ninghao Liu", "Haitao Mi", "Dong Yu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "45d36dd32b30bf4faee6d9a72e3cdfe91edff465bbd886bdfc3c27c3da475b82", "submitted_at": "2025-07-08T07:07:53Z", "updated_at": "2025-07-08T07:07:53Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2507.05720v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2507.05720", "identifiers": {"arxiv": "2507.05720"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2507.05720"}], "name": "MobileGUI-RL: Advancing Mobile GUI Agent through Reinforcement Learning in Online Environment", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-07-08", "relations": [], "review_status": "curated", "scope": "core", "summary": "MobileGUI-RL constructs an online training curriculum through agent exploration and task filtering. It adapts GRPO with trajectory-aware advantages and rewards for task completion and execution efficiency, evaluated on three online mobile-agent benchmarks.", "topics": ["mobile-navigation", "reinforcement-learning", "data-generation", "efficiency"], "updated_at": "2025-07-08", "url": "https://arxiv.org/abs/2507.05720"}
{"aliases": [], "authors": ["Qinglong Yang", "Haoming Li", "Haotian Zhao", "Xiaokai Yan", "Jingtao Ding", "Fengli Xu", "Yong Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "947c85efc189635a118b823b8ca0787ff6b1c922d1af8b21a8686b7df9a27cf8", "submitted_at": "2025-06-09T06:38:41Z", "updated_at": "2026-03-16T09:14:01Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2507.21071v2"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:49Z", "role": "primary", "sha256": "c208e36ae75cef4b32cfcb2f8182dd4aaaa218e3119bb3d4a2c20a5832f8b1ee", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2506.10387%2C2506.14866%2C2506.20332%2C2507.00008%2C2507.02592%2C2507.03730%2C2507.04227%2C2507.05445%2C2507.05673%2C2507.05720%2C2507.05791%2C2507.14293%2C2507.14799%2C2507.15846%2C2507.16853%2C2507.19132%2C2507.19478%2C2507.21071%2C2507.22025%2C2507.22358%2C2507.23779%2C2508.00935%2C2508.01330%2C2508.03700%2C2508.03923%2C2508.04010%2C2508.04026%2C2508.04482%2C2508.04700%2C2508.05615%2C2508.05731%2C2508.09057%2C2508.09123%2C2508.13167%2C2508.14040%2C2508.15144%2C2508.18689%2C2508.19679%2C2508.20096%2C2509.02544%2C2509.03891%2C2509.06477%2C2509.10723%2C2509.11250%2C2509.11543%2C2509.15221%2C2509.15566%2C2509.15738%2C2509.18119%2C2510.01354&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T17:11:36Z", "role": "primary", "sha256": "90a03448678a584f29ef11ae2afe1ef7040b4f3658c492f6e48eafe8fb6d2fa2", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?id_list=2506.10387%2C2506.14866%2C2506.20332%2C2507.00008%2C2507.02592%2C2507.03730%2C2507.04227%2C2507.05445%2C2507.05673%2C2507.05720%2C2507.05791%2C2507.14293%2C2507.14799%2C2507.15846%2C2507.16853%2C2507.19132%2C2507.19478%2C2507.21071%2C2507.22025%2C2507.22358%2C2507.23779%2C2508.00935%2C2508.01330%2C2508.03700%2C2508.03923%2C2508.04010%2C2508.04026%2C2508.04482%2C2508.04700%2C2508.05615%2C2508.05731%2C2508.09057%2C2508.09123%2C2508.13167%2C2508.14040%2C2508.15144%2C2508.18689%2C2508.19679%2C2508.20096%2C2509.02544%2C2509.03891%2C2509.06477%2C2509.10723%2C2509.11250%2C2509.11543%2C2509.15221%2C2509.15566%2C2509.15738%2C2509.18119%2C2510.01354&max_results=50"}], "id": "arxiv:2507.21071", "identifiers": {"arxiv": "2507.21071"}, "kind": "paper", "last_verified_at": "2026-10-17T17:11:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2507.21071"}], "name": "FingerTip 20K: A Benchmark for Proactive and Personalized Mobile LLM Agents", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2025-06-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "FingerTip 20K gathers 20K human Android demonstrations from users' long-term everyday app use, with user context, to evaluate proactive task suggestion and preference-aware task execution.", "topics": ["evaluation", "data-generation", "mobile-navigation", "interaction-history"], "updated_at": "2026-03-16", "url": "https://arxiv.org/abs/2507.21071"}
{"aliases": [], "authors": ["Yuhan Guo", "Cong Guo", "Aiwen Sun", "Hongliang He", "Xinyu Yang", "Yue Lu", "Yingji Zhang", "Xuntao Guo", "Dong Zhang", "Jianzhuang Liu", "Jiang Duan", "Yijia Xiao", "Liangjian Wen", "Hai-Ming Xu", "Yong Dai"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "7406b48ed9826d51215d955c31c3c67cd2ce9e08bf9a93a55ac4e85a546315f1", "submitted_at": "2025-08-03T17:17:52Z", "updated_at": "2026-07-05T06:17:25Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2508.01858v3"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:14:13+00:00", "role": "primary", "sha256": "1c8ad21928ad5e15bba474a060d225b7da000b37878a8e98dfe5ce94008e3086", "supports": ["scope", "environments"], "url": "https://raw.githubusercontent.com/Gnonymous/Web-CogReasoner/HEAD/README.md"}], "id": "arxiv:2508.01858", "identifiers": {"arxiv": "2508.01858"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2508.01858"}], "name": "Web-CogReasoner: Towards Multimodal Knowledge-Induced Cognitive Reasoning for Web Agents", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-08-03", "relations": [], "review_status": "curated", "scope": "core", "summary": "Web-CogReasoner separates factual and conceptual web knowledge from procedural reasoning, using a structured curriculum to train a multimodal agent. Web-CogDataset and Web-CogBench support study of these capabilities alongside web-navigation evaluation.", "topics": ["data-generation", "planning", "evaluation", "web-navigation"], "updated_at": "2026-07-05", "url": "https://arxiv.org/abs/2508.01858"}
{"aliases": [], "authors": ["Zhihao Luo", "Wentao Yan", "Jingyu Gong", "Min Wang", "Zhizhong Zhang", "Xuhong Wang", "Yuan Xie", "Xin Tan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e892ab93511f4164f37ef79ac0473e3c1eed7d6ec11fb6c8ec82092f26c5ca68", "submitted_at": "2025-08-04T04:28:18Z", "updated_at": "2026-05-02T12:32:44Z", "version": 4, "versioned_url": "https://arxiv.org/abs/2508.02046v4"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-18T04:21:52Z", "role": "primary", "sha256": "c194d15b7e422aae48b4e05216a1f1d84471b6666718b9d3dd4c9d250c356e77", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2210.03347%2C2403.04984%2C2405.07090%2C2406.06282%2C2409.00088%2C2409.10811%2C2411.10943%2C2503.09241%2C2503.20279%2C2503.23350%2C2505.05440%2C2505.10593%2C2505.16422%2C2505.21279%2C2505.24878%2C2506.02153%2C2506.02839%2C2506.05079%2C2506.07976%2C2508.02046%2C2511.07885%2C2511.11248%2C2512.21720%2C2603.20351%2C2604.19299&max_results=25"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:21:52Z", "role": "primary", "sha256": "c194d15b7e422aae48b4e05216a1f1d84471b6666718b9d3dd4c9d250c356e77", "supports": ["name", "kind", "scope", "environments", "topics", "summary", "notes"], "url": "https://export.arxiv.org/api/query?id_list=2210.03347%2C2403.04984%2C2405.07090%2C2406.06282%2C2409.00088%2C2409.10811%2C2411.10943%2C2503.09241%2C2503.20279%2C2503.23350%2C2505.05440%2C2505.10593%2C2505.16422%2C2505.21279%2C2505.24878%2C2506.02153%2C2506.02839%2C2506.05079%2C2506.07976%2C2508.02046%2C2511.07885%2C2511.11248%2C2512.21720%2C2603.20351%2C2604.19299&max_results=25"}], "id": "arxiv:2508.02046", "identifiers": {"arxiv": "2508.02046"}, "kind": "paper", "last_verified_at": "2026-10-18T04:21:52Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2508.02046"}], "name": "NaviMaster: Learning a Unified Policy for GUI and Embodied Navigation Tasks", "notes": "Reviewed against the complete official arXiv abstract. Experimental results and linked artifact availability have not been independently reproduced.", "published_at": "2025-08-04", "relations": [], "review_status": "curated", "scope": "core", "summary": "NaviMaster trains a shared policy for GUI and embodied navigation using a common visual-trajectory formulation. It combines trajectory collection, reinforcement learning and distance-aware rewards, with evaluation across both task families.", "topics": ["reinforcement-learning", "data-generation", "robustness"], "updated_at": "2026-05-02", "url": "https://arxiv.org/abs/2508.02046"}
{"aliases": [], "authors": ["Liujian Tang", "Shaokang Dong", "Yijia Huang", "Minqi Xiang", "Hongtao Ruan", "Bin Wang", "Shuo Li", "Zhiheng Xi", "Zhihui Cao", "Hailiang Pang", "Heng Kong", "He Yang", "Mingxu Chai", "Zhilin Gao", "Xingyu Liu", "Yingnan Fu", "Jiaming Liu", "Xuanjing Huang", "Yu-Gang Jiang", "Tao Gui", "Qi Zhang", "Kang Wang", "Yunke Zhang", "Yuran Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d5175e5ed0061660fa950512300daa878eb90e5c98bdac8f80d478414b0a2931", "submitted_at": "2025-07-19T12:33:43Z", "updated_at": "2025-09-11T14:28:11Z", "version": 5, "versioned_url": "https://arxiv.org/abs/2508.03700v5"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2508.03700", "identifiers": {"arxiv": "2508.03700"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2508.03700"}], "name": "MagicGUI: A Foundational Mobile GUI Agent with Scalable Data Pipeline and Reinforcement Fine-tuning", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-07-19", "relations": [], "review_status": "curated", "scope": "core", "summary": "Combines a mobile GUI data pipeline, perception and planning capabilities, and reinforcement fine-tuning for interactive task execution.", "topics": ["mobile-navigation", "data-generation", "reinforcement-learning"], "updated_at": "2025-09-11", "url": "https://arxiv.org/abs/2508.03700"}
{"aliases": [], "authors": ["Yuhao Cheng", "Liang Tang", "Shuxian Li", "Yukang Huo", "Tiaonan Duan", "Kaer Huang", "Yanzhe Jing", "Yiqiang Yan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d54ad74695a60cb040d88fcb85549912d7d4ecf6ca2ed78e636d05f8f17d1795", "submitted_at": "2025-08-06T02:57:22Z", "updated_at": "2026-01-22T05:22:03Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2508.04037v2"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2508.04037", "identifiers": {"arxiv": "2508.04037"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2508.04037"}], "name": "Evolving in Tasks: Empowering the Multi-modality Large Language Model as the Computer Use Agent", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-08-06", "relations": [], "review_status": "curated", "scope": "core", "summary": "The Self-Evolution Agent combines automatic generation of verifiable computer-use trajectories with step-wise reinforcement learning for long-horizon training. It also describes a model-enhancement procedure that integrates grounding and planning capabilities without an additional training stage.", "topics": ["data-generation", "reinforcement-learning", "grounding", "planning"], "updated_at": "2026-01-22", "url": "https://arxiv.org/abs/2508.04037"}
{"aliases": [], "authors": ["Zeyi Sun", "Ziyu Liu", "Yuhang Zang", "Yuhang Cao", "Xiaoyi Dong", "Tong Wu", "Dahua Lin", "Jiaqi Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "268c38e6f11f11771a1251fc522112ff052557b5753dff6cadf2ba49c0ec04b2", "submitted_at": "2025-08-06T17:58:46Z", "updated_at": "2025-08-12T15:11:53Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2508.04700v2"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2508.04700", "identifiers": {"arxiv": "2508.04700"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2508.04700"}], "name": "SEAgent: Self-Evolving Computer Use Agent with Autonomous Learning from Experience", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-08-06", "relations": [], "review_status": "curated", "scope": "core", "summary": "SEAgent explores unfamiliar software using a world-state assessor and an automatically generated task curriculum. It learns from failed and successful interactions through adversarial imitation and GRPO, then combines specialist experience into a generalist agent evaluated on five OSWorld software environments.", "topics": ["reinforcement-learning", "data-generation", "world-models", "imitation-learning"], "updated_at": "2025-08-12", "url": "https://arxiv.org/abs/2508.04700"}
{"aliases": [], "authors": ["Xinyuan Wang", "Bowen Wang", "Dunjie Lu", "Junlin Yang", "Tianbao Xie", "Junli Wang", "Jiaqi Deng", "Xiaole Guo", "Yiheng Xu", "Chen Henry Wu", "Zhennan Shen", "Zhuokai Li", "Ryan Li", "Xiaochuan Li", "Junda Chen", "Boyuan Zheng", "Peihang Li", "Fangyu Lei", "Ruisheng Cao", "Yeqiao Fu", "Dongchan Shin", "Martin Shin", "Jiarui Hu", "Yuyan Wang", "Jixuan Chen", "Yuxiao Ye", "Danyang Zhang", "Dikang Du", "Hao Hu", "Huarong Chen", "Zaida Zhou", "Haotian Yao", "Ziwei Chen", "Qizheng Gu", "Yipu Wang", "Heng Wang", "Diyi Yang", "Victor Zhong", "Flood Sung", "Y. Charles", "Zhilin Yang", "Tao Yu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "40e46f27128d066358fcee6279b26bb2b708dacaf1f7b5794b755520fe892243", "submitted_at": "2025-08-12T17:52:32Z", "updated_at": "2025-10-04T17:38:02Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2508.09123v3"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2508.09123", "identifiers": {"arxiv": "2508.09123"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2508.09123"}], "name": "OpenCUA: Open Foundations for Computer-Use Agents", "notes": "Editorial summary based on the primary paper abstract; experiments and released artifacts have not been independently reproduced.", "published_at": "2025-08-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "Presents demonstration collection infrastructure, the AgentNet dataset, and a training pipeline for open computer-use models.", "topics": ["data-generation", "planning"], "updated_at": "2025-10-04", "url": "https://arxiv.org/abs/2508.09123"}
{"aliases": [], "authors": ["Jiabo Ye", "Xi Zhang", "Haiyang Xu", "Haowei Liu", "Junyang Wang", "Zhaoqing Zhu", "Ziwei Zheng", "Feiyu Gao", "Junjie Cao", "Zhengxi Lu", "Jitong Liao", "Qi Zheng", "Fei Huang", "Jingren Zhou", "Ming Yan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "629421d60bb8beeba3f087854ebaddc5461f9af0e5fb0bc572d0c20394ba9dcd", "submitted_at": "2025-08-21T00:39:12Z", "updated_at": "2025-09-01T05:38:34Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2508.15144v2"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:49Z", "role": "primary", "sha256": "c208e36ae75cef4b32cfcb2f8182dd4aaaa218e3119bb3d4a2c20a5832f8b1ee", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2506.10387%2C2506.14866%2C2506.20332%2C2507.00008%2C2507.02592%2C2507.03730%2C2507.04227%2C2507.05445%2C2507.05673%2C2507.05720%2C2507.05791%2C2507.14293%2C2507.14799%2C2507.15846%2C2507.16853%2C2507.19132%2C2507.19478%2C2507.21071%2C2507.22025%2C2507.22358%2C2507.23779%2C2508.00935%2C2508.01330%2C2508.03700%2C2508.03923%2C2508.04010%2C2508.04026%2C2508.04482%2C2508.04700%2C2508.05615%2C2508.05731%2C2508.09057%2C2508.09123%2C2508.13167%2C2508.14040%2C2508.15144%2C2508.18689%2C2508.19679%2C2508.20096%2C2509.02544%2C2509.03891%2C2509.06477%2C2509.10723%2C2509.11250%2C2509.11543%2C2509.15221%2C2509.15566%2C2509.15738%2C2509.18119%2C2510.01354&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:49Z", "role": "primary", "sha256": "c208e36ae75cef4b32cfcb2f8182dd4aaaa218e3119bb3d4a2c20a5832f8b1ee", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2506.10387%2C2506.14866%2C2506.20332%2C2507.00008%2C2507.02592%2C2507.03730%2C2507.04227%2C2507.05445%2C2507.05673%2C2507.05720%2C2507.05791%2C2507.14293%2C2507.14799%2C2507.15846%2C2507.16853%2C2507.19132%2C2507.19478%2C2507.21071%2C2507.22025%2C2507.22358%2C2507.23779%2C2508.00935%2C2508.01330%2C2508.03700%2C2508.03923%2C2508.04010%2C2508.04026%2C2508.04482%2C2508.04700%2C2508.05615%2C2508.05731%2C2508.09057%2C2508.09123%2C2508.13167%2C2508.14040%2C2508.15144%2C2508.18689%2C2508.19679%2C2508.20096%2C2509.02544%2C2509.03891%2C2509.06477%2C2509.10723%2C2509.11250%2C2509.11543%2C2509.15221%2C2509.15566%2C2509.15738%2C2509.18119%2C2510.01354&max_results=50"}], "id": "arxiv:2508.15144", "identifiers": {"arxiv": "2508.15144"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:49Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2508.15144"}], "name": "Mobile-Agent-v3: Fundamental Agents for GUI Automation", "notes": "Summary and topic tags reviewed against the primary abstract. Reported experiments, task counts, and artifact availability are not independently reproduced.", "published_at": "2025-08-21", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces GUI-Owl models and Mobile-Agent-v3 orchestration, using environment-generated trajectories to support GUI capabilities.", "topics": ["data-generation", "planning", "mobile-navigation"], "updated_at": "2025-09-01", "url": "https://arxiv.org/abs/2508.15144"}
{"aliases": [], "authors": ["Xianren Zhang", "Shreyas Prasad", "Di Wang", "Qiuhai Zeng", "Suhang Wang", "Wenbo Yan", "Mat Hans"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6c8336a707723227a9d651a3e9347e45907f28ed7421b577761ed57f7d70593d", "submitted_at": "2025-08-18T21:58:43Z", "updated_at": "2026-04-21T00:38:54Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2508.15832v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2508.15832", "identifiers": {"arxiv": "2508.15832"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2508.15832"}], "name": "A Functionality-Grounded Benchmark for Evaluating Web Agents in E-commerce Domains", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-08-18", "relations": [], "review_status": "curated", "scope": "core", "summary": "Amazon-Bench generates functionality-specific web tasks from page content and interactive controls, covering operations beyond product search. Its evaluator checks task completion and unintended changes to account state, such as address or purchase settings.", "topics": ["evaluation", "safety-security", "data-generation", "web-navigation"], "updated_at": "2026-04-21", "url": "https://arxiv.org/abs/2508.15832"}
{"aliases": [], "authors": ["Zhixiong Zeng", "Jing Huang", "Liming Zheng", "Wenkang Han", "Yufeng Zhong", "Lei Chen", "Longrong Yang", "Yingjie Chu", "Yuzhi He", "Lin Ma"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "b2a14819c7d7e8258947c9db8cc89012f4d133eaced4e4b904c3bfd1761181e5", "submitted_at": "2025-08-29T16:40:57Z", "updated_at": "2025-08-29T16:40:57Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2508.21767v1"}}, "environments": ["desktop", "cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2508.21767", "identifiers": {"arxiv": "2508.21767"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2508.21767"}], "name": "UItron: Foundational GUI Agent with Advanced Perception and Planning", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-08-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "UItron combines GUI perception and planning data engineering with supervised training and curriculum reinforcement learning in connected mobile and PC environments. It also collects more than one million operation steps across 100 Chinese apps and builds offline and online evaluation environments.", "topics": ["data-generation", "reinforcement-learning", "planning", "grounding"], "updated_at": "2025-08-29", "url": "https://arxiv.org/abs/2508.21767"}
{"aliases": [], "authors": ["Cheng Zhang", "Erhu Feng", "Xi Zhao", "Yisheng Zhao", "Wangbo Gong", "Jiahui Sun", "Dong Du", "Zhichao Hua", "Yubin Xia", "Haibo Chen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "dbf26f28382245db085fd48f15e905bfa31f1c289a9a7f0799d35440069b442a", "submitted_at": "2025-08-30T15:24:47Z", "updated_at": "2025-08-30T15:24:47Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2509.00531v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T07:57:56Z", "role": "primary", "sha256": "3429cc2fc7737050cac964c976293fc4793e52c5913636c1f67d04243fbefb6a", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2508.20410%2C2508.20890%2C2508.21456%2C2509.00088%2C2509.00531%2C2509.00969%2C2509.01684%2C2509.03536%2C2509.04018%2C2509.04908%2C2509.05368%2C2509.05831%2C2509.06917%2C2509.07098%2C2509.07553%2C2509.08820%2C2509.10345%2C2509.10540%2C2509.11253%2C2509.12543%2C2509.12987%2C2509.13615%2C2509.14547%2C2509.14744%2C2509.15061%2C2509.15237%2C2509.17336%2C2509.19185%2C2509.19199%2C2509.21072%2C2509.21291%2C2509.21766%2C2509.21799%2C2509.22715%2C2509.22830%2C2509.23738%2C2509.24704%2C2509.25455%2C2509.25693%2C2510.00172%2C2510.00515%2C2510.02190%2C2510.04491%2C2510.05096%2C2510.05156%2C2510.05788%2C2510.07284%2C2510.07915%2C2510.08787%2C2510.09517&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:20:43Z", "role": "primary", "sha256": "b61937e1f046c4158b92c2f9d34dddea52a757f0e49b79f30efd73db24484abe", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/abs/2509.00531v1"}], "id": "arxiv:2509.00531", "identifiers": {"arxiv": "2509.00531"}, "kind": "paper", "last_verified_at": "2026-10-17T11:20:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2509.00531"}], "name": "MobiAgent: A Systematic Framework for Customizable Mobile Agents", "notes": "Reviewed against the exact-version official arXiv abstract page; its complete title, ordered author list, visible version, and normalized abstract SHA match the retained snapshot. The original API response body was not locally available. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2025-08-30", "relations": [], "review_status": "curated", "scope": "core", "summary": "Presents MobiAgent, combining MobiMind agent models, the AgentRR acceleration framework, and MobiFlow evaluation. It also introduces an AI-assisted data-collection pipeline for mobile GUI tasks.", "topics": ["mobile-navigation", "efficiency", "data-generation"], "updated_at": "2025-08-30", "url": "https://arxiv.org/abs/2509.00531"}
{"aliases": [], "authors": ["Longrong Yang", "Zhixiong Zeng", "Yufeng Zhong", "Jing Huang", "Liming Zheng", "Lei Chen", "Haibo Qiu", "Zequn Qin", "Lin Ma", "Xi Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "227630b3a257557262ed25eb21906a129eb35c6c5d2de7c5754f919dd281c4bd", "submitted_at": "2025-09-02T13:47:54Z", "updated_at": "2025-09-02T13:47:54Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2509.02322v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:32:52Z", "role": "primary", "sha256": "0b730323b2ce85c39aa7d534665901a480ff06540c73f1f80832fcc0735b8789", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2310.08588%2C2402.09205%2C2406.05720%2C2407.21075%2C2410.07166%2C2503.03743%2C2508.04037%2C2508.10833%2C2508.11360%2C2508.13024%2C2508.13186%2C2508.14104%2C2509.01619%2C2509.01720%2C2509.02322%2C2509.04198%2C2509.07764%2C2509.21823%2C2509.22256%2C2509.23263%2C2509.23866%2C2509.25047%2C2509.26539%2C2510.00413%2C2510.00536%2C2510.01524%2C2510.02204%2C2510.02250%2C2510.02418%2C2510.03204%2C2510.03612%2C2510.04039%2C2510.04373%2C2510.04607%2C2510.06223%2C2510.06607%2C2510.09872%2C2510.10666%2C2510.11035%2C2510.11221%2C2510.11892%2C2510.13543%2C2510.15455%2C2510.15863%2C2510.16252%2C2510.18113%2C2510.18314%2C2510.19205%2C2510.22732%2C2511.06101&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:32:52Z", "role": "primary", "sha256": "0b730323b2ce85c39aa7d534665901a480ff06540c73f1f80832fcc0735b8789", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?id_list=2310.08588%2C2402.09205%2C2406.05720%2C2407.21075%2C2410.07166%2C2503.03743%2C2508.04037%2C2508.10833%2C2508.11360%2C2508.13024%2C2508.13186%2C2508.14104%2C2509.01619%2C2509.01720%2C2509.02322%2C2509.04198%2C2509.07764%2C2509.21823%2C2509.22256%2C2509.23263%2C2509.23866%2C2509.25047%2C2509.26539%2C2510.00413%2C2510.00536%2C2510.01524%2C2510.02204%2C2510.02250%2C2510.02418%2C2510.03204%2C2510.03612%2C2510.04039%2C2510.04373%2C2510.04607%2C2510.06223%2C2510.06607%2C2510.09872%2C2510.10666%2C2510.11035%2C2510.11221%2C2510.11892%2C2510.13543%2C2510.15455%2C2510.15863%2C2510.16252%2C2510.18113%2C2510.18314%2C2510.19205%2C2510.22732%2C2511.06101&max_results=50"}], "id": "arxiv:2509.02322", "identifiers": {"arxiv": "2509.02322"}, "kind": "paper", "last_verified_at": "2026-10-16T06:32:52Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2509.02322"}], "name": "OmniActor: A Generalist GUI and Embodied Agent for 2D&3D Worlds", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2025-09-02", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces OmniActor, which shares shallow model layers while separating deeper parameters to train jointly on GUI and embodied interactions. A unified action representation and mixed training data support evaluation in both task families.", "topics": ["action-prediction", "data-generation", "imitation-learning"], "updated_at": "2025-09-02", "url": "https://arxiv.org/abs/2509.02322"}
{"aliases": [], "authors": ["Haoming Wang", "Haoyang Zou", "Huatong Song", "Jiazhan Feng", "Junjie Fang", "Junting Lu", "Longxiang Liu", "Qinyu Luo", "Shihao Liang", "Shijue Huang", "Wanjun Zhong", "Yining Ye", "Yujia Qin", "Yuwen Xiong", "Yuxin Song", "Zhiyong Wu", "Aoyan Li", "Bo Li", "Chen Dun", "Chong Liu", "Daoguang Zan", "Fuxing Leng", "Hanbin Wang", "Hao Yu", "Haobin Chen", "Hongyi Guo", "Jing Su", "Jingjia Huang", "Kai Shen", "Kaiyu Shi", "Lin Yan", "Peiyao Zhao", "Pengfei Liu", "Qinghao Ye", "Renjie Zheng", "Shulin Xin", "Wayne Xin Zhao", "Wen Heng", "Wenhao Huang", "Wenqian Wang", "Xiaobo Qin", "Yi Lin", "Youbin Wu", "Zehui Chen", "Zihao Wang", "Baoquan Zhong", "Xinchun Zhang", "Xujing Li", "Yuanfan Li", "Zhongkai Zhao", "Chengquan Jiang", "Faming Wu", "Haotian Zhou", "Jinlin Pang", "Li Han", "Qi Liu", "Qianli Ma", "Siyao Liu", "Songhua Cai", "Wenqi Fu", "Xin Liu", "Yaohui Wang", "Zhi Zhang", "Bo Zhou", "Guoliang Li", "Jiajun Shi", "Jiale Yang", "Jie Tang", "Li Li", "Qihua Han", "Taoran Lu", "Woyu Lin", "Xiaokang Tong", "Xinyao Li", "Yichi Zhang", "Yu Miao", "Zhengxuan Jiang", "Zili Li", "Ziyuan Zhao", "Chenxin Li", "Dehua Ma", "Feng Lin", "Ge Zhang", "Haihua Yang", "Hangyu Guo", "Hongda Zhu", "Jiaheng Liu", "Junda Du", "Kai Cai", "Kuanye Li", "Lichen Yuan", "Meilan Han", "Minchao Wang", "Shuyue Guo", "Tianhao Cheng", "Xiaobo Ma", "Xiaojun Xiao", "Xiaolong Huang", "Xinjie Chen", "Yidi Du", "Yilin Chen", "Yiwen Wang", "Zhaojian Li", "Zhenzhu Yang", "Zhiyuan Zeng", "Chaolin Jin", "Chen Li", "Hao Chen", "Haoli Chen", "Jian Chen", "Qinghao Zhao", "Guang Shi"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "800e417f4a2f720beb436ddd2e7bc60f0dd2483b44ac007a72a5eab2c6001ff8", "submitted_at": "2025-09-02T17:44:45Z", "updated_at": "2025-09-05T14:59:27Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2509.02544v2"}}, "environments": ["web", "desktop", "windows", "android", "cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2509.02544", "identifiers": {"arxiv": "2509.02544"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2509.02544"}], "name": "UI-TARS-2 Technical Report: Advancing GUI Agent with Multi-Turn Reinforcement Learning", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-09-02", "relations": [], "review_status": "curated", "scope": "core", "summary": "UI-TARS-2 combines scalable trajectory generation, stabilized multi-turn reinforcement learning, and a sandbox that supports GUI, filesystem, and terminal actions. The report evaluates the resulting agent across web, desktop, Android, and game tasks and analyzes training stability.", "topics": ["reinforcement-learning", "data-generation", "sandbox", "tool-integration"], "updated_at": "2025-09-05", "url": "https://arxiv.org/abs/2509.02544"}
{"aliases": [], "authors": ["Wanfu Wang", "Qipeng Huang", "Guangquan Xue", "Xiaobo Liang", "Juntao Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "875e2452985bae9e67d7071121ab13f5205307ecc30bf33f237048a44e75b3bc", "submitted_at": "2025-09-04T14:17:01Z", "updated_at": "2025-09-04T14:17:01Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2509.04243v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2509.04243", "identifiers": {"arxiv": "2509.04243"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2509.04243"}], "name": "Learning Active Perception via Self-Evolving Preference Optimization for GUI Grounding", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-09-04", "relations": [], "review_status": "curated", "scope": "core", "summary": "LASER builds preference data for multi-step GUI perception by combining Monte Carlo quality estimates with region-overlap evaluation. Its self-evolving training encourages relevant region selection and adapts the amount of reasoning before predicting screen coordinates.", "topics": ["grounding", "data-generation", "screen-understanding"], "updated_at": "2025-09-04", "url": "https://arxiv.org/abs/2509.04243"}
{"aliases": [], "authors": ["Zhaoyang Liu", "Jingjing Xie", "Zichen Ding", "Zehao Li", "Bowen Yang", "Zhenyu Wu", "Xuehui Wang", "Qiushi Sun", "Shi Liu", "Weiyun Wang", "Shenglong Ye", "Qingyun Li", "Xuan Dong", "Yue Yu", "Chenyu Lu", "YunXiang Mo", "Yao Yan", "Zeyue Tian", "Xiao Zhang", "Yuan Huang", "Yiqian Liu", "Weijie Su", "Gen Luo", "Xiangyu Yue", "Biqing Qi", "Kai Chen", "Bowen Zhou", "Yu Qiao", "Qifeng Chen", "Wenhai Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "70d09da7d4990118b2b0745cbddf2bab25852ba3ec26625902fee94cdab37c8d", "submitted_at": "2025-09-18T17:59:22Z", "updated_at": "2025-09-19T05:29:03Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2509.15221v2"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2509.15221", "identifiers": {"arxiv": "2509.15221"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2509.15221"}], "name": "ScaleCUA: Scaling Open-Source Computer Use Agents with Cross-Platform Data", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-09-18", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces a cross-platform computer-use dataset and agents trained with a collection pipeline combining automated interaction and human input.", "topics": ["data-generation", "automation"], "updated_at": "2025-09-19", "url": "https://arxiv.org/abs/2509.15221"}
{"aliases": [], "authors": ["Shaojie Zhang", "Ruoceng Zhang", "Pei Fu", "Shaokang Wang", "Jiahui Yang", "Xin Du", "Shiqi Cui", "Bin Qin", "Ying Huang", "Zhenbo Luo", "Jian Luan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d190eefae9309d87f4ea4780d3483ae0ebd834e0bf70e7f2af204c3788d87e7e", "submitted_at": "2025-09-19T04:03:44Z", "updated_at": "2025-10-27T11:34:58Z", "version": 4, "versioned_url": "https://arxiv.org/abs/2509.15566v4"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2509.15566", "identifiers": {"arxiv": "2509.15566"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2509.15566"}], "name": "BTL-UI: Blink-Think-Link Reasoning Model for GUI Agent", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-09-19", "relations": [], "review_status": "curated", "scope": "core", "summary": "BTL-UI separates GUI interaction into screen-region attention, planning and executable-command generation. It adds an automated annotation pipeline and rule-based process/outcome rewards for training, with evaluation on both static understanding and dynamic interaction.", "topics": ["grounding", "planning", "data-generation", "reinforcement-learning"], "updated_at": "2025-10-27", "url": "https://arxiv.org/abs/2509.15566"}
{"aliases": [], "authors": ["Musen Lin", "Minghao Liu", "Taoran Lu", "Lichen Yuan", "Yiwei Liu", "Haonan Xu", "Yu Miao", "Yuhao Chao", "Zhaojian Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "035048459d2b303721ef0738212829d1fa3314ccd0736b6649cc786c6e975ad4", "submitted_at": "2025-09-19T08:09:18Z", "updated_at": "2025-09-19T08:09:18Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2509.15738v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2509.15738", "identifiers": {"arxiv": "2509.15738"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2509.15738"}], "name": "GUI-ReWalk: Massive Data Generation for GUI Agent via Stochastic Exploration and Intent-Aware Reasoning", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-09-19", "relations": [], "review_status": "curated", "scope": "core", "summary": "GUI-ReWalk begins with stochastic interface exploration and then infers goals to generate coherent action trajectories. It supports long-horizon workflows across applications and studies training on the resulting data using grounding and navigation benchmarks.", "topics": ["data-generation", "cross-app", "long-horizon", "planning"], "updated_at": "2025-09-19", "url": "https://arxiv.org/abs/2509.15738"}
{"aliases": [], "authors": ["Hongxin Li", "Jingran Su", "Jingfan Chen", "Zheng Ju", "Yuntao Chen", "Qing Li", "Zhaoxiang Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "97b7c0690ba7deb7dbbdac6ac94dd82e3864c440cb4325bf143ad735919bb08b", "submitted_at": "2025-09-22T03:04:53Z", "updated_at": "2025-09-22T03:04:53Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2509.17328v1"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2509.17328", "identifiers": {"arxiv": "2509.17328"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2509.17328"}], "name": "UIPro: Unleashing Superior Interaction Capability For GUI Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-09-22", "relations": [], "review_status": "curated", "scope": "core", "summary": "UIPro pretrains GUI understanding on 20.6 million tasks and then harmonizes interaction datasets through a unified action space. Continued fine-tuning uses the merged multi-platform data to train action prediction for generalist GUI operation.", "topics": ["data-generation", "action-prediction", "grounding", "imitation-learning"], "updated_at": "2025-09-22", "url": "https://arxiv.org/abs/2509.17328"}
{"aliases": [], "authors": ["Junyu Lu", "Songxin Zhang", "Zejian Xie", "Zhuoyang Song", "Jiaxing Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "7b85c9bae829bad343dd6f57c866ea100859fa0ad67f5c1a3ac9b3e53d617486", "submitted_at": "2025-09-22T15:40:31Z", "updated_at": "2025-09-22T15:40:31Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2509.17917v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2509.17917", "identifiers": {"arxiv": "2509.17917"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2509.17917"}], "name": "Orcust: Stepwise-Feedback Reinforcement Learning for GUI Agent", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-09-22", "relations": [], "review_status": "curated", "scope": "core", "summary": "Orcust combines principle-constrained reward modeling with GUI trajectory collection in instrumented virtual machines. Environment-checkable and model-derived principles guide stepwise feedback for grounding, basic operations, and complete task execution.", "topics": ["reinforcement-learning", "reward-models", "data-generation", "sandbox"], "updated_at": "2025-09-22", "url": "https://arxiv.org/abs/2509.17917"}
{"aliases": [], "authors": ["Renqi Chen", "Zeyin Tao", "Jianming Guo", "Jingzhe Zhu", "Yiheng Peng", "Qingqing Sun", "Tianyi Zhang", "Shuai Chen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d23b9f33282f77b9d4965078ca993bcf21163dc1d3210aaa4e554d4fc12b37c5", "submitted_at": "2025-09-26T07:05:01Z", "updated_at": "2026-04-13T03:14:30Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2509.21982v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2509.21982", "identifiers": {"arxiv": "2509.21982"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2509.21982"}], "name": "RISK: A Framework for GUI Agents in E-commerce Risk Management", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-09-26", "relations": [], "review_status": "curated", "scope": "core", "summary": "RISK combines browser-interaction training data, a benchmark, and reinforcement fine-tuning for e-commerce risk-management workflows. Its trajectories and rewards distinguish individual actions from multi-step stateful tasks involving deeply embedded web content.", "topics": ["industry", "data-generation", "evaluation", "reinforcement-learning"], "updated_at": "2026-04-13", "url": "https://arxiv.org/abs/2509.21982"}
{"aliases": [], "authors": ["Pengxiang Li", "Zechen Hu", "Zirui Shang", "Jingrong Wu", "Yang Liu", "Hui Liu", "Zhi Gao", "Chenrui Shi", "Bofei Zhang", "Zihao Zhang", "Xiaochuan Shi", "Zedong YU", "Yuwei Wu", "Xinxiao Wu", "Yunde Jia", "Liuyu Xiang", "Zhaofeng He", "Qing Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a96e774920489ef366ad462972a938863a3e714672e9c1f4b784d57eee34c4ca", "submitted_at": "2025-09-28T13:19:20Z", "updated_at": "2025-09-28T13:19:20Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2509.23866v1"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2509.23866", "identifiers": {"arxiv": "2509.23866"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2509.23866"}], "name": "Efficient Multi-turn RL for GUI Agents via Decoupled Training and Adaptive Data Curation", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-09-28", "relations": [], "review_status": "curated", "scope": "core", "summary": "DART separates GUI reinforcement learning into asynchronous environment, rollout, data-management, and training services. It curates experience by task difficulty, successful trajectories, and step uncertainty while correcting rollout-policy mismatch, with evaluation on OSWorld.", "topics": ["reinforcement-learning", "efficiency", "data-generation", "sandbox"], "updated_at": "2025-09-28", "url": "https://arxiv.org/abs/2509.23866"}
{"aliases": [], "authors": ["Ram Ramrakhya", "Andrew Szot", "Omar Attia", "Yuhao Yang", "Anh Nguyen", "Bogdan Mazoure", "Zhe Gan", "Harsh Agrawal", "Alexander Toshev"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d41da9264d664abbeacccae889f9f14f403fc4edbd7453ed4ff276689592fe68", "submitted_at": "2025-09-29T17:00:02Z", "updated_at": "2025-09-29T17:00:02Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2509.25047v1"}}, "environments": ["android", "linux", "desktop", "cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:32:52Z", "role": "primary", "sha256": "0b730323b2ce85c39aa7d534665901a480ff06540c73f1f80832fcc0735b8789", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2310.08588%2C2402.09205%2C2406.05720%2C2407.21075%2C2410.07166%2C2503.03743%2C2508.04037%2C2508.10833%2C2508.11360%2C2508.13024%2C2508.13186%2C2508.14104%2C2509.01619%2C2509.01720%2C2509.02322%2C2509.04198%2C2509.07764%2C2509.21823%2C2509.22256%2C2509.23263%2C2509.23866%2C2509.25047%2C2509.26539%2C2510.00413%2C2510.00536%2C2510.01524%2C2510.02204%2C2510.02250%2C2510.02418%2C2510.03204%2C2510.03612%2C2510.04039%2C2510.04373%2C2510.04607%2C2510.06223%2C2510.06607%2C2510.09872%2C2510.10666%2C2510.11035%2C2510.11221%2C2510.11892%2C2510.13543%2C2510.15455%2C2510.15863%2C2510.16252%2C2510.18113%2C2510.18314%2C2510.19205%2C2510.22732%2C2511.06101&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:32:52Z", "role": "primary", "sha256": "0b730323b2ce85c39aa7d534665901a480ff06540c73f1f80832fcc0735b8789", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2310.08588%2C2402.09205%2C2406.05720%2C2407.21075%2C2410.07166%2C2503.03743%2C2508.04037%2C2508.10833%2C2508.11360%2C2508.13024%2C2508.13186%2C2508.14104%2C2509.01619%2C2509.01720%2C2509.02322%2C2509.04198%2C2509.07764%2C2509.21823%2C2509.22256%2C2509.23263%2C2509.23866%2C2509.25047%2C2509.26539%2C2510.00413%2C2510.00536%2C2510.01524%2C2510.02204%2C2510.02250%2C2510.02418%2C2510.03204%2C2510.03612%2C2510.04039%2C2510.04373%2C2510.04607%2C2510.06223%2C2510.06607%2C2510.09872%2C2510.10666%2C2510.11035%2C2510.11221%2C2510.11892%2C2510.13543%2C2510.15455%2C2510.15863%2C2510.16252%2C2510.18113%2C2510.18314%2C2510.19205%2C2510.22732%2C2511.06101&max_results=50"}], "id": "arxiv:2509.25047", "identifiers": {"arxiv": "2509.25047"}, "kind": "paper", "last_verified_at": "2026-10-16T06:32:52Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2509.25047"}], "name": "Scaling Synthetic Task Generation for Agents via Exploration", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-09-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "AutoPlay explores application states and functions to generate tasks, then uses a model executor and verifier to construct demonstrations. It produces 20,000 tasks across 20 Android apps and 10,000 tasks across 13 Ubuntu apps for supervised and reinforcement learning of UI agents.", "topics": ["data-generation", "reinforcement-learning", "imitation-learning"], "updated_at": "2025-09-29", "url": "https://arxiv.org/abs/2509.25047"}
{"aliases": [], "authors": ["Sofiya Garkot", "Maksym Shamrai", "Ivan Synytsia", "Mariya Hirna"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6e4377fd6b510455afa1eb04c64e8e0b551e89d3b54ce31f241f283233985e1c", "submitted_at": "2025-10-16T19:03:45Z", "updated_at": "2026-03-24T14:39:38Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2510.16051v2"}}, "environments": ["macos", "desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:52Z", "role": "primary", "sha256": "8623b0ff7ef9d2e8ba484cd67359e7b8817e78e9ca22b8f27432c2533f5b90a0", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2510.01670%2C2510.04673%2C2510.10073%2C2510.12200%2C2510.14621%2C2510.16051%2C2510.17790%2C2510.18488%2C2510.18596%2C2510.19488%2C2510.19949%2C2510.20286%2C2510.22009%2C2510.24168%2C2510.24411%2C2510.24563%2C2511.00810%2C2511.04307%2C2511.07332%2C2511.09157%2C2511.12254%2C2511.13087%2C2511.15567%2C2511.16590%2C2511.19663%2C2511.20597%2C2511.20766%2C2512.08529%2C2512.10371%2C2512.10962%2C2512.12634%2C2512.14014%2C2512.15431%2C2512.16295%2C2512.16501%2C2512.19432%2C2512.22894%2C2512.24965%2C2601.02399%2C2601.06031%2C2601.07181%2C2601.07262%2C2601.07779%2C2601.08406%2C2601.09923%2C2601.11044%2C2601.11868%2C2601.12822%2C2601.13060%2C2601.15876&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:52Z", "role": "primary", "sha256": "8623b0ff7ef9d2e8ba484cd67359e7b8817e78e9ca22b8f27432c2533f5b90a0", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2510.01670%2C2510.04673%2C2510.10073%2C2510.12200%2C2510.14621%2C2510.16051%2C2510.17790%2C2510.18488%2C2510.18596%2C2510.19488%2C2510.19949%2C2510.20286%2C2510.22009%2C2510.24168%2C2510.24411%2C2510.24563%2C2511.00810%2C2511.04307%2C2511.07332%2C2511.09157%2C2511.12254%2C2511.13087%2C2511.15567%2C2511.16590%2C2511.19663%2C2511.20597%2C2511.20766%2C2512.08529%2C2512.10371%2C2512.10962%2C2512.12634%2C2512.14014%2C2512.15431%2C2512.16295%2C2512.16501%2C2512.19432%2C2512.22894%2C2512.24965%2C2601.02399%2C2601.06031%2C2601.07181%2C2601.07262%2C2601.07779%2C2601.08406%2C2601.09923%2C2601.11044%2C2601.11868%2C2601.12822%2C2601.13060%2C2601.15876&max_results=50"}], "id": "arxiv:2510.16051", "identifiers": {"arxiv": "2510.16051"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:52Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2510.16051"}], "name": "GUIrilla: A Scalable Framework for Automated Desktop UI Exploration", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-10-16", "relations": [], "review_status": "curated", "scope": "core", "summary": "GUIrilla systematically explores macOS interfaces to collect interaction traces and accessibility metadata for downstream GUI research. It organizes these observations into hierarchical MacApp Trees and describes an accessibility-driven collection library.", "topics": ["data-generation", "screen-understanding", "recording"], "updated_at": "2026-03-24", "url": "https://arxiv.org/abs/2510.16051"}
{"aliases": [], "authors": ["Yuhao Yang", "Zhen Yang", "Zi-Yi Dou", "Anh Nguyen", "Keen You", "Omar Attia", "Andrew Szot", "Michael Feng", "Ram Ramrakhya", "Alexander Toshev", "Chao Huang", "Yinfei Yang", "Zhe Gan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "33569f6ccca9c7e2675796191124424e76636828179c300d35da01ad9ccef1ce", "submitted_at": "2025-10-20T17:48:26Z", "updated_at": "2026-05-26T11:29:37Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2510.17790v3"}}, "environments": ["desktop", "windows"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2510.17790", "identifiers": {"arxiv": "2510.17790"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2510.17790"}], "name": "UltraCUA: A Foundation Model for Computer Use Agents with Hybrid Action", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-10-20", "relations": [], "review_status": "curated", "scope": "core", "summary": "UltraCUA trains a shared policy for primitive GUI actions and high-level tool calls using synthesized tasks and hybrid trajectories. Supervised fine-tuning followed by online reinforcement learning is evaluated on OSWorld and WindowsAgentArena.", "topics": ["tool-integration", "reinforcement-learning", "data-generation"], "updated_at": "2026-05-26", "url": "https://arxiv.org/abs/2510.17790"}
{"aliases": [], "authors": ["Ho Fai Leung", "Xiaoyan Xi", "Fei Zuo"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "33b21a63d51148f2b5c30478c302e3df9b92e03e2e106a1c68224f2c835a8608", "submitted_at": "2025-10-21T10:11:33Z", "updated_at": "2025-10-21T10:11:33Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2510.18488v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2510.18488", "identifiers": {"arxiv": "2510.18488"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2510.18488"}], "name": "AndroidControl-Curated: Revealing the True Potential of GUI Agents through Benchmark Purification", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-10-21", "relations": [], "review_status": "curated", "scope": "core", "summary": "AndroidControl-Curated audits ambiguities and factual errors in AndroidControl and produces a refined evaluation dataset. The paper also studies post-training a small Magma-R1 model on a curated subset to examine the effect of data quality on GUI action assessment.", "topics": ["evaluation", "data-generation", "grounding"], "updated_at": "2025-10-21", "url": "https://arxiv.org/abs/2510.18488"}
{"aliases": [], "authors": ["Dunjie Lu", "Yiheng Xu", "Junli Wang", "Haoyuan Wu", "Xinyuan Wang", "Zekun Wang", "Junlin Yang", "Hongjin Su", "Jixuan Chen", "Junda Chen", "Yuchen Mao", "Jingren Zhou", "Junyang Lin", "Binyuan Hui", "Tao Yu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "ebb064e1a9c3e2255677d9b4cd9ce33d2a97dd26a379c321721d576e7c5d6399", "submitted_at": "2025-10-22T11:25:48Z", "updated_at": "2025-10-22T11:25:48Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2510.19488v1"}}, "environments": ["desktop", "web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2510.19488", "identifiers": {"arxiv": "2510.19488"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2510.19488"}], "name": "VideoAgentTrek: Computer Use Pretraining from Unlabeled Videos", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-10-22", "relations": [], "review_status": "curated", "scope": "core", "summary": "Extracts structured GUI actions from screen-recorded tutorials using inverse dynamics, then uses the inferred trajectories for computer-use training.", "topics": ["data-generation", "video"], "updated_at": "2025-10-22", "url": "https://arxiv.org/abs/2510.19488"}
{"aliases": [], "authors": ["Yujian Liu", "Ze Wang", "Hao Chen", "Ximeng Sun", "Xiaodong Yu", "Jialian Wu", "Jiang Liu", "Emad Barsoum", "Zicheng Liu", "Shiyu Chang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a91413c1e80869860d411402ab49f6e5968787b3bca83f0a9066a0e9c2f75d40", "submitted_at": "2025-11-06T07:29:02Z", "updated_at": "2025-11-06T07:29:02Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2511.04137v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2511.04137", "identifiers": {"arxiv": "2511.04137"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2511.04137"}], "name": "Learning from Online Videos at Inference Time for Computer-Use Agents", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-11-06", "relations": [], "review_status": "curated", "scope": "core", "summary": "This work retrieves online video tutorials and converts them into structured GUI action demonstrations with short objectives. A two-stage selection process supplies a relevant local trajectory as in-context guidance at each execution step.", "topics": ["video", "memory", "planning", "data-generation"], "updated_at": "2025-11-06", "url": "https://arxiv.org/abs/2511.04137"}
{"aliases": [], "authors": ["Jian Mu", "Chaoyun Zhang", "Chiming Ni", "Lu Wang", "Bo Qiao", "Kartik Mathur", "Qianhui Wu", "Yuhang Xie", "Xiaojun Ma", "Mengyu Zhou", "Si Qin", "Liqun Li", "Yu Kang", "Minghua Ma", "Qingwei Lin", "Saravan Rajmohan", "Dongmei Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6b063a4d2b3a552aa05a153745978473a7d165f2c6b4b97e4c91eff11b7b0f80", "submitted_at": "2025-11-06T12:19:02Z", "updated_at": "2025-11-10T12:27:15Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2511.04307v2"}}, "environments": ["desktop", "windows"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2511.04307", "identifiers": {"arxiv": "2511.04307"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2511.04307"}], "name": "GUI-360$^\\circ$: A Comprehensive Dataset and Benchmark for Computer-Using Agents", "notes": "Summary and topic tags reviewed against the primary abstract. Reported experiments, task counts, and artifact availability are not independently reproduced.", "published_at": "2025-11-06", "relations": [], "review_status": "curated", "scope": "core", "summary": "Collects Windows application trajectories with screenshots and action traces for grounding, screen parsing, and action prediction.", "topics": ["data-generation", "grounding", "evaluation"], "updated_at": "2025-11-10", "url": "https://arxiv.org/abs/2511.04307"}
{"aliases": [], "authors": ["Zhaoyang Wang", "Yiming Liang", "Xuchao Zhang", "Qianhui Wu", "Siwei Han", "Anson Bastos", "Rujia Wang", "Chetan Bansal", "Baolin Peng", "Jianfeng Gao", "Saravan Rajmohan", "Huaxiu Yao"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "7420d51203de06ed3afe6f2695b8612ec36dc072765a4f2394f366f9e6bf8838", "submitted_at": "2025-11-08T18:45:33Z", "updated_at": "2026-04-13T06:03:02Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2511.06101v3"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2511.06101", "identifiers": {"arxiv": "2511.06101"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2511.06101"}], "name": "SynthAgent: Adapting Web Agents with Synthetic Supervision", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2025-11-08", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces SynthAgent, which generates tasks by exploring website elements and refines them when observations reveal conflicts. A later trajectory-refinement stage removes misaligned behavior before fine-tuning agents for the target websites.", "topics": ["data-generation", "imitation-learning", "web-navigation"], "updated_at": "2026-04-13", "url": "https://arxiv.org/abs/2511.06101"}
{"aliases": [], "authors": ["Xiangwu Guo", "Difei Gao", "Mike Zheng Shou"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "8a5f43873a47e5b47250cedd3d860abe2dc5ec9b85fe657444bf2b7378f3e8f7", "submitted_at": "2025-11-09T15:13:45Z", "updated_at": "2025-11-09T15:13:45Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2511.06417v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2511.06417", "identifiers": {"arxiv": "2511.06417"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2511.06417"}], "name": "AUTO-Explorer: Automated Data Collection for GUI Agent", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2025-11-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces Auto-Explorer, an automated GUI exploration method that records software states for training-data collection. UIXplore evaluates exploration coverage, and collected data is used to fine-tune and test GUI element grounding.", "topics": ["data-generation", "grounding", "evaluation"], "updated_at": "2025-11-09", "url": "https://arxiv.org/abs/2511.06417"}
{"aliases": [], "authors": ["Aarash Feizi", "Shravan Nayak", "Xiangru Jian", "Kevin Qinghong Lin", "Kaixin Li", "Rabiul Awal", "Xing Han Lù", "Johan Obando-Ceron", "Juan A. Rodriguez", "Nicolas Chapados", "David Vazquez", "Adriana Romero-Soriano", "Reihaneh Rabbany", "Perouz Taslakian", "Christopher Pal", "Spandana Gella", "Sai Rajeswar"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "0f2b80f294ddf5e3331dcb47a836864032f88f2fdbe35588bfef000cd9ceddc8", "submitted_at": "2025-11-10T17:35:21Z", "updated_at": "2026-06-09T21:30:32Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2511.07332v2"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2511.07332", "identifiers": {"arxiv": "2511.07332"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2511.07332"}], "name": "Grounding Computer Use Agents on Human Demonstrations", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-11-10", "relations": [], "review_status": "curated", "scope": "core", "summary": "GroundCUA turns expert desktop demonstrations into densely annotated screenshots and grounding instructions. The GroundNext model family uses this supervision and subsequent reinforcement learning, with both grounding benchmarks and an OSWorld agent configuration used for evaluation.", "topics": ["grounding", "data-generation", "evaluation", "reinforcement-learning"], "updated_at": "2026-06-09", "url": "https://arxiv.org/abs/2511.07332"}
{"aliases": [], "authors": ["Yuan Zhao", "Hualei Zhu", "Tingyu Jiang", "Shen Li", "Xiaohang Xu", "Hao Henry Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "101d38ed439c41ebfa826e990697b502b9126728956295d25b0ea9d539ff23e3", "submitted_at": "2025-11-13T03:41:02Z", "updated_at": "2025-11-13T03:41:02Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2511.10705v1"}}, "environments": ["web", "android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2511.10705", "identifiers": {"arxiv": "2511.10705"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2511.10705"}], "name": "Co-EPG: A Framework for Co-Evolution of Planning and Grounding in Autonomous GUI Agents", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-11-13", "relations": [], "review_status": "curated", "scope": "core", "summary": "Co-EPG alternates reinforcement learning for a planner with updates to a grounding model using generated training examples. The evolving grounder provides feedback to the planner, and evaluation uses Multimodal Mind2Web and AndroidControl.", "topics": ["grounding", "planning", "reinforcement-learning", "data-generation"], "updated_at": "2025-11-13", "url": "https://arxiv.org/abs/2511.10705"}
{"aliases": [], "authors": ["Zengyi Qin", "Jinyuan Chen", "Yunze Man", "Shengcao Cao", "Ziqi Pang", "Zhuoyuan Wang", "Han Fang", "Ling Zhu", "Zixin Xie", "Zibu Wei", "Tianshu Ran", "Haoran Geng", "Ray Pan", "Qizhen Sun", "Zachary Bright", "Yuyang Cai", "Chongye Yang", "Jiace Zhao", "Tianrui Liu", "Han Cao", "Yeyang Zhou", "Rui Wang", "Song Wang", "Xiang Ren", "Bo Zhang", "Yutong Ban", "Pieter Abbeel", "Brian Anthony"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "755d41575ae8ee5d07be0022c8f078c742da4628f12ecd1cad984269441021cf", "submitted_at": "2025-11-11T20:03:38Z", "updated_at": "2026-04-02T02:42:36Z", "version": 5, "versioned_url": "https://arxiv.org/abs/2511.11672v5"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2511.11672", "identifiers": {"arxiv": "2511.11672"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2511.11672"}], "name": "OSGym: Scalable OS Infra for Computer Use Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-11-11", "relations": [], "review_status": "curated", "scope": "core", "summary": "OSGym supplies full operating-system sandboxes with GUIs for parallel computer-use data collection and training. It combines per-replica state management, hardware-aware orchestration, copy-on-write virtual-machine disks, and layered failure recovery.", "topics": ["sandbox", "efficiency", "reinforcement-learning", "data-generation"], "updated_at": "2026-04-02", "url": "https://arxiv.org/abs/2511.11672"}
{"aliases": [], "authors": ["Ahmed Awadallah", "Yash Lara", "Raghav Magazine", "Hussein Mozannar", "Akshay Nambi", "Yash Pandya", "Aravind Rajeswaran", "Corby Rosset", "Alexey Taymanov", "Vibhav Vineet", "Spencer Whitehead", "Andrew Zhao"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "f0e23ba9b31a01956836e67329652882f84fe8ebef1075385a6af5e68cdae936", "submitted_at": "2025-11-24T19:56:28Z", "updated_at": "2025-11-24T19:56:28Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2511.19663v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2511.19663", "identifiers": {"arxiv": "2511.19663"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2511.19663"}], "name": "Fara-7B: An Efficient Agentic Model for Computer Use", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-11-24", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces FaraGen synthetic web trajectories, a compact screenshot-based action model, and WebTailBench for underrepresented browser tasks.", "topics": ["data-generation", "efficiency", "web-navigation"], "updated_at": "2025-11-24", "url": "https://arxiv.org/abs/2511.19663"}
{"aliases": [], "authors": ["Karen Ullrich", "Jingtong Su", "Claudia Shi", "Arjun Subramonian", "Amir Bar", "Ivan Evtimov", "Nikolaos Tsilivis", "Randall Balestriero", "Julia Kempe", "Mark Ibrahim"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2c288e7c42b5eedeea3f27a5622e87981f101f472db05ead6d6bede1371500f0", "submitted_at": "2025-11-25T19:00:22Z", "updated_at": "2025-11-25T19:00:22Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2511.20766v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:52Z", "role": "primary", "sha256": "8623b0ff7ef9d2e8ba484cd67359e7b8817e78e9ca22b8f27432c2533f5b90a0", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2510.01670%2C2510.04673%2C2510.10073%2C2510.12200%2C2510.14621%2C2510.16051%2C2510.17790%2C2510.18488%2C2510.18596%2C2510.19488%2C2510.19949%2C2510.20286%2C2510.22009%2C2510.24168%2C2510.24411%2C2510.24563%2C2511.00810%2C2511.04307%2C2511.07332%2C2511.09157%2C2511.12254%2C2511.13087%2C2511.15567%2C2511.16590%2C2511.19663%2C2511.20597%2C2511.20766%2C2512.08529%2C2512.10371%2C2512.10962%2C2512.12634%2C2512.14014%2C2512.15431%2C2512.16295%2C2512.16501%2C2512.19432%2C2512.22894%2C2512.24965%2C2601.02399%2C2601.06031%2C2601.07181%2C2601.07262%2C2601.07779%2C2601.08406%2C2601.09923%2C2601.11044%2C2601.11868%2C2601.12822%2C2601.13060%2C2601.15876&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:52Z", "role": "primary", "sha256": "8623b0ff7ef9d2e8ba484cd67359e7b8817e78e9ca22b8f27432c2533f5b90a0", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?id_list=2510.01670%2C2510.04673%2C2510.10073%2C2510.12200%2C2510.14621%2C2510.16051%2C2510.17790%2C2510.18488%2C2510.18596%2C2510.19488%2C2510.19949%2C2510.20286%2C2510.22009%2C2510.24168%2C2510.24411%2C2510.24563%2C2511.00810%2C2511.04307%2C2511.07332%2C2511.09157%2C2511.12254%2C2511.13087%2C2511.15567%2C2511.16590%2C2511.19663%2C2511.20597%2C2511.20766%2C2512.08529%2C2512.10371%2C2512.10962%2C2512.12634%2C2512.14014%2C2512.15431%2C2512.16295%2C2512.16501%2C2512.19432%2C2512.22894%2C2512.24965%2C2601.02399%2C2601.06031%2C2601.07181%2C2601.07262%2C2601.07779%2C2601.08406%2C2601.09923%2C2601.11044%2C2601.11868%2C2601.12822%2C2601.13060%2C2601.15876&max_results=50"}], "id": "arxiv:2511.20766", "identifiers": {"arxiv": "2511.20766"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:52Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2511.20766"}], "name": "OpenApps: Simulating Environment Variations to Measure UI-Agent Reliability", "notes": "Reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2025-11-25", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces OpenApps, a configurable app ecosystem for measuring UI-agent reliability under changes in appearance and content. Its evaluation compares task success and failure behavior across many versions of the same apps.", "topics": ["evaluation", "robustness", "data-generation"], "updated_at": "2025-11-25", "url": "https://arxiv.org/abs/2511.20766"}
{"aliases": [], "authors": ["Haolong Yan", "Yeqing Shen", "Xin Huang", "Jia Wang", "Kaijun Tan", "Zhixuan Liang", "Hongxin Li", "Zheng Ge", "Osamu Yoshie", "Si Li", "Xiangyu Zhang", "Daxin Jiang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "fda844ac6767e6d9dfd16919042469e2070ec7f092e0b01fb5a243ed6eb4c282", "submitted_at": "2025-12-02T05:11:23Z", "updated_at": "2025-12-02T05:11:23Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2512.02423v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:32:55Z", "role": "primary", "sha256": "e6061c8dd9c00a70fcf1d4325b671c50a6912063ecbe2b6359175ae4622983a0", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2511.13091%2C2511.21398%2C2511.22235%2C2512.00756%2C2512.00846%2C2512.01174%2C2512.01763%2C2512.02423%2C2512.04105%2C2512.05941%2C2512.07725%2C2512.09396%2C2512.11611%2C2512.12692%2C2512.17221%2C2512.19396%2C2512.21302%2C2512.22009%2C2512.22047%2C2512.22322%2C2512.23128%2C2601.02371%2C2601.02439%2C2601.03641%2C2601.04035%2C2601.04126%2C2601.04500%2C2601.05787%2C2601.06899%2C2601.08605%2C2601.09636%2C2601.09770%2C2601.11631%2C2601.12349%2C2601.16356%2C2601.17418%2C2601.18197%2C2601.18305%2C2601.19199%2C2601.20732%2C2601.21352%2C2601.21872%2C2601.21961%2C2601.22149%2C2601.22491%2C2601.22528%2C2602.02468%2C2602.05354%2C2602.06090%2C2602.06351&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:32:55Z", "role": "primary", "sha256": "e6061c8dd9c00a70fcf1d4325b671c50a6912063ecbe2b6359175ae4622983a0", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?id_list=2511.13091%2C2511.21398%2C2511.22235%2C2512.00756%2C2512.00846%2C2512.01174%2C2512.01763%2C2512.02423%2C2512.04105%2C2512.05941%2C2512.07725%2C2512.09396%2C2512.11611%2C2512.12692%2C2512.17221%2C2512.19396%2C2512.21302%2C2512.22009%2C2512.22047%2C2512.22322%2C2512.23128%2C2601.02371%2C2601.02439%2C2601.03641%2C2601.04035%2C2601.04126%2C2601.04500%2C2601.05787%2C2601.06899%2C2601.08605%2C2601.09636%2C2601.09770%2C2601.11631%2C2601.12349%2C2601.16356%2C2601.17418%2C2601.18197%2C2601.18305%2C2601.19199%2C2601.20732%2C2601.21352%2C2601.21872%2C2601.21961%2C2601.22149%2C2601.22491%2C2601.22528%2C2602.02468%2C2602.05354%2C2602.06090%2C2602.06351&max_results=50"}], "id": "arxiv:2512.02423", "identifiers": {"arxiv": "2512.02423"}, "kind": "paper", "last_verified_at": "2026-10-16T06:32:55Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2512.02423"}], "name": "GUI Exploration Lab: Enhancing Screen Navigation in Agents via Multi-Turn Reinforcement Learning", "notes": "Reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2025-12-02", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces GUI Exploration Lab, a simulation engine for composing screens, interface elements, and navigation graphs. It studies how supervised learning, single-turn RL, and multiturn RL affect GUI navigation and exploration.", "topics": ["reinforcement-learning", "evaluation", "data-generation"], "updated_at": "2025-12-02", "url": "https://arxiv.org/abs/2512.02423"}
{"aliases": [], "authors": ["Yifei He", "Pranit Chawla", "Yaser Souri", "Subhojit Som", "Xia Song"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a7ccc7923f4e6f2ea296b4a6bfc28ba4dc7a52ed256008f153f13fccac3f018f", "submitted_at": "2025-11-22T23:12:56Z", "updated_at": "2026-02-04T19:36:03Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2512.10962v3"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2512.10962", "identifiers": {"arxiv": "2512.10962"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2512.10962"}], "name": "WebSTAR: Scalable Data Synthesis for Computer Use Agents with Step-Level Filtering", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2025-11-22", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces a pipeline that filters individual steps from noisy computer-agent rollouts and augments them with reasoning for training. WebSTAR and WebSCORE support GUI-policy supervision and a distilled process reward model, StepRM.", "topics": ["data-generation", "reward-models", "imitation-learning"], "updated_at": "2026-02-04", "url": "https://arxiv.org/abs/2512.10962"}
{"aliases": [], "authors": ["Chunyi Li", "Longfei Li", "Zicheng Zhang", "Xiaohong Liu", "Min Tang", "Weisi Lin", "Guangtao Zhai"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "484a51656db4466d5e8630d64fa78d3cf10043568488f5a2e9bae679fa800cbe", "submitted_at": "2025-12-12T14:49:32Z", "updated_at": "2025-12-12T14:49:32Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2512.11611v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2512.11611", "identifiers": {"arxiv": "2512.11611"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2512.11611"}], "name": "Using GUI Agent for Electronic Design Automation", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-12-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "GUI-EDA collects more than 2,000 screenshot-answer-action pairs from engineers working across five CAD tools and five physical domains. The accompanying benchmark evaluates more than 30 GUI agents on specialized electronic-design workflows.", "topics": ["industry", "data-generation", "evaluation", "workflow"], "updated_at": "2025-12-12", "url": "https://arxiv.org/abs/2512.11611"}
{"aliases": [], "authors": ["Haolong Yan", "Jia Wang", "Xin Huang", "Yeqing Shen", "Ziyang Meng", "Zhimin Fan", "Kaijun Tan", "Jin Gao", "Lieyu Shi", "Mi Yang", "Shiliang Yang", "Zhirui Wang", "Brian Li", "Kang An", "Chenyang Li", "Lei Lei", "Mengmeng Duan", "Danxun Liang", "Guodong Liu", "Hang Cheng", "Hao Wu", "Jie Dong", "Junhao Huang", "Mei Chen", "Renjie Yu", "Shunshan Li", "Xu Zhou", "Yiting Dai", "Yineng Deng", "Yingdan Liang", "Zelin Chen", "Wen Sun", "Chengxu Yan", "Chunqin Xu", "Dong Li", "Fengqiong Xiao", "Guanghao Fan", "Guopeng Li", "Guozhen Peng", "Hongbing Li", "Hang Li", "Hongming Chen", "Jingjing Xie", "Jianyong Li", "Jingyang Zhang", "Jiaju Ren", "Jiayu Yuan", "Jianpeng Yin", "Kai Cao", "Liang Zhao", "Liguo Tan", "Liying Shi", "Mengqiang Ren", "Min Xu", "Manjiao Liu", "Mao Luo", "Mingxin Wan", "Na Wang", "Nan Wu", "Ning Wang", "Peiyao Ma", "Qingzhou Zhang", "Qiao Wang", "Qinlin Zeng", "Qiong Gao", "Qiongyao Li", "Shangwu Zhong", "Shuli Gao", "Shaofan Liu", "Shisi Gao", "Shuang Luo", "Xingbin Liu", "Xiaojia Liu", "Xiaojie Hou", "Xin Liu", "Xuanti Feng", "Xuedan Cai", "Xuan Wen", "Xianwei Zhu", "Xin Liang", "Xin Liu", "Xin Zhou", "Yifan Sui", "Yingxiu Zhao", "Yukang Shi", "Yunfang Xu", "Yuqing Zeng", "Yixun Zhang", "Zejia Weng", "Zhonghao Yan", "Zhiguo Huang", "Zhuoyu Wang", "Zihan Yan", "Zheng Ge", "Jing Li", "Yibo Zhu", "Binxing Jiao", "Xiangyu Zhang", "Daxin Jiang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2fc824bdeabca3451fd895447a0321c507da1ac7bc7cc24184c7768992549daf", "submitted_at": "2025-12-17T13:26:30Z", "updated_at": "2025-12-19T17:36:21Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2512.15431v2"}}, "environments": ["android", "desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:52Z", "role": "primary", "sha256": "8623b0ff7ef9d2e8ba484cd67359e7b8817e78e9ca22b8f27432c2533f5b90a0", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2510.01670%2C2510.04673%2C2510.10073%2C2510.12200%2C2510.14621%2C2510.16051%2C2510.17790%2C2510.18488%2C2510.18596%2C2510.19488%2C2510.19949%2C2510.20286%2C2510.22009%2C2510.24168%2C2510.24411%2C2510.24563%2C2511.00810%2C2511.04307%2C2511.07332%2C2511.09157%2C2511.12254%2C2511.13087%2C2511.15567%2C2511.16590%2C2511.19663%2C2511.20597%2C2511.20766%2C2512.08529%2C2512.10371%2C2512.10962%2C2512.12634%2C2512.14014%2C2512.15431%2C2512.16295%2C2512.16501%2C2512.19432%2C2512.22894%2C2512.24965%2C2601.02399%2C2601.06031%2C2601.07181%2C2601.07262%2C2601.07779%2C2601.08406%2C2601.09923%2C2601.11044%2C2601.11868%2C2601.12822%2C2601.13060%2C2601.15876&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:52Z", "role": "primary", "sha256": "8623b0ff7ef9d2e8ba484cd67359e7b8817e78e9ca22b8f27432c2533f5b90a0", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2510.01670%2C2510.04673%2C2510.10073%2C2510.12200%2C2510.14621%2C2510.16051%2C2510.17790%2C2510.18488%2C2510.18596%2C2510.19488%2C2510.19949%2C2510.20286%2C2510.22009%2C2510.24168%2C2510.24411%2C2510.24563%2C2511.00810%2C2511.04307%2C2511.07332%2C2511.09157%2C2511.12254%2C2511.13087%2C2511.15567%2C2511.16590%2C2511.19663%2C2511.20597%2C2511.20766%2C2512.08529%2C2512.10371%2C2512.10962%2C2512.12634%2C2512.14014%2C2512.15431%2C2512.16295%2C2512.16501%2C2512.19432%2C2512.22894%2C2512.24965%2C2601.02399%2C2601.06031%2C2601.07181%2C2601.07262%2C2601.07779%2C2601.08406%2C2601.09923%2C2601.11044%2C2601.11868%2C2601.12822%2C2601.13060%2C2601.15876&max_results=50"}], "id": "arxiv:2512.15431", "identifiers": {"arxiv": "2512.15431"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:52Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2512.15431"}], "name": "Step-GUI Technical Report", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-12-17", "relations": [], "review_status": "curated", "scope": "core", "summary": "Step-GUI uses calibrated trajectory rewards to turn generated GUI interaction data into training signals for compact agent models. The report also introduces GUI-MCP for hierarchical GUI operations and local task delegation, and AndroidDaily for static-action and end-to-end mobile evaluation.", "topics": ["data-generation", "reward-models", "tool-integration", "evaluation", "mobile-navigation"], "updated_at": "2025-12-19", "url": "https://arxiv.org/abs/2512.15431"}
{"aliases": [], "authors": ["Zhenyu Wu", "Jingjing Xie", "Zehao Li", "Bowen Yang", "Qiushi Sun", "Zhaoyang Liu", "Zhoumianze Liu", "Yu Qiao", "Xiangyu Yue", "Zun Wang", "Zichen Ding"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "67a1bd40b117b34e5fb994b1c2c93cd285714097d1bc9730baeeaba6c31a7b8a", "submitted_at": "2025-12-18T08:29:50Z", "updated_at": "2025-12-18T08:29:50Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2512.16295v1"}}, "environments": ["web", "desktop", "android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:52Z", "role": "primary", "sha256": "8623b0ff7ef9d2e8ba484cd67359e7b8817e78e9ca22b8f27432c2533f5b90a0", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2510.01670%2C2510.04673%2C2510.10073%2C2510.12200%2C2510.14621%2C2510.16051%2C2510.17790%2C2510.18488%2C2510.18596%2C2510.19488%2C2510.19949%2C2510.20286%2C2510.22009%2C2510.24168%2C2510.24411%2C2510.24563%2C2511.00810%2C2511.04307%2C2511.07332%2C2511.09157%2C2511.12254%2C2511.13087%2C2511.15567%2C2511.16590%2C2511.19663%2C2511.20597%2C2511.20766%2C2512.08529%2C2512.10371%2C2512.10962%2C2512.12634%2C2512.14014%2C2512.15431%2C2512.16295%2C2512.16501%2C2512.19432%2C2512.22894%2C2512.24965%2C2601.02399%2C2601.06031%2C2601.07181%2C2601.07262%2C2601.07779%2C2601.08406%2C2601.09923%2C2601.11044%2C2601.11868%2C2601.12822%2C2601.13060%2C2601.15876&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:52Z", "role": "primary", "sha256": "8623b0ff7ef9d2e8ba484cd67359e7b8817e78e9ca22b8f27432c2533f5b90a0", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2510.01670%2C2510.04673%2C2510.10073%2C2510.12200%2C2510.14621%2C2510.16051%2C2510.17790%2C2510.18488%2C2510.18596%2C2510.19488%2C2510.19949%2C2510.20286%2C2510.22009%2C2510.24168%2C2510.24411%2C2510.24563%2C2511.00810%2C2511.04307%2C2511.07332%2C2511.09157%2C2511.12254%2C2511.13087%2C2511.15567%2C2511.16590%2C2511.19663%2C2511.20597%2C2511.20766%2C2512.08529%2C2512.10371%2C2512.10962%2C2512.12634%2C2512.14014%2C2512.15431%2C2512.16295%2C2512.16501%2C2512.19432%2C2512.22894%2C2512.24965%2C2601.02399%2C2601.06031%2C2601.07181%2C2601.07262%2C2601.07779%2C2601.08406%2C2601.09923%2C2601.11044%2C2601.11868%2C2601.12822%2C2601.13060%2C2601.15876&max_results=50"}], "id": "arxiv:2512.16295", "identifiers": {"arxiv": "2512.16295"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:52Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2512.16295"}], "name": "OS-Oracle: A Comprehensive Framework for Cross-Platform GUI Critic Models", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-12-18", "relations": [], "review_status": "curated", "scope": "core", "summary": "OS-Oracle synthesizes GUI critic examples and trains a pre-action critic with supervised learning followed by consistency-preserving GRPO. It introduces OS-Critic Bench across mobile, web, and desktop settings and tests critic-assisted agents in OSWorld and AndroidWorld.", "topics": ["reward-models", "reinforcement-learning", "data-generation", "evaluation"], "updated_at": "2025-12-18", "url": "https://arxiv.org/abs/2512.16295"}
{"aliases": [], "authors": ["Beitong Zhou", "Zhexiao Huang", "Yuan Guo", "Zhangxuan Gu", "Tianyu Xia", "Zichen Luo", "Fei Tang", "Dehan Kong", "Yanyi Shang", "Suling Ou", "Zhenlin Guo", "Changhua Meng", "Shuheng Shen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "4bc8a1da3993ead9b36af47a9332fe9a1a2427101ad0d91da654b4a4b25141df", "submitted_at": "2025-12-18T13:09:09Z", "updated_at": "2025-12-18T13:09:09Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2512.16501v1"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:52Z", "role": "primary", "sha256": "8623b0ff7ef9d2e8ba484cd67359e7b8817e78e9ca22b8f27432c2533f5b90a0", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2510.01670%2C2510.04673%2C2510.10073%2C2510.12200%2C2510.14621%2C2510.16051%2C2510.17790%2C2510.18488%2C2510.18596%2C2510.19488%2C2510.19949%2C2510.20286%2C2510.22009%2C2510.24168%2C2510.24411%2C2510.24563%2C2511.00810%2C2511.04307%2C2511.07332%2C2511.09157%2C2511.12254%2C2511.13087%2C2511.15567%2C2511.16590%2C2511.19663%2C2511.20597%2C2511.20766%2C2512.08529%2C2512.10371%2C2512.10962%2C2512.12634%2C2512.14014%2C2512.15431%2C2512.16295%2C2512.16501%2C2512.19432%2C2512.22894%2C2512.24965%2C2601.02399%2C2601.06031%2C2601.07181%2C2601.07262%2C2601.07779%2C2601.08406%2C2601.09923%2C2601.11044%2C2601.11868%2C2601.12822%2C2601.13060%2C2601.15876&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:52Z", "role": "primary", "sha256": "8623b0ff7ef9d2e8ba484cd67359e7b8817e78e9ca22b8f27432c2533f5b90a0", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2510.01670%2C2510.04673%2C2510.10073%2C2510.12200%2C2510.14621%2C2510.16051%2C2510.17790%2C2510.18488%2C2510.18596%2C2510.19488%2C2510.19949%2C2510.20286%2C2510.22009%2C2510.24168%2C2510.24411%2C2510.24563%2C2511.00810%2C2511.04307%2C2511.07332%2C2511.09157%2C2511.12254%2C2511.13087%2C2511.15567%2C2511.16590%2C2511.19663%2C2511.20597%2C2511.20766%2C2512.08529%2C2512.10371%2C2512.10962%2C2512.12634%2C2512.14014%2C2512.15431%2C2512.16295%2C2512.16501%2C2512.19432%2C2512.22894%2C2512.24965%2C2601.02399%2C2601.06031%2C2601.07181%2C2601.07262%2C2601.07779%2C2601.08406%2C2601.09923%2C2601.11044%2C2601.11868%2C2601.12822%2C2601.13060%2C2601.15876&max_results=50"}], "id": "arxiv:2512.16501", "identifiers": {"arxiv": "2512.16501"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:52Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2512.16501"}], "name": "VenusBench-GD: A Comprehensive Multi-Platform GUI Benchmark for Diverse Grounding Tasks", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-12-18", "relations": [], "review_status": "curated", "scope": "core", "summary": "VenusBench-GD is a bilingual, cross-platform GUI grounding benchmark organized into basic and advanced tasks. Its annotation pipeline and six-subtask taxonomy support comparisons of general multimodal models and GUI-specialized models across different grounding demands.", "topics": ["grounding", "evaluation", "data-generation", "robustness"], "updated_at": "2025-12-18", "url": "https://arxiv.org/abs/2512.16501"}
{"aliases": [], "authors": ["Runze Li", "Yuwen Zhai", "Bo Xu", "LiWu Xu", "Nian Shi", "Wei Zhang", "Ran Lin", "Liang Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "03cdf8514fa9f036b4614a880d842d49bb9cb0a821858b7215784fba209faa62", "submitted_at": "2025-12-22T13:42:18Z", "updated_at": "2026-04-10T06:20:00Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2512.19396v3"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2512.19396", "identifiers": {"arxiv": "2512.19396"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2512.19396"}], "name": "EchoTrail-GUI: Building Actionable Memory for GUI Agents via Critic-Guided Self-Exploration", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-12-22", "relations": [], "review_status": "curated", "scope": "core", "summary": "EchoTrail-GUI builds a memory of successful GUI trajectories through autonomous exploration filtered by a reward model. It retrieves relevant trajectories as in-context guidance for new tasks and evaluates the approach on AndroidWorld and AndroidLab.", "topics": ["memory", "data-generation", "reward-models", "mobile-navigation"], "updated_at": "2026-04-10", "url": "https://arxiv.org/abs/2512.19396"}
{"aliases": [], "authors": ["Hanzhang Zhou", "Xu Zhang", "Panrong Tong", "Jianan Zhang", "Liangyu Chen", "Quyu Kong", "Chenglin Cai", "Chen Liu", "Yue Wang", "Jingren Zhou", "Steven Hoi"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a89eb3fd173680c127245933d252d4e245b7179d77bca4b463d50800c78f7238", "submitted_at": "2025-12-26T14:51:52Z", "updated_at": "2025-12-26T14:51:52Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2512.22047v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2512.22047", "identifiers": {"arxiv": "2512.22047"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2512.22047"}], "name": "MAI-UI Technical Report: Real-World Centric Foundation GUI Agents", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-12-26", "relations": [], "review_status": "curated", "scope": "core", "summary": "MAI-UI combines GUI navigation, native user interaction, and MCP tool calls in a family of policies trained through generated data and online reinforcement learning. The report also studies device-cloud routing and scaling of parallel environments and interaction budgets, with grounding and Android navigation evaluation.", "topics": ["human-in-the-loop", "tool-integration", "reinforcement-learning", "data-generation", "efficiency"], "updated_at": "2025-12-26", "url": "https://arxiv.org/abs/2512.22047"}
{"aliases": ["ShowUI-pi", "ShowUI Pi"], "authors": ["Siyuan Hu", "Kevin Qinghong Lin", "Mike Zheng Shou"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "b7d1f358b450f2a6c3deb00f1b16c54d443f48fda3a7dc70c77456bebf35ad7c", "submitted_at": "2025-12-31T16:51:14Z", "updated_at": "2025-12-31T16:51:14Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2512.24965v1"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:52Z", "role": "primary", "sha256": "8623b0ff7ef9d2e8ba484cd67359e7b8817e78e9ca22b8f27432c2533f5b90a0", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2510.01670%2C2510.04673%2C2510.10073%2C2510.12200%2C2510.14621%2C2510.16051%2C2510.17790%2C2510.18488%2C2510.18596%2C2510.19488%2C2510.19949%2C2510.20286%2C2510.22009%2C2510.24168%2C2510.24411%2C2510.24563%2C2511.00810%2C2511.04307%2C2511.07332%2C2511.09157%2C2511.12254%2C2511.13087%2C2511.15567%2C2511.16590%2C2511.19663%2C2511.20597%2C2511.20766%2C2512.08529%2C2512.10371%2C2512.10962%2C2512.12634%2C2512.14014%2C2512.15431%2C2512.16295%2C2512.16501%2C2512.19432%2C2512.22894%2C2512.24965%2C2601.02399%2C2601.06031%2C2601.07181%2C2601.07262%2C2601.07779%2C2601.08406%2C2601.09923%2C2601.11044%2C2601.11868%2C2601.12822%2C2601.13060%2C2601.15876&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:52Z", "role": "primary", "sha256": "8623b0ff7ef9d2e8ba484cd67359e7b8817e78e9ca22b8f27432c2533f5b90a0", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2510.01670%2C2510.04673%2C2510.10073%2C2510.12200%2C2510.14621%2C2510.16051%2C2510.17790%2C2510.18488%2C2510.18596%2C2510.19488%2C2510.19949%2C2510.20286%2C2510.22009%2C2510.24168%2C2510.24411%2C2510.24563%2C2511.00810%2C2511.04307%2C2511.07332%2C2511.09157%2C2511.12254%2C2511.13087%2C2511.15567%2C2511.16590%2C2511.19663%2C2511.20597%2C2511.20766%2C2512.08529%2C2512.10371%2C2512.10962%2C2512.12634%2C2512.14014%2C2512.15431%2C2512.16295%2C2512.16501%2C2512.19432%2C2512.22894%2C2512.24965%2C2601.02399%2C2601.06031%2C2601.07181%2C2601.07262%2C2601.07779%2C2601.08406%2C2601.09923%2C2601.11044%2C2601.11868%2C2601.12822%2C2601.13060%2C2601.15876&max_results=50"}], "id": "arxiv:2512.24965", "identifiers": {"arxiv": "2512.24965"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:52Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2512.24965"}], "name": "ShowUI-$π$: Flow-based Generative Models as GUI Dexterous Hands", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-12-31", "relations": [], "review_status": "curated", "scope": "core", "summary": "Models discrete clicks and continuous drags with flow-based action generation, and introduces drag trajectories and the ScreenDrag benchmark.", "topics": ["grounding", "evaluation", "data-generation"], "updated_at": "2025-12-31", "url": "https://arxiv.org/abs/2512.24965"}
{"aliases": [], "authors": ["Hao Bai", "Alexey Taymanov", "Tong Zhang", "Aviral Kumar", "Spencer Whitehead"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "ae8751c4fae3b4f838eba59ffa8be37a1a5f4fa497dfd69fa3075f8581547f51", "submitted_at": "2026-01-05T09:35:11Z", "updated_at": "2026-05-02T18:45:09Z", "version": 6, "versioned_url": "https://arxiv.org/abs/2601.02439v6"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2601.02439", "identifiers": {"arxiv": "2601.02439"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2601.02439"}], "name": "WebGym: Scaling Training Environments for Visual Web Agents with Realistic Tasks", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-01-05", "relations": [], "review_status": "curated", "scope": "core", "summary": "WebGym supplies realistic web tasks with rubric-based evaluation and an asynchronous rollout system for reinforcement learning. It studies how training on an agent’s own interaction traces scales with task coverage and transfers to unseen websites.", "topics": ["web-navigation", "reinforcement-learning", "data-generation", "evaluation", "efficiency"], "updated_at": "2026-05-02", "url": "https://arxiv.org/abs/2601.02439"}
{"aliases": [], "authors": ["Ziyun Zhang", "Zezhou Wang", "Xiaoyi Zhang", "Zongyu Guo", "Jiahao Li", "Bin Li", "Yan Lu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d4d0a59f1eb8197b4d389f5f40957fb8aeb1e22d0739f9877b26d57ed70fe59b", "submitted_at": "2026-01-07T17:40:08Z", "updated_at": "2026-06-30T13:54:32Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2601.04126v3"}}, "environments": ["web", "desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2601.04126", "identifiers": {"arxiv": "2601.04126"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2601.04126"}], "name": "InfiniteWeb: Scalable Web Environment Synthesis for GUI Agent Training", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-01-07", "relations": [], "review_status": "curated", "scope": "core", "summary": "InfiniteWeb generates functional, interconnected websites and task evaluators using shared specifications and task-driven development. It evaluates website construction and the transfer of GUI policies trained in these synthetic web environments to OSWorld and Online-Mind2Web.", "topics": ["data-generation", "reinforcement-learning", "evaluation", "sandbox"], "updated_at": "2026-06-30", "url": "https://arxiv.org/abs/2601.04126"}
{"aliases": [], "authors": ["Zezhou Wang", "Ziyun Zhang", "Xiaoyi Zhang", "Zhuzhong Qian", "Yan Lu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e342723078b50e612df749df80eadfd16e0fe29f5875777c1b3189d8b5981512", "submitted_at": "2026-01-09T13:26:38Z", "updated_at": "2026-02-10T15:12:17Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2601.05787v2"}}, "environments": ["desktop", "web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2601.05787", "identifiers": {"arxiv": "2601.05787"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2601.05787"}], "name": "From Off-Policy to On-Policy: Enhancing GUI Agents via Bi-level Expert-to-Policy Assimilation", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-01-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "BEPA adapts a small set of expert GUI traces into policy-aligned guidance through trajectories reachable by the learner and a dynamically updated task cache. It evaluates this two-level approach to reinforcement learning on desktop and web benchmarks.", "topics": ["reinforcement-learning", "data-generation", "evaluation"], "updated_at": "2026-02-10", "url": "https://arxiv.org/abs/2601.05787"}
{"aliases": [], "authors": ["Yichun Zhang", "Xiangwu Guo", "Yauhong Goh", "Jessica Hu", "Zhiheng Chen", "Xin Wang", "Difei Gao", "Mike Zheng Shou"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "60c6a513f53c16ea6d989a7b2be56ad26b83e531bcea46b00f2e08476b686c97", "submitted_at": "2026-01-12T04:04:20Z", "updated_at": "2026-01-12T04:04:20Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2601.07181v1"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2601.07181", "identifiers": {"arxiv": "2601.07181"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2601.07181"}], "name": "ShowUI-Aloha: Human-Taught GUI Agent", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-01-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "Transforms human screen recordings and input events into annotated demonstrations used by planning and execution components for GUI tasks.", "topics": ["recording", "workflow", "data-generation"], "updated_at": "2026-01-12", "url": "https://arxiv.org/abs/2601.07181"}
{"aliases": [], "authors": ["Wenqi Zhang", "Yulin Shen", "Changyue Jiang", "Jiarun Dai", "Geng Hong", "Xudong Pan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "b57586682d4675387dea2687e7c15d70cc10aecac883e35fe1812c39abcbc1dc", "submitted_at": "2026-01-19T08:32:09Z", "updated_at": "2026-01-19T08:32:09Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2601.12822v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2601.12822", "identifiers": {"arxiv": "2601.12822"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2601.12822"}], "name": "MirrorGuard: Toward Secure Computer-Use Agents via Simulation-to-Real Reasoning Correction", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-01-19", "relations": [], "review_status": "curated", "scope": "core", "summary": "MirrorGuard trains a reasoning-correction defense using GUI risk trajectories generated in a text-based neural-symbolic simulator. It evaluates whether corrections transfer to computer-use agents operating real interfaces, keeping simulated training separate from deployment tests.", "topics": ["safety-security", "data-generation", "robustness"], "updated_at": "2026-01-19", "url": "https://arxiv.org/abs/2601.12822"}
{"aliases": [], "authors": ["Zecheng Li", "Zhihui Cao", "Wenke Huang", "Yudong Zhang", "Keying Qi", "Rui Wang", "Zeyu Zheng", "Jian Zhao", "Hao Zhu", "Hengxin Wu", "Yuran Wang", "Guitao Fan", "Guokun Wu", "Yicong Liu", "Zhilin Gao", "Haikun Xu", "He Yang", "Minqi Xiang", "Xingyu Liu", "Zuojian Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "aa8cb4137e862bf8031a55bc8f81e1d6c903733f2cfac9cf758e5f2f92884481", "submitted_at": "2026-01-19T13:50:43Z", "updated_at": "2026-01-19T13:50:43Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2601.13060v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2601.13060", "identifiers": {"arxiv": "2601.13060"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2601.13060"}], "name": "MagicGUI-RMS: A Multi-Agent Reward Model System for Self-Evolving GUI Agents via Automated Feedback Reflux", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-01-19", "relations": [], "review_status": "curated", "scope": "core", "summary": "MagicGUI-RMS combines domain-specific and general reward models to assess GUI actions and propose corrections. A structured data pipeline and feedback-recycling process support repeated reward-model and agent improvement.", "topics": ["reward-models", "data-generation", "evaluation", "robustness"], "updated_at": "2026-01-19", "url": "https://arxiv.org/abs/2601.13060"}
{"aliases": [], "authors": ["Taofeng Xue", "Chong Peng", "Mianqiu Huang", "Linsen Guo", "Tiancheng Han", "Haozhe Wang", "Jianing Wang", "Xiaocheng Zhang", "Xin Yang", "Dengchang Zhao", "Jinrui Ding", "Xiandi Ma", "Yuchen Xie", "Peng Pei", "Xunliang Cai", "Xipeng Qiu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "38b2798ce9ebd69a46202f0fa8ceca859394acda4f5173b5e30248914c8c35ac", "submitted_at": "2026-01-22T11:36:43Z", "updated_at": "2026-01-23T08:52:00Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2601.15876v2"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2601.15876", "identifiers": {"arxiv": "2601.15876"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2601.15876"}], "name": "EvoCUA: Evolving Computer Use Agents via Learning from Scalable Synthetic Experience", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-01-22", "relations": [], "review_status": "curated", "scope": "core", "summary": "EvoCUA combines task synthesis with executable validators, asynchronous sandbox rollouts and iterative policy updates. It turns successful routines and analyzed failures into supervision and evaluates the evolving agent on OSWorld.", "topics": ["data-generation", "sandbox", "long-horizon", "self-reflection"], "updated_at": "2026-01-23", "url": "https://arxiv.org/abs/2601.15876"}
{"aliases": [], "authors": ["Shaokang Wang", "Pei Fu", "Ruoceng Zhang", "Shaojie Zhang", "Xiuwen Xi", "Jiahui Yang", "Bin Qin", "Ying Huang", "Zhenbo Luo", "Jian Luan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "ccb46b86b0c0b854ed2f1e17f9e12bf031c5ed1349f3d3c44d3276b2a3a195ee", "submitted_at": "2026-01-26T06:29:41Z", "updated_at": "2026-06-26T07:19:55Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2601.18197v2"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:32:55Z", "role": "primary", "sha256": "e6061c8dd9c00a70fcf1d4325b671c50a6912063ecbe2b6359175ae4622983a0", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2511.13091%2C2511.21398%2C2511.22235%2C2512.00756%2C2512.00846%2C2512.01174%2C2512.01763%2C2512.02423%2C2512.04105%2C2512.05941%2C2512.07725%2C2512.09396%2C2512.11611%2C2512.12692%2C2512.17221%2C2512.19396%2C2512.21302%2C2512.22009%2C2512.22047%2C2512.22322%2C2512.23128%2C2601.02371%2C2601.02439%2C2601.03641%2C2601.04035%2C2601.04126%2C2601.04500%2C2601.05787%2C2601.06899%2C2601.08605%2C2601.09636%2C2601.09770%2C2601.11631%2C2601.12349%2C2601.16356%2C2601.17418%2C2601.18197%2C2601.18305%2C2601.19199%2C2601.20732%2C2601.21352%2C2601.21872%2C2601.21961%2C2601.22149%2C2601.22491%2C2601.22528%2C2602.02468%2C2602.05354%2C2602.06090%2C2602.06351&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:32:55Z", "role": "primary", "sha256": "e6061c8dd9c00a70fcf1d4325b671c50a6912063ecbe2b6359175ae4622983a0", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2511.13091%2C2511.21398%2C2511.22235%2C2512.00756%2C2512.00846%2C2512.01174%2C2512.01763%2C2512.02423%2C2512.04105%2C2512.05941%2C2512.07725%2C2512.09396%2C2512.11611%2C2512.12692%2C2512.17221%2C2512.19396%2C2512.21302%2C2512.22009%2C2512.22047%2C2512.22322%2C2512.23128%2C2601.02371%2C2601.02439%2C2601.03641%2C2601.04035%2C2601.04126%2C2601.04500%2C2601.05787%2C2601.06899%2C2601.08605%2C2601.09636%2C2601.09770%2C2601.11631%2C2601.12349%2C2601.16356%2C2601.17418%2C2601.18197%2C2601.18305%2C2601.19199%2C2601.20732%2C2601.21352%2C2601.21872%2C2601.21961%2C2601.22149%2C2601.22491%2C2601.22528%2C2602.02468%2C2602.05354%2C2602.06090%2C2602.06351&max_results=50"}], "id": "arxiv:2601.18197", "identifiers": {"arxiv": "2601.18197"}, "kind": "paper", "last_verified_at": "2026-10-16T06:32:55Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2601.18197"}], "name": "GAIA: A Data Flywheel System for Training GUI Test-Time Scaling Critic Models", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-01-26", "relations": [], "review_status": "curated", "scope": "core", "summary": "GAIA trains a GUI action critic on positive and negative examples, then uses that critic to collect improved examples for another training round. It studies how recycling these data affects the test-time action choices of different GUI agents.", "topics": ["reward-models", "data-generation", "evaluation"], "updated_at": "2026-06-26", "url": "https://arxiv.org/abs/2601.18197"}
{"aliases": [], "authors": ["Xuan Wang", "Siyuan Su", "Quantong Fu", "Yongxiang Hu", "Yangfan Zhou"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "cb90eed33e989d38cd4058d54ba7d073f1b8e5c1453a9106dfe5e533da7b08a1", "submitted_at": "2026-01-26T09:35:10Z", "updated_at": "2026-08-20T07:50:44Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2601.18305v2"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2601.18305", "identifiers": {"arxiv": "2601.18305"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2601.18305"}], "name": "SwipeGen: Bridging the Execution Gap in GUI Agents via Human-like Swipe Synthesis", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-01-26", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces SwipeGen for synthesizing varied swipe trajectories and SwipeBench for assessing their execution quality. The study examines how richer control of trajectory, speed, and timing changes GUI agents’ interaction performance.", "topics": ["action-prediction", "data-generation", "evaluation"], "updated_at": "2026-08-20", "url": "https://arxiv.org/abs/2601.18305"}
{"aliases": [], "authors": ["Le Zhang", "Yixiong Xiao", "Xinjiang Lu", "Jingjia Cao", "Yusai Zhao", "Jingbo Zhou", "Lang An", "Zikan Feng", "Wanxiang Sha", "Yu Shi", "Congxi Xiao", "Jian Xiong", "Yankai Zhang", "Hua Wu", "Haifeng Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "ad764cf86a2ecab81f3f730d6d26ea582544a355a90a8be55be220c98aabd7bf", "submitted_at": "2026-01-28T08:45:17Z", "updated_at": "2026-01-28T08:45:17Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2601.20380v1"}}, "environments": ["android", "desktop", "linux"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2601.20380", "identifiers": {"arxiv": "2601.20380"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2601.20380"}], "name": "OmegaUse: Building a General-Purpose GUI Agent for Autonomous Task Execution", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-01-28", "relations": [], "review_status": "curated", "scope": "core", "summary": "OmegaUse combines exploratory and taxonomy-guided data synthesis with supervised fine-tuning followed by GRPO for GUI control. The report introduces OS-Nav for offline Android and Ubuntu evaluation alongside established grounding and action-prediction benchmarks.", "topics": ["data-generation", "reinforcement-learning", "grounding", "evaluation"], "updated_at": "2026-01-28", "url": "https://arxiv.org/abs/2601.20380"}
{"aliases": [], "authors": ["Hang Ding", "Peidong Liu", "Junqiao Wang", "Ziwei Ji", "Meng Cao", "Rongzhao Zhang", "Lynn Ai", "Eric Yang", "Tianyu Shi", "Lei Yu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "aedb5ed90c91d72f62f5ab60be5374b6f431f887425bc760883ce3fb85a46d43", "submitted_at": "2026-01-29T18:59:07Z", "updated_at": "2026-04-17T18:27:53Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2601.22149v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2601.22149", "identifiers": {"arxiv": "2601.22149"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2601.22149"}], "name": "DynaWeb: Model-Based Reinforcement Learning of Web Agents", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-01-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "DynaWeb learns a world model that predicts web-page representations after actions and uses it to generate policy-training rollouts. It interleaves these rollouts with expert trajectories and evaluates the trained web agents on WebArena and WebVoyager.", "topics": ["world-models", "reinforcement-learning", "data-generation", "web-navigation"], "updated_at": "2026-04-17", "url": "https://arxiv.org/abs/2601.22149"}
{"aliases": [], "authors": ["Linjia Kang", "Zhimin Wang", "Yongkang Zhang", "Duo Wu", "Jinghe Wang", "Ming Ma", "Haopeng Yan", "Zhi Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "0d66365682d0ea4cee744c6c59d3a585c9272a31da5f8ff7c27983d719f070a7", "submitted_at": "2026-01-30T10:03:20Z", "updated_at": "2026-01-30T10:03:20Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2601.22781v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2601.22781", "identifiers": {"arxiv": "2601.22781"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2601.22781"}], "name": "Learning with Challenges: Adaptive Difficulty-Aware Data Generation for Mobile GUI Agent Training", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-01-30", "relations": [], "review_status": "curated", "scope": "core", "summary": "MobileGen estimates a mobile agent’s strengths across structural and semantic task difficulty, then samples new training targets around that profile. A controllable multi-agent generator creates the corresponding instructions and GUI trajectories.", "topics": ["data-generation", "mobile-navigation", "evaluation"], "updated_at": "2026-01-30", "url": "https://arxiv.org/abs/2601.22781"}
{"aliases": [], "authors": ["Woosung Koh", "Sungjun Han", "Segyu Lee", "Se-Young Yun", "Jamin Shin"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "7c29d1404b02c4f5ef2cdaf78bdb0f9062b1b85ec1a154408bc8f49446c28d08", "submitted_at": "2026-02-02T03:12:16Z", "updated_at": "2026-05-23T02:41:44Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2602.01576v2"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:56Z", "role": "primary", "sha256": "c22be2d5c6a84b2c367759ce2b8f41f5a178fccd0534e832802c9d57b40918e8", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2601.16206%2C2601.17722%2C2601.18113%2C2601.18842%2C2601.20162%2C2601.20335%2C2601.20380%2C2601.20650%2C2601.21123%2C2601.22781%2C2602.00575%2C2602.01576%2C2602.01725%2C2602.02419%2C2602.02548%2C2602.02995%2C2602.03255%2C2602.03792%2C2602.05429%2C2602.05832%2C2602.06075%2C2602.06391%2C2602.07153%2C2602.07624%2C2602.08235%2C2602.08995%2C2602.09082%2C2602.09222%2C2602.09310%2C2602.09662%2C2602.10139%2C2602.11514%2C2602.11724%2C2602.11750%2C2602.14093%2C2602.14276%2C2602.14296%2C2602.14721%2C2602.16855%2C2602.17049%2C2602.17365%2C2602.20502%2C2602.21858%2C2602.22190%2C2603.05044%2C2603.05295%2C2603.07978%2C2603.08533%2C2603.10178%2C2603.12138&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:56Z", "role": "primary", "sha256": "c22be2d5c6a84b2c367759ce2b8f41f5a178fccd0534e832802c9d57b40918e8", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2601.16206%2C2601.17722%2C2601.18113%2C2601.18842%2C2601.20162%2C2601.20335%2C2601.20380%2C2601.20650%2C2601.21123%2C2601.22781%2C2602.00575%2C2602.01576%2C2602.01725%2C2602.02419%2C2602.02548%2C2602.02995%2C2602.03255%2C2602.03792%2C2602.05429%2C2602.05832%2C2602.06075%2C2602.06391%2C2602.07153%2C2602.07624%2C2602.08235%2C2602.08995%2C2602.09082%2C2602.09222%2C2602.09310%2C2602.09662%2C2602.10139%2C2602.11514%2C2602.11724%2C2602.11750%2C2602.14093%2C2602.14276%2C2602.14296%2C2602.14721%2C2602.16855%2C2602.17049%2C2602.17365%2C2602.20502%2C2602.21858%2C2602.22190%2C2603.05044%2C2603.05295%2C2603.07978%2C2603.08533%2C2603.10178%2C2603.12138&max_results=50"}], "id": "arxiv:2602.01576", "identifiers": {"arxiv": "2602.01576"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:56Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2602.01576"}], "name": "Generative Visual Code Mobile World Models", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-02-02", "relations": [], "review_status": "curated", "scope": "core", "summary": "gWorld predicts future mobile GUI states as renderable web code and supplies an automated pipeline for code-based training data. The study evaluates visual state prediction and how stronger world models affect downstream mobile policies.", "topics": ["world-models", "data-generation", "screen-understanding", "mobile-navigation"], "updated_at": "2026-05-23", "url": "https://arxiv.org/abs/2602.01576"}
{"aliases": [], "authors": ["Xiaoce Wang", "Guibin Zhang", "Junzhe Li", "Jinzhe Tu", "Chun Li", "Ming Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "49b0dbff31dcf8c966bebd367ae05984e46a7fd13b2cfb1ad5ef061332b43f00", "submitted_at": "2026-01-30T08:38:05Z", "updated_at": "2026-01-30T08:38:05Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2602.02548v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2602.02548", "identifiers": {"arxiv": "2602.02548"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2602.02548"}], "name": "ToolTok: Tool Tokenization for Efficient and Generalizable GUI Agents", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-01-30", "relations": [], "review_status": "curated", "scope": "core", "summary": "ToolTok represents GUI operations with learned tool tokens and frames interaction as progressive tool selection. Semantic anchoring and an easy-to-hard curriculum teach tool meanings before visual pathfinding, with evaluation across grounding scenarios.", "topics": ["tool-integration", "grounding", "data-generation"], "updated_at": "2026-01-30", "url": "https://arxiv.org/abs/2602.02548"}
{"aliases": [], "authors": ["Rui Lv", "Juncheng Mo", "Tianyi Chu", "Chen Rao", "Hongyi Jing", "Jiajie Teng", "Jiafu Chen", "Shiqi Zhang", "Liangzi Ding", "Shuo Fang", "Huaizhong Lin", "Ziqiang Dang", "Chenguang Ma", "Lei Zhao"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "84d822655c810958cff6035584542d7e151d765734ec7030f841c15561ee5677", "submitted_at": "2026-02-05T08:19:39Z", "updated_at": "2026-02-05T08:19:39Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2602.05429v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2602.05429", "identifiers": {"arxiv": "2602.05429"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2602.05429"}], "name": "M$^2$-Miner: Multi-Agent Enhanced MCTS for Mobile GUI Agent Data Mining", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-02-05", "relations": [], "review_status": "curated", "scope": "core", "summary": "M2-Miner uses multi-agent-guided Monte Carlo tree search to collect mobile GUI intent–trajectory pairs. Intent recycling and progressive collector training aim to increase the diversity and usefulness of the mined interaction data.", "topics": ["data-generation", "mobile-navigation", "planning"], "updated_at": "2026-02-05", "url": "https://arxiv.org/abs/2602.05429"}
{"aliases": [], "authors": ["Jinbiao Wei", "Yilun Zhao", "Kangqi Ni", "Arman Cohan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "ce896d4d846bf07fedc68b7f952532c4fefe24dcb49b51556e0d6c2bc648bb91", "submitted_at": "2026-02-06T19:55:26Z", "updated_at": "2026-04-12T06:09:23Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2602.07153v2"}}, "environments": ["desktop", "windows"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2602.07153", "identifiers": {"arxiv": "2602.07153"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2602.07153"}], "name": "ANCHOR: Branch-Point Data Generation for GUI Agents", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-02-06", "relations": [], "review_status": "curated", "scope": "core", "summary": "ANCHOR expands verified desktop demonstrations by generating new tasks at meaningful branch points and executing their instructions. State-aware verification and action filtering refine the resulting trajectories, which are evaluated through fine-tuned agents on OSWorld and WindowsAgentArena.", "topics": ["data-generation", "planning", "evaluation"], "updated_at": "2026-04-12", "url": "https://arxiv.org/abs/2602.07153"}
{"aliases": [], "authors": ["Jiacheng Liu", "Yaxin Luo", "Jiacheng Cui", "Xinyi Shang", "Xiaohan Zhao", "Zhiqiang Shen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "3178684a322fcf07f847473139efdddf0b02213af468f1afbd285549d15bde56", "submitted_at": "2026-02-09T18:55:33Z", "updated_at": "2026-06-21T08:25:39Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2602.09012v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2602.09012", "identifiers": {"arxiv": "2602.09012"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2602.09012"}], "name": "Next-Gen CAPTCHAs: Leveraging the Cognitive Gap for Scalable and Diverse GUI-Agent Defense", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-02-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "Next-Gen CAPTCHAs introduces generated interactive challenges intended to distinguish human users from GUI agents. The benchmark targets perception, memory, decisions and actions, with backend-supported challenge types designed for scalable instance generation.", "topics": ["evaluation", "safety-security", "data-generation"], "updated_at": "2026-06-21", "url": "https://arxiv.org/abs/2602.09012"}
{"aliases": [], "authors": ["Ananya Gubbi Mohanbabu", "Rosiana Natalie", "Brandon Kim", "Anhong Guo", "Amy Pavel"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "c3606d090749737d4223a9f826400c2b2aebde245c623391f2a8c6702b2468f8", "submitted_at": "2026-02-10T01:00:55Z", "updated_at": "2026-02-10T01:00:55Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2602.09310v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2602.09310", "identifiers": {"arxiv": "2602.09310"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2602.09310"}], "name": "A11y-CUA Dataset: Characterizing the Accessibility Gap in Computer Use Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-02-10", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces A11y-CUA, a dataset comparing how blind or low-vision users and sighted users perform everyday computer tasks. It contrasts those traces with agent behavior under default, keyboard-only, and magnifier conditions to identify barriers to shared control.", "topics": ["human-in-the-loop", "evaluation", "data-generation"], "updated_at": "2026-02-10", "url": "https://arxiv.org/abs/2602.09310"}
{"aliases": [], "authors": ["Deyang Jiang", "Jing Huang", "Xuanle Zhao", "Lei Chen", "Liming Zheng", "Fanfan Liu", "Haibo Qiu", "Peng Shi", "Zhixiong Zeng"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "b5080050b85e38c8c03078c96c206e92d4ed63d106d55983563d51fcbc9f66c5", "submitted_at": "2026-02-10T11:16:57Z", "updated_at": "2026-02-10T11:16:57Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2602.09662v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:56Z", "role": "primary", "sha256": "c22be2d5c6a84b2c367759ce2b8f41f5a178fccd0534e832802c9d57b40918e8", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2601.16206%2C2601.17722%2C2601.18113%2C2601.18842%2C2601.20162%2C2601.20335%2C2601.20380%2C2601.20650%2C2601.21123%2C2601.22781%2C2602.00575%2C2602.01576%2C2602.01725%2C2602.02419%2C2602.02548%2C2602.02995%2C2602.03255%2C2602.03792%2C2602.05429%2C2602.05832%2C2602.06075%2C2602.06391%2C2602.07153%2C2602.07624%2C2602.08235%2C2602.08995%2C2602.09082%2C2602.09222%2C2602.09310%2C2602.09662%2C2602.10139%2C2602.11514%2C2602.11724%2C2602.11750%2C2602.14093%2C2602.14276%2C2602.14296%2C2602.14721%2C2602.16855%2C2602.17049%2C2602.17365%2C2602.20502%2C2602.21858%2C2602.22190%2C2603.05044%2C2603.05295%2C2603.07978%2C2603.08533%2C2603.10178%2C2603.12138&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:56Z", "role": "primary", "sha256": "c22be2d5c6a84b2c367759ce2b8f41f5a178fccd0534e832802c9d57b40918e8", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2601.16206%2C2601.17722%2C2601.18113%2C2601.18842%2C2601.20162%2C2601.20335%2C2601.20380%2C2601.20650%2C2601.21123%2C2601.22781%2C2602.00575%2C2602.01576%2C2602.01725%2C2602.02419%2C2602.02548%2C2602.02995%2C2602.03255%2C2602.03792%2C2602.05429%2C2602.05832%2C2602.06075%2C2602.06391%2C2602.07153%2C2602.07624%2C2602.08235%2C2602.08995%2C2602.09082%2C2602.09222%2C2602.09310%2C2602.09662%2C2602.10139%2C2602.11514%2C2602.11724%2C2602.11750%2C2602.14093%2C2602.14276%2C2602.14296%2C2602.14721%2C2602.16855%2C2602.17049%2C2602.17365%2C2602.20502%2C2602.21858%2C2602.22190%2C2603.05044%2C2603.05295%2C2603.07978%2C2603.08533%2C2603.10178%2C2603.12138&max_results=50"}], "id": "arxiv:2602.09662", "identifiers": {"arxiv": "2602.09662"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:56Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2602.09662"}], "name": "TreeCUA: Efficiently Scaling GUI Automation with Tree-Structured Verifiable Evolution", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-02-10", "relations": [], "review_status": "curated", "scope": "core", "summary": "TreeCUA organizes GUI exploration trajectories as reusable trees and balances deeper tasks against broader coverage. Its data pipeline combines action verification, memory-backed recovery and a DPO variant that uses information from adjacent branches.", "topics": ["data-generation", "planning", "memory", "efficiency"], "updated_at": "2026-02-10", "url": "https://arxiv.org/abs/2602.09662"}
{"aliases": [], "authors": ["Yuhao Zheng", "Li'an Zhong", "Yi Wang", "Rui Dai", "Kaikui Liu", "Xiangxiang Chu", "Linyuan Lv", "Philip Torr", "Kevin Qinghong Lin"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "55287e2fe8a5e94671e4aab8e9d4a907ba9b6c83306d9b3f61b74adccd5c5059", "submitted_at": "2026-02-10T14:56:19Z", "updated_at": "2026-02-10T14:56:19Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2602.09856v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:32:59Z", "role": "primary", "sha256": "d494c76e512656f76a5912eac0768bbbdd50014387c779a0618a40c1722c399b", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2602.07008%2C2602.07283%2C2602.07787%2C2602.09856%2C2602.10356%2C2602.10814%2C2602.10915%2C2602.11524%2C2602.11858%2C2602.12276%2C2602.12544%2C2602.12852%2C2602.13559%2C2602.13653%2C2602.14865%2C2602.15384%2C2602.17003%2C2602.17245%2C2602.17588%2C2602.24142%2C2603.00503%2C2603.00676%2C2603.02951%2C2603.04364%2C2603.04949%2C2603.07024%2C2603.07432%2C2603.08013%2C2603.08316%2C2603.08938%2C2603.10268%2C2603.10291%2C2603.10505%2C2603.10577%2C2603.12710%2C2603.14248%2C2603.14448%2C2603.17441%2C2603.17826%2C2603.20278%2C2603.20340%2C2603.22529%2C2603.23007%2C2603.23559%2C2603.24160%2C2603.24533%2C2603.26041%2C2603.26266%2C2604.00073%2C2604.00842&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:32:59Z", "role": "primary", "sha256": "d494c76e512656f76a5912eac0768bbbdd50014387c779a0618a40c1722c399b", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2602.07008%2C2602.07283%2C2602.07787%2C2602.09856%2C2602.10356%2C2602.10814%2C2602.10915%2C2602.11524%2C2602.11858%2C2602.12276%2C2602.12544%2C2602.12852%2C2602.13559%2C2602.13653%2C2602.14865%2C2602.15384%2C2602.17003%2C2602.17245%2C2602.17588%2C2602.24142%2C2603.00503%2C2603.00676%2C2603.02951%2C2603.04364%2C2603.04949%2C2603.07024%2C2603.07432%2C2603.08013%2C2603.08316%2C2603.08938%2C2603.10268%2C2603.10291%2C2603.10505%2C2603.10577%2C2603.12710%2C2603.14248%2C2603.14448%2C2603.17441%2C2603.17826%2C2603.20278%2C2603.20340%2C2603.22529%2C2603.23007%2C2603.23559%2C2603.24160%2C2603.24533%2C2603.26041%2C2603.26266%2C2604.00073%2C2604.00842&max_results=50"}], "id": "arxiv:2602.09856", "identifiers": {"arxiv": "2602.09856"}, "kind": "paper", "last_verified_at": "2026-10-16T06:32:59Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2602.09856"}], "name": "Code2World: A GUI World Model via Renderable Code Generation", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-02-10", "relations": [], "review_status": "curated", "scope": "core", "summary": "Code2World predicts future GUI screens through renderable code, trained on AndroidCode trajectories with supervised learning and render-aware reinforcement learning. It evaluates visual prediction and downstream AndroidWorld navigation with the learned world model.", "topics": ["world-models", "data-generation", "reinforcement-learning", "mobile-navigation"], "updated_at": "2026-02-10", "url": "https://arxiv.org/abs/2602.09856"}
{"aliases": [], "authors": ["Tianci Xue", "Zeyi Liao", "Tianneng Shi", "Zilu Wang", "Kai Zhang", "Dawn Song", "Yu Su", "Huan Sun"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d3385966e7b483b33738b3c8e6c9a4d9cf17afa671e1f2e49de4f946cbc5f251", "submitted_at": "2026-02-10T23:06:02Z", "updated_at": "2026-05-11T00:11:46Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2602.10356v2"}}, "environments": ["desktop", "linux", "windows"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:29:29+00:00", "role": "primary", "sha256": "fba520c76ca9268af9239fbeaf5836e21932ab41ce53ccba71f47e3fc767028a", "supports": ["summary", "environments", "scope"], "url": "https://arxiv.org/html/2602.10356v2"}], "id": "arxiv:2602.10356", "identifiers": {"arxiv": "2602.10356"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2602.10356"}], "name": "Autonomous Continual Learning for Environment Adaptation of Computer-Use Agents", "notes": "The annotation was reviewed against the complete retained official abstract and the specified sections of the same recorded paper version. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-02-10", "relations": [], "review_status": "curated", "scope": "core", "summary": "ACuRL uses autonomous interface exploration and an adaptive task curriculum to continually train computer-use policies with reinforcement learning. CUAJudge supplies outcome feedback, and experiments study adaptation across desktop applications and environment changes.", "topics": ["reinforcement-learning", "data-generation", "reward-models", "robustness"], "updated_at": "2026-05-11", "url": "https://arxiv.org/abs/2602.10356"}
{"aliases": [], "authors": ["Lajanugen Logeswaran", "Jaekyeom Kim", "Sungryull Sohn", "Creighton Glasscock", "Honglak Lee"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "555285a1ab51e02661067bdf8bcc3fa7f6ee0fdd63f52f786a03eb2483bd1a31", "submitted_at": "2026-02-13T02:52:18Z", "updated_at": "2026-02-13T02:52:18Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2602.12544v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2602.12544", "identifiers": {"arxiv": "2602.12544"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2602.12544"}], "name": "Scaling Web Agent Training through Automatic Data Generation and Fine-grained Evaluation", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-02-13", "relations": [], "review_status": "curated", "scope": "core", "summary": "This work uses constraint-based progress evaluation to retain partially successful web trajectories for training. It introduces BookingArena for structured booking tasks and studies a distilled agent trained with the resulting data.", "topics": ["data-generation", "evaluation", "web-navigation"], "updated_at": "2026-02-13", "url": "https://arxiv.org/abs/2602.12544"}
{"aliases": [], "authors": ["Yuan Cao", "Dezhi Ran", "Mengzhou Wu", "Yuzhe Guo", "Xin Chen", "Ang Li", "Gang Cao", "Gong Zhi", "Hao Yu", "Linyi Li", "Wei Yang", "Tao Xie"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "29b4c8a5bda1df237fad33ce483018b3476684f904841ed0598ec4286234b858", "submitted_at": "2026-02-15T10:58:01Z", "updated_at": "2026-02-15T10:58:01Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2602.14093v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2602.14093", "identifiers": {"arxiv": "2602.14093"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2602.14093"}], "name": "GUI-GENESIS: Automated Synthesis of Efficient Environments with Verifiable Rewards for GUI Agent Post-Training", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-02-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "GUI-GENESIS reconstructs applications as lightweight web environments with executable assertions for reward computation. It evaluates the cost of these training environments and transfer from synthetic interaction training to held-out real tasks.", "topics": ["data-generation", "sandbox", "reinforcement-learning", "evaluation"], "updated_at": "2026-02-15", "url": "https://arxiv.org/abs/2602.14093"}
{"aliases": [], "authors": ["A. Said Gurbuz", "Sunghwan Hong", "Ahmed Nassar", "Marc Pollefeys", "Peter Staar"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d7c6e23f704562c4ae4db9e0b4f8f2a8dde1318efc08876ee57f5967b178257e", "submitted_at": "2026-02-15T19:00:02Z", "updated_at": "2026-05-01T17:32:06Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2602.14276v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:56Z", "role": "primary", "sha256": "c22be2d5c6a84b2c367759ce2b8f41f5a178fccd0534e832802c9d57b40918e8", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2601.16206%2C2601.17722%2C2601.18113%2C2601.18842%2C2601.20162%2C2601.20335%2C2601.20380%2C2601.20650%2C2601.21123%2C2601.22781%2C2602.00575%2C2602.01576%2C2602.01725%2C2602.02419%2C2602.02548%2C2602.02995%2C2602.03255%2C2602.03792%2C2602.05429%2C2602.05832%2C2602.06075%2C2602.06391%2C2602.07153%2C2602.07624%2C2602.08235%2C2602.08995%2C2602.09082%2C2602.09222%2C2602.09310%2C2602.09662%2C2602.10139%2C2602.11514%2C2602.11724%2C2602.11750%2C2602.14093%2C2602.14276%2C2602.14296%2C2602.14721%2C2602.16855%2C2602.17049%2C2602.17365%2C2602.20502%2C2602.21858%2C2602.22190%2C2603.05044%2C2603.05295%2C2603.07978%2C2603.08533%2C2603.10178%2C2603.12138&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:56Z", "role": "primary", "sha256": "c22be2d5c6a84b2c367759ce2b8f41f5a178fccd0534e832802c9d57b40918e8", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2601.16206%2C2601.17722%2C2601.18113%2C2601.18842%2C2601.20162%2C2601.20335%2C2601.20380%2C2601.20650%2C2601.21123%2C2601.22781%2C2602.00575%2C2602.01576%2C2602.01725%2C2602.02419%2C2602.02548%2C2602.02995%2C2602.03255%2C2602.03792%2C2602.05429%2C2602.05832%2C2602.06075%2C2602.06391%2C2602.07153%2C2602.07624%2C2602.08235%2C2602.08995%2C2602.09082%2C2602.09222%2C2602.09310%2C2602.09662%2C2602.10139%2C2602.11514%2C2602.11724%2C2602.11750%2C2602.14093%2C2602.14276%2C2602.14296%2C2602.14721%2C2602.16855%2C2602.17049%2C2602.17365%2C2602.20502%2C2602.21858%2C2602.22190%2C2603.05044%2C2603.05295%2C2603.07978%2C2603.08533%2C2603.10178%2C2603.12138&max_results=50"}], "id": "arxiv:2602.14276", "identifiers": {"arxiv": "2602.14276"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:56Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2602.14276"}], "name": "ScreenParse: Moving Beyond Sparse Grounding with Complete Screen Parsing Supervision", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-02-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "ScreenParse supplies dense annotations of UI elements, types and text in web screenshots through an automated rendering and filtering pipeline. ScreenVLM learns to output structured screen markup, and the study evaluates dense parsing and transfer to GUI grounding.", "topics": ["screen-understanding", "data-generation", "grounding", "evaluation"], "updated_at": "2026-05-01", "url": "https://arxiv.org/abs/2602.14276"}
{"aliases": [], "authors": ["Yifan Wu", "Yiran Peng", "Yiyu Chen", "Jianhao Ruan", "Zijie Zhuang", "Cheng Yang", "Jiayi Zhang", "Man Chen", "Yenchi Tseng", "Zhaoyang Yu", "Liang Chen", "Yuyao Zhai", "Bang Liu", "Chenglin Wu", "Yuyu Luo"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "3773ec51f5316b9006f93aeafd7b654bf0e7418253c0af16ef7b7c00fddea102", "submitted_at": "2026-02-15T20:03:19Z", "updated_at": "2026-02-15T20:03:19Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2602.14296v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:56Z", "role": "primary", "sha256": "c22be2d5c6a84b2c367759ce2b8f41f5a178fccd0534e832802c9d57b40918e8", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2601.16206%2C2601.17722%2C2601.18113%2C2601.18842%2C2601.20162%2C2601.20335%2C2601.20380%2C2601.20650%2C2601.21123%2C2601.22781%2C2602.00575%2C2602.01576%2C2602.01725%2C2602.02419%2C2602.02548%2C2602.02995%2C2602.03255%2C2602.03792%2C2602.05429%2C2602.05832%2C2602.06075%2C2602.06391%2C2602.07153%2C2602.07624%2C2602.08235%2C2602.08995%2C2602.09082%2C2602.09222%2C2602.09310%2C2602.09662%2C2602.10139%2C2602.11514%2C2602.11724%2C2602.11750%2C2602.14093%2C2602.14276%2C2602.14296%2C2602.14721%2C2602.16855%2C2602.17049%2C2602.17365%2C2602.20502%2C2602.21858%2C2602.22190%2C2603.05044%2C2603.05295%2C2603.07978%2C2603.08533%2C2603.10178%2C2603.12138&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:56Z", "role": "primary", "sha256": "c22be2d5c6a84b2c367759ce2b8f41f5a178fccd0534e832802c9d57b40918e8", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2601.16206%2C2601.17722%2C2601.18113%2C2601.18842%2C2601.20162%2C2601.20335%2C2601.20380%2C2601.20650%2C2601.21123%2C2601.22781%2C2602.00575%2C2602.01576%2C2602.01725%2C2602.02419%2C2602.02548%2C2602.02995%2C2602.03255%2C2602.03792%2C2602.05429%2C2602.05832%2C2602.06075%2C2602.06391%2C2602.07153%2C2602.07624%2C2602.08235%2C2602.08995%2C2602.09082%2C2602.09222%2C2602.09310%2C2602.09662%2C2602.10139%2C2602.11514%2C2602.11724%2C2602.11750%2C2602.14093%2C2602.14276%2C2602.14296%2C2602.14721%2C2602.16855%2C2602.17049%2C2602.17365%2C2602.20502%2C2602.21858%2C2602.22190%2C2603.05044%2C2603.05295%2C2603.07978%2C2603.08533%2C2603.10178%2C2603.12138&max_results=50"}], "id": "arxiv:2602.14296", "identifiers": {"arxiv": "2602.14296"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:56Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2602.14296"}], "name": "AutoWebWorld: Synthesizing Infinite Verifiable Web Environments via Finite State Machines", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-02-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "AutoWebWorld turns finite-state task models into interactive websites whose actions and goal states can be checked programmatically. It generates verified trajectories for training and evaluates transfer to WebVoyager and Online-Mind2Web.", "topics": ["data-generation", "sandbox", "web-navigation", "evaluation"], "updated_at": "2026-02-15", "url": "https://arxiv.org/abs/2602.14296"}
{"aliases": [], "authors": ["Zikai Xiao", "Jianhong Tu", "Chuhang Zou", "Yuxin Zuo", "Zhi Li", "Peng Wang", "Bowen Yu", "Fei Huang", "Junyang Lin", "Zuozhu Liu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2dca0e4219a788205bbb060911d0f004d1db4cf2ff8a4962c04fd5777ea194d7", "submitted_at": "2026-02-16T13:06:49Z", "updated_at": "2026-02-16T13:06:49Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2602.14721v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2602.14721", "identifiers": {"arxiv": "2602.14721"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2602.14721"}], "name": "WebWorld: A Large-Scale World Model for Web Agent Training", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2026-02-16", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces WebWorld, an open-web simulator trained on interaction trajectories for generating agent training data and testing candidate actions. WebWorld-Bench evaluates simulated transitions, alongside downstream web-agent training experiments.", "topics": ["world-models", "data-generation", "evaluation"], "updated_at": "2026-02-16", "url": "https://arxiv.org/abs/2602.14721"}
{"aliases": [], "authors": ["Faria Huq", "Zora Zhiruo Wang", "Zhanqiu Guo", "Venu Arvind Arangarajan", "Tianyue Ou", "Frank Xu", "Shuyan Zhou", "Graham Neubig", "Jeffrey P. Bigham"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "fe83d38ece7cd8f39d932611abaf7a170e7cb9b7c9744bd1176eb42e73be85b5", "submitted_at": "2026-02-19T18:11:28Z", "updated_at": "2026-07-07T23:37:46Z", "version": 4, "versioned_url": "https://arxiv.org/abs/2602.17588v4"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2602.17588", "identifiers": {"arxiv": "2602.17588"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2602.17588"}], "name": "Modeling Distinct Human Interaction in Web Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-02-19", "relations": [], "review_status": "curated", "scope": "core", "summary": "CowCorpus records interleaved human and agent actions during web navigation and identifies distinct patterns of user intervention. Models trained to anticipate these interventions are integrated into live web agents and assessed in a user study.", "topics": ["human-in-the-loop", "web-navigation", "data-generation", "action-prediction"], "updated_at": "2026-07-07", "url": "https://arxiv.org/abs/2602.17588"}
{"aliases": [], "authors": ["Md Farhan Ishmam", "Kenneth Marino"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "1901125b21431e41eacb5d5f0fc7a439d9a637c43762d441ece46a90e135051b", "submitted_at": "2026-03-05T08:43:06Z", "updated_at": "2026-10-14T10:16:21Z", "version": 4, "versioned_url": "https://arxiv.org/abs/2603.04949v4"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2603.04949", "identifiers": {"arxiv": "2603.04949"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2603.04949"}], "name": "TimeWarp: Evaluating Web Agents by Revisiting the Past", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-03-05", "relations": [], "review_status": "curated", "scope": "core", "summary": "TimeWarp tests web agents across multiple versions of the same sites, varying visual design, frontend code and workflows. TimeTraj transfers plans across versions to collect training trajectories and study robustness to interface change.", "topics": ["evaluation", "robustness", "web-navigation", "data-generation"], "updated_at": "2026-10-14", "url": "https://arxiv.org/abs/2603.04949"}
{"aliases": [], "authors": ["Sicheng Fan", "Qingyun Shi", "Shengze Xu", "Shengbo Cai", "Tieyong Zeng", "Li Ling", "Yanyi Shang", "Dehan Kong"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "fcca36d9a28ace835ede3fa5a93cb9c54608f164240c18bfe007bc945273b460", "submitted_at": "2026-03-05T10:51:34Z", "updated_at": "2026-04-14T08:14:14Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2603.05044v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2603.05044", "identifiers": {"arxiv": "2603.05044"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2603.05044"}], "name": "WebFactory: Automated Compression of Foundational Language Intelligence into Grounded Web Agents", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2026-03-05", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces WebFactory, a GUI-agent training pipeline that synthesizes websites, generates tasks and trajectories, and trains with decomposed rewards. Offline and online transfer evaluations examine agents trained on the generated environments.", "topics": ["data-generation", "reinforcement-learning", "web-navigation"], "updated_at": "2026-04-14", "url": "https://arxiv.org/abs/2603.05044"}
{"aliases": [], "authors": ["Sicheng Fan", "Rui Wan", "Yifei Leng", "Gaoning Liang", "Li Ling", "Yanyi Shang", "Dehan Kong"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "db52246d9f5f305046e6dd14f2ef15440c371678e1db0d05282c84a3667c6f10", "submitted_at": "2026-03-05T15:37:34Z", "updated_at": "2026-04-14T08:41:56Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2603.05295v3"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:56Z", "role": "primary", "sha256": "c22be2d5c6a84b2c367759ce2b8f41f5a178fccd0534e832802c9d57b40918e8", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2601.16206%2C2601.17722%2C2601.18113%2C2601.18842%2C2601.20162%2C2601.20335%2C2601.20380%2C2601.20650%2C2601.21123%2C2601.22781%2C2602.00575%2C2602.01576%2C2602.01725%2C2602.02419%2C2602.02548%2C2602.02995%2C2602.03255%2C2602.03792%2C2602.05429%2C2602.05832%2C2602.06075%2C2602.06391%2C2602.07153%2C2602.07624%2C2602.08235%2C2602.08995%2C2602.09082%2C2602.09222%2C2602.09310%2C2602.09662%2C2602.10139%2C2602.11514%2C2602.11724%2C2602.11750%2C2602.14093%2C2602.14276%2C2602.14296%2C2602.14721%2C2602.16855%2C2602.17049%2C2602.17365%2C2602.20502%2C2602.21858%2C2602.22190%2C2603.05044%2C2603.05295%2C2603.07978%2C2603.08533%2C2603.10178%2C2603.12138&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:56Z", "role": "primary", "sha256": "c22be2d5c6a84b2c367759ce2b8f41f5a178fccd0534e832802c9d57b40918e8", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2601.16206%2C2601.17722%2C2601.18113%2C2601.18842%2C2601.20162%2C2601.20335%2C2601.20380%2C2601.20650%2C2601.21123%2C2601.22781%2C2602.00575%2C2602.01576%2C2602.01725%2C2602.02419%2C2602.02548%2C2602.02995%2C2602.03255%2C2602.03792%2C2602.05429%2C2602.05832%2C2602.06075%2C2602.06391%2C2602.07153%2C2602.07624%2C2602.08235%2C2602.08995%2C2602.09082%2C2602.09222%2C2602.09310%2C2602.09662%2C2602.10139%2C2602.11514%2C2602.11724%2C2602.11750%2C2602.14093%2C2602.14276%2C2602.14296%2C2602.14721%2C2602.16855%2C2602.17049%2C2602.17365%2C2602.20502%2C2602.21858%2C2602.22190%2C2603.05044%2C2603.05295%2C2603.07978%2C2603.08533%2C2603.10178%2C2603.12138&max_results=50"}], "id": "arxiv:2603.05295", "identifiers": {"arxiv": "2603.05295"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:56Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2603.05295"}], "name": "WebChain: A Large-Scale Human-Annotated Dataset of Real-World Web Interaction Traces", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-03-05", "relations": [], "review_status": "curated", "scope": "core", "summary": "WebChain aligns human web-interaction trajectories with visual, structural and action data for agent training. It also introduces WebChainBench and a mid-training approach that separates spatial grounding from planning.", "topics": ["data-generation", "evaluation", "grounding", "planning"], "updated_at": "2026-04-14", "url": "https://arxiv.org/abs/2603.05295"}
{"aliases": [], "authors": ["Jiateng Liu", "Zhenhailong Wang", "Rushi Wang", "Bingxuan Li", "Jeonghwan Kim", "Aditi Tiwari", "Pengfei Yu", "Denghui Zhang", "Heng Ji"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "77eb68f31d616b8e542dbb41ada67f28bdc21f29f7da3f04a0077c0fa2d10196", "submitted_at": "2026-03-09T05:27:56Z", "updated_at": "2026-03-09T05:27:56Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2603.07978v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2603.07978", "identifiers": {"arxiv": "2603.07978"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2603.07978"}], "name": "OSExpert: Computer-Use Agents Learning Professional Skills via Exploration", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2026-03-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "Uses GUI-based depth-first exploration to verify elementary software functions, assemble action primitives, and construct curricula of composed skills. OSExpert-Eval examines how the resulting procedural knowledge supports complex computer-use tasks.", "topics": ["planning", "data-generation", "evaluation"], "updated_at": "2026-03-09", "url": "https://arxiv.org/abs/2603.07978"}
{"aliases": [], "authors": ["Hyungjoo Chae", "Jungsoo Park", "Alan Ritter"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "178b2140d0c70ded39fede20bc91d1459d0a4bf65c5cb6a63c09c9db73828513", "submitted_at": "2026-03-11T07:58:53Z", "updated_at": "2026-03-11T07:58:53Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2603.10505v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2603.10505", "identifiers": {"arxiv": "2603.10505"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2603.10505"}], "name": "Safe and Scalable Web Agent Learning via Recreated Websites", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced. The abstract promises a release upon acceptance; this annotation does not verify artifact availability.", "published_at": "2026-03-11", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces VeriEnv, which recreates websites as executable training environments and exposes internal state through a Python SDK for task generation and deterministic reward checks. Experiments examine web-agent learning across new and expanded site environments.", "topics": ["data-generation", "sandbox", "reinforcement-learning"], "updated_at": "2026-03-11", "url": "https://arxiv.org/abs/2603.10505"}
{"aliases": [], "authors": ["Rui Shao", "Ruize Gao", "Bin Xie", "Yixing Li", "Kaiwen Zhou", "Shuai Wang", "Weili Guan", "Gongwei Chen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6421a7928b502cd431964e6f84a1a58d4cd16dbbf1d9b3d1ecc1c2b3dae863bf", "submitted_at": "2026-03-12T16:40:59Z", "updated_at": "2026-03-12T16:40:59Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2603.12138v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2603.12138", "identifiers": {"arxiv": "2603.12138"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2603.12138"}], "name": "HATS: Hardness-Aware Trajectory Synthesis for GUI Agents", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2026-03-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces HATS, which directs GUI trajectory exploration toward semantically ambiguous actions and iteratively checks instruction-execution alignment. Feedback from trajectory refinement updates the exploration signal used to collect subsequent training examples.", "topics": ["data-generation", "robustness", "action-prediction"], "updated_at": "2026-03-12", "url": "https://arxiv.org/abs/2603.12138"}
{"aliases": [], "authors": ["Nathan Zhao"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e3b429d277ec776af18977a72aaeb95000a4fedf66aa89c0366ab02120d4cdbe", "submitted_at": "2026-03-18T04:41:16Z", "updated_at": "2026-03-18T04:41:16Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2603.17357v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2603.17357", "identifiers": {"arxiv": "2603.17357"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:39Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2603.17357"}], "name": "WebPII: Benchmarking Visual PII Detection for Computer-Use Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-03-18", "relations": [], "review_status": "curated", "scope": "core", "summary": "WebPII provides synthetic e-commerce screenshots with fine-grained annotations for sensitive information, including partially filled forms and transaction identifiers. It also trains WebRedact to detect visual PII as a privacy component for computer-use agents.", "topics": ["evaluation", "data-generation", "safety-security", "screen-understanding"], "updated_at": "2026-03-18", "url": "https://arxiv.org/abs/2603.17357"}
{"aliases": [], "authors": ["Xiangru Jian", "Shravan Nayak", "Kevin Qinghong Lin", "Aarash Feizi", "Kaixin Li", "Patrice Bechard", "Spandana Gella", "Sai Rajeswar"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "15237fda2f3799314009635017487749dbedc9b1a0ffdd6b56a8d28eaf01481d", "submitted_at": "2026-03-25T15:52:56Z", "updated_at": "2026-03-25T15:52:56Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2603.24440v1"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2603.24440", "identifiers": {"arxiv": "2603.24440"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:39Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2603.24440"}], "name": "CUA-Suite: Massive Human-annotated Video Demonstrations for Computer-Use Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-03-25", "relations": [], "review_status": "curated", "scope": "core", "summary": "CUA-Suite combines continuous expert desktop demonstrations in VideoCUA with the UI-Vision evaluation benchmark and GroundCUA grounding data. Screen recordings, cursor traces and layered annotations support study of temporal behavior and professional software workflows.", "topics": ["data-generation", "video", "evaluation", "grounding", "workflow"], "updated_at": "2026-03-25", "url": "https://arxiv.org/abs/2603.24440"}
{"aliases": [], "authors": ["Zichuan Lin", "Feiyu Liu", "Yijun Yang", "Jiafei Lyu", "Yiming Gao", "Yicheng Liu", "Zhicong Lu", "Yangbin Yu", "Mingyu Yang", "Junyou Li", "Deheng Ye", "Jie Jiang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "aa03c8716edf782ce1a86e404107aa2569be63484dfedb472d3710dc82f405d0", "submitted_at": "2026-03-25T17:10:29Z", "updated_at": "2026-03-25T17:10:29Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2603.24533v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2603.24533", "identifiers": {"arxiv": "2603.24533"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:39Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2603.24533"}], "name": "UI-Voyager: A Self-Evolving GUI Agent Learning via Failed Experience", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2026-03-25", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces a two-stage mobile GUI learner that first co-evolves trajectories and policies through rejection fine-tuning. Group Relative Self-Distillation then uses successful rollout branches to provide step-level supervision for failed trajectories, evaluated on AndroidWorld.", "topics": ["imitation-learning", "data-generation", "mobile-navigation"], "updated_at": "2026-03-25", "url": "https://arxiv.org/abs/2603.24533"}
{"aliases": [], "authors": ["Mengzhou Wu", "Yuzhe Guo", "Yuan Cao", "Haochuan Lu", "Songhe Zhu", "Pingzhe Qu", "Xin Chen", "Kang Qin", "Zhongpu Wang", "Xiaode Zhang", "Xinyi Wang", "Wei Dai", "Gang Cao", "Yuetang Deng", "Zhi Gong", "Dezhi Ran", "Linyi Li", "Wei Yang", "Tao Xie"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "bac86c2ff2bf18feb3edfbe76285d33718d0c6d9564aff5a9d62f8778cdc5147", "submitted_at": "2026-02-11T17:02:38Z", "updated_at": "2026-02-11T17:02:38Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2604.02345v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2604.02345", "identifiers": {"arxiv": "2604.02345"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2604.02345"}], "name": "UI-Oceanus: Scaling GUI Agents with Synthetic Environmental Dynamics", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-02-11", "relations": [], "review_status": "curated", "scope": "core", "summary": "UI-Oceanus converts autonomous GUI exploration and execution feedback into supervision for predicting future interface states. It compares forward and inverse objectives during continual pretraining and examines downstream offline and online navigation as synthetic data scales.", "topics": ["world-models", "data-generation", "state-tracking", "evaluation"], "updated_at": "2026-02-11", "url": "https://arxiv.org/abs/2604.02345"}
{"aliases": [], "authors": ["Pranjal Aggarwal", "Graham Neubig", "Sean Welleck"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "fdd9cb6daad8832261ebe861fc1cb74429b819305e14ecfeb9698616013e6ac8", "submitted_at": "2026-04-07T17:38:15Z", "updated_at": "2026-04-07T17:38:15Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2604.06126v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:32:00Z", "role": "primary", "sha256": "4d09ebfb093e43a4c44adef86c8fa8374f90d907fc33bd8e12a1315e62d5a583", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2603.12823%2C2603.14707%2C2603.15039%2C2603.17357%2C2603.18429%2C2603.19191%2C2603.24440%2C2603.25226%2C2603.25864%2C2603.26211%2C2603.26648%2C2603.29318%2C2604.01535%2C2604.02947%2C2604.04399%2C2604.05477%2C2604.06126%2C2604.06182%2C2604.06367%2C2604.07429%2C2604.08455%2C2604.08516%2C2604.08523%2C2604.09155%2C2604.09574%2C2604.09587%2C2604.09815%2C2604.09937%2C2604.11259%2C2604.11784%2C2604.12284%2C2604.13019%2C2604.13822%2C2604.14113%2C2604.14262%2C2604.14956%2C2604.15093%2C2604.15309%2C2604.17817%2C2604.17821%2C2604.18860%2C2604.24964%2C2604.25380%2C2604.25562%2C2604.27151%2C2604.27253%2C2604.27776%2C2605.00642%2C2605.06393%2C2605.06664&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:32:00Z", "role": "primary", "sha256": "4d09ebfb093e43a4c44adef86c8fa8374f90d907fc33bd8e12a1315e62d5a583", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?id_list=2603.12823%2C2603.14707%2C2603.15039%2C2603.17357%2C2603.18429%2C2603.19191%2C2603.24440%2C2603.25226%2C2603.25864%2C2603.26211%2C2603.26648%2C2603.29318%2C2604.01535%2C2604.02947%2C2604.04399%2C2604.05477%2C2604.06126%2C2604.06182%2C2604.06367%2C2604.07429%2C2604.08455%2C2604.08516%2C2604.08523%2C2604.09155%2C2604.09574%2C2604.09587%2C2604.09815%2C2604.09937%2C2604.11259%2C2604.11784%2C2604.12284%2C2604.13019%2C2604.13822%2C2604.14113%2C2604.14262%2C2604.14956%2C2604.15093%2C2604.15309%2C2604.17817%2C2604.17821%2C2604.18860%2C2604.24964%2C2604.25380%2C2604.25562%2C2604.27151%2C2604.27253%2C2604.27776%2C2605.00642%2C2605.06393%2C2605.06664&max_results=50"}], "id": "arxiv:2604.06126", "identifiers": {"arxiv": "2604.06126"}, "kind": "paper", "last_verified_at": "2026-10-16T06:32:00Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2604.06126"}], "name": "Gym-Anything: Turn any Software into an Agent Environment", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2026-04-07", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces Gym-Anything, where a setup agent configures software environments and an independent auditor checks the resulting evidence. CUA-World supplies long-horizon tasks across professional applications, with experiments on trajectory distillation and test-time auditing.", "topics": ["data-generation", "sandbox", "long-horizon", "evaluation"], "updated_at": "2026-04-07", "url": "https://arxiv.org/abs/2604.06126"}
{"aliases": [], "authors": ["Tanmay Gupta", "Piper Wolters", "Zixian Ma", "Peter Sushko", "Rock Yuren Pang", "Diego Llanes", "Yue Yang", "Taira Anderson", "Boyuan Zheng", "Zhongzheng Ren", "Harsh Trivedi", "Taylor Blanton", "Caleb Ouellette", "Winson Han", "Ali Farhadi", "Ranjay Krishna"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "dbc0d512a11b88a9036d06d0fcc26a8ab0d0bc54fad986d4b6225c8073897bc2", "submitted_at": "2026-04-09T17:54:02Z", "updated_at": "2026-04-09T17:54:02Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2604.08516v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2604.08516", "identifiers": {"arxiv": "2604.08516"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:39Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2604.08516"}], "name": "MolmoWeb: Open Visual Web Agent and Open Data for the Open Web", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced. The abstract states that artifacts will be released; this annotation does not independently verify release status or licenses.", "published_at": "2026-04-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "Presents MolmoWebMix, combining synthetic browser trajectories, human demonstrations and GUI perception data, and MolmoWeb screenshot-to-action policies trained from it. Browser benchmarks also test parallel rollouts with best-of-N selection.", "topics": ["web-navigation", "data-generation", "imitation-learning"], "updated_at": "2026-04-09", "url": "https://arxiv.org/abs/2604.08516"}
{"aliases": [], "authors": ["Tiantian He", "Yihang Chen", "Keyue Jiang", "Ka Yiu Lee", "Kaiwen Zhou", "Kun Shao", "Shuai Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "dd2bc0ee1955dbc8e98a9306a3d6260251cbab89d36ac117e4dc49b6fa99766e", "submitted_at": "2026-04-10T18:46:05Z", "updated_at": "2026-04-10T18:46:05Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2604.09815v1"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2604.09815", "identifiers": {"arxiv": "2604.09815"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:39Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2604.09815"}], "name": "EE-MCP: Self-Evolving MCP-GUI Agents via Automated Environment Generation and Experience Learning", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-04-10", "relations": [], "review_status": "curated", "scope": "core", "summary": "EE-MCP studies application-dependent mixtures of GUI actions and MCP calls using generated environments, trajectory collection and filtered training. An experience bank extracts rules from trajectory comparisons for inference-time improvement, distinct from model distillation.", "topics": ["tool-integration", "data-generation", "imitation-learning", "memory"], "updated_at": "2026-04-10", "url": "https://arxiv.org/abs/2604.09815"}
{"aliases": [], "authors": ["Peng Yuan", "Yuyang Yin", "Yuxuan Cai", "Zheng Wei"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "1db8d789fb8ba2d5e61ca24536670fd83491c305f532adfecf7193821f7eea52", "submitted_at": "2026-04-13T04:45:27Z", "updated_at": "2026-04-13T04:45:27Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2604.10988v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2604.10988", "identifiers": {"arxiv": "2604.10988"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:39Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2604.10988"}], "name": "WebForge: Breaking the Realism-Reproducibility-Scalability Trilemma in Browser Agent Benchmark", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2026-04-13", "relations": [], "review_status": "curated", "scope": "core", "summary": "Presents WebForge, a pipeline that plans, generates, refines and validates self-contained interactive websites. WebForge-Bench uses these environments to evaluate browser agents across domains and controlled difficulty levels.", "topics": ["data-generation", "evaluation", "web-navigation"], "updated_at": "2026-04-13", "url": "https://arxiv.org/abs/2604.10988"}
{"aliases": [], "authors": ["Yulin Chen", "Tri Cao", "Haoran Li", "Yue Liu", "Yibo Li", "Yufei He", "Le Minh Khoi", "Yangqiu Song", "Shuicheng Yan", "Bryan Hooi"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "64d4323050a5ad0f7265231337eb22a4c0823cee41587099c8bdc5db6b320dea", "submitted_at": "2026-04-14T04:50:35Z", "updated_at": "2026-04-14T04:50:35Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2604.12284v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2604.12284", "identifiers": {"arxiv": "2604.12284"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:39Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2604.12284"}], "name": "WebAgentGuard: A Reasoning-Driven Guard Model for Detecting Prompt Injection Attacks in Web Agents", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-04-14", "relations": [], "review_status": "curated", "scope": "core", "summary": "WebAgentGuard separates prompt-injection detection from the operating web agent through a parallel multimodal guard. The guard is trained on synthetic visual/textual examples using reasoning-focused supervised fine-tuning followed by reinforcement learning.", "topics": ["safety-security", "data-generation", "imitation-learning", "reinforcement-learning", "evaluation"], "updated_at": "2026-04-14", "url": "https://arxiv.org/abs/2604.12284"}
{"aliases": [], "authors": ["Kanzhi Cheng", "Zehao Li", "Zheng Ma", "Nuo Chen", "Jialin Cao", "Qiushi Sun", "Zichen Ding", "Fangzhi Xu", "Hang Yan", "Jiajun Chen", "Anh Tuan Luu", "Jianbing Zhang", "Lewei Lu", "Dahua Lin"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "12c2e5e2ae9bdc871bb63d195194c5387129accf5e8b26d12b21650b92fe82b6", "submitted_at": "2026-04-16T14:53:08Z", "updated_at": "2026-04-16T14:53:08Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2604.15093v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:32:00Z", "role": "primary", "sha256": "4d09ebfb093e43a4c44adef86c8fa8374f90d907fc33bd8e12a1315e62d5a583", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2603.12823%2C2603.14707%2C2603.15039%2C2603.17357%2C2603.18429%2C2603.19191%2C2603.24440%2C2603.25226%2C2603.25864%2C2603.26211%2C2603.26648%2C2603.29318%2C2604.01535%2C2604.02947%2C2604.04399%2C2604.05477%2C2604.06126%2C2604.06182%2C2604.06367%2C2604.07429%2C2604.08455%2C2604.08516%2C2604.08523%2C2604.09155%2C2604.09574%2C2604.09587%2C2604.09815%2C2604.09937%2C2604.11259%2C2604.11784%2C2604.12284%2C2604.13019%2C2604.13822%2C2604.14113%2C2604.14262%2C2604.14956%2C2604.15093%2C2604.15309%2C2604.17817%2C2604.17821%2C2604.18860%2C2604.24964%2C2604.25380%2C2604.25562%2C2604.27151%2C2604.27253%2C2604.27776%2C2605.00642%2C2605.06393%2C2605.06664&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:32:00Z", "role": "primary", "sha256": "4d09ebfb093e43a4c44adef86c8fa8374f90d907fc33bd8e12a1315e62d5a583", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?id_list=2603.12823%2C2603.14707%2C2603.15039%2C2603.17357%2C2603.18429%2C2603.19191%2C2603.24440%2C2603.25226%2C2603.25864%2C2603.26211%2C2603.26648%2C2603.29318%2C2604.01535%2C2604.02947%2C2604.04399%2C2604.05477%2C2604.06126%2C2604.06182%2C2604.06367%2C2604.07429%2C2604.08455%2C2604.08516%2C2604.08523%2C2604.09155%2C2604.09574%2C2604.09587%2C2604.09815%2C2604.09937%2C2604.11259%2C2604.11784%2C2604.12284%2C2604.13019%2C2604.13822%2C2604.14113%2C2604.14262%2C2604.14956%2C2604.15093%2C2604.15309%2C2604.17817%2C2604.17821%2C2604.18860%2C2604.24964%2C2604.25380%2C2604.25562%2C2604.27151%2C2604.27253%2C2604.27776%2C2605.00642%2C2605.06393%2C2605.06664&max_results=50"}], "id": "arxiv:2604.15093", "identifiers": {"arxiv": "2604.15093"}, "kind": "paper", "last_verified_at": "2026-10-16T06:32:00Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2604.15093"}], "name": "OpenMobile: Building Open Mobile Agents with Task and Trajectory Synthesis", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2026-04-16", "relations": [], "review_status": "curated", "scope": "core", "summary": "Presents OpenMobile, which builds exploration-based environment memory to synthesize task instructions and switches between learner and expert policies during trajectory collection. The resulting data includes recovery behavior and is tested through mobile-agent training and benchmark-overlap analysis.", "topics": ["data-generation", "imitation-learning", "mobile-navigation"], "updated_at": "2026-04-16", "url": "https://arxiv.org/abs/2604.15093"}
{"aliases": [], "authors": ["Hongxin Li", "Xiping Wang", "Jingran Su", "Zheng Ju", "Yuntao Chen", "Qing Li", "Zhaoxiang Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "65325912365effe8adbc39301a378d7526c60d1185691f3a89cb8acd4aa53c3f", "submitted_at": "2026-04-27T13:06:27Z", "updated_at": "2026-04-27T13:06:27Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2604.24441v1"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:02Z", "role": "primary", "sha256": "e436adf2735cb537ee4b24aa0c3f36930a50c2f683cc6868e7c40df602bdb6b8", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2604.00892%2C2604.01221%2C2604.01676%2C2604.02345%2C2604.03465%2C2604.05157%2C2604.06134%2C2604.06240%2C2604.06995%2C2604.07277%2C2604.07831%2C2604.07929%2C2604.08005%2C2604.09581%2C2604.10261%2C2604.10577%2C2604.10988%2C2604.11201%2C2604.18847%2C2604.21375%2C2604.24441%2C2604.26020%2C2605.00551%2C2605.02168%2C2605.07505%2C2605.10347%2C2605.12294%2C2605.15040%2C2605.16883%2C2605.27134%2C2605.28258%2C2606.01790%2C2606.02031%2C2606.02208%2C2606.03203%2C2606.03236%2C2606.03951%2C2606.04391%2C2606.04627%2C2606.04701%2C2606.05233%2C2606.05597%2C2606.05697%2C2606.06560%2C2606.06708%2C2606.07027%2C2606.07074%2C2606.09399%2C2606.09447%2C2606.09764&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:02Z", "role": "primary", "sha256": "e436adf2735cb537ee4b24aa0c3f36930a50c2f683cc6868e7c40df602bdb6b8", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?id_list=2604.00892%2C2604.01221%2C2604.01676%2C2604.02345%2C2604.03465%2C2604.05157%2C2604.06134%2C2604.06240%2C2604.06995%2C2604.07277%2C2604.07831%2C2604.07929%2C2604.08005%2C2604.09581%2C2604.10261%2C2604.10577%2C2604.10988%2C2604.11201%2C2604.18847%2C2604.21375%2C2604.24441%2C2604.26020%2C2605.00551%2C2605.02168%2C2605.07505%2C2605.10347%2C2605.12294%2C2605.15040%2C2605.16883%2C2605.27134%2C2605.28258%2C2606.01790%2C2606.02031%2C2606.02208%2C2606.03203%2C2606.03236%2C2606.03951%2C2606.04391%2C2606.04627%2C2606.04701%2C2606.05233%2C2606.05597%2C2606.05697%2C2606.06560%2C2606.06708%2C2606.07027%2C2606.07074%2C2606.09399%2C2606.09447%2C2606.09764&max_results=50"}], "id": "arxiv:2604.24441", "identifiers": {"arxiv": "2604.24441"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:02Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2604.24441"}], "name": "AutoGUI-v2: A Comprehensive Multi-Modal GUI Functionality Understanding Benchmark", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-04-27", "relations": [], "review_status": "curated", "scope": "core", "summary": "AutoGUI-v2 evaluates interface functionality, grounding and predicted interaction outcomes using hierarchical regions derived through VLM-human annotation. Its tasks span multiple operating systems and expose differences between functional localization, captioning and transition-logic understanding.", "topics": ["screen-understanding", "grounding", "state-tracking", "evaluation", "data-generation"], "updated_at": "2026-04-27", "url": "https://arxiv.org/abs/2604.24441"}
{"aliases": [], "authors": ["Fazle Elahi Faisal", "Qianhui Wu", "Baolin Peng", "Jianfeng Gao"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "276d8a62192c0ad576a1d986bb27d3978963b93466a822858e3313957dd83a81", "submitted_at": "2026-04-29T22:57:35Z", "updated_at": "2026-04-29T22:57:35Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2604.27253v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2604.27253", "identifiers": {"arxiv": "2604.27253"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:39Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2604.27253"}], "name": "AutoSurfer -- Teaching Web Agents through Comprehensive Surfing, Learning, and Modeling", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-04-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "AutoSurfer explores website actions breadth-first, retaining page and trajectory knowledge to guide task synthesis and trajectory refinement. It evaluates the resulting training data by fine-tuning a web agent on WebArena and examining synthesized-task diversity.", "topics": ["web-navigation", "data-generation", "imitation-learning", "evaluation"], "updated_at": "2026-04-29", "url": "https://arxiv.org/abs/2604.27253"}
{"aliases": [], "authors": ["Oğuzhan Fatih Kar", "Roman Bachmann", "Yuanzheng Gong", "Anders Boesen Lindbo Larsen", "Afshin Dehghan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2e1fb158dad88d32f8754ac27f12346ba8fbe5f9a5e736fba3e81390071895a3", "submitted_at": "2026-05-07T17:17:10Z", "updated_at": "2026-05-07T17:17:10Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2605.06761v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2605.06761", "identifiers": {"arxiv": "2605.06761"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:39Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2605.06761"}], "name": "Weblica: Scalable and Reproducible Training Environments for Visual Web Agents", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2026-05-07", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces Weblica, which combines HTTP-level capture and replay with website-grounded environment synthesis. The resulting interactive web environments support reinforcement learning and tests of navigation transfer and inference-time scaling.", "topics": ["sandbox", "data-generation", "reinforcement-learning"], "updated_at": "2026-05-07", "url": "https://arxiv.org/abs/2605.06761"}
{"aliases": [], "authors": ["Weikai Xu", "Kun Huang", "Yunren Feng", "Jiaxing Li", "Yuhan Chen", "Yuxuan Liu", "Zhizheng Jiang", "Heng Qu", "Pengzhi Gao", "Wei Liu", "Jian Luan", "Xiaolin Hu", "Bo An"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "4374ec1b7c44804bb78ea56bd0aa3214083419f37780d79c5fcfb68468727bbf", "submitted_at": "2026-05-11T10:49:31Z", "updated_at": "2026-05-22T05:43:30Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2605.10347v2"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2605.10347", "identifiers": {"arxiv": "2605.10347"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:39Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2605.10347"}], "name": "How Mobile World Model Guides GUI Agents?", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2026-05-11", "relations": [], "review_status": "curated", "scope": "core", "summary": "Compares mobile world models that predict state changes as delta text, full text, images or renderable code. Downstream Android benchmarks examine synthetic training trajectories and test-time guidance, including limits of post-action reflection.", "topics": ["world-models", "data-generation", "evaluation"], "updated_at": "2026-05-22", "url": "https://arxiv.org/abs/2605.10347"}
{"aliases": [], "authors": ["Xuhao Hu", "Xi Zhang", "Haiyang Xu", "Kyle Qiao", "Jingyi Yang", "Xuanjing Huang", "Jing Shao", "Ming Yan", "Jieping Ye"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "09128e90b925ac0cd2cf6e967853180bbef1e97a0960c22065f1b0b50c30d58b", "submitted_at": "2026-05-12T17:57:04Z", "updated_at": "2026-09-19T14:53:46Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2605.12481v2"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2605.12481", "identifiers": {"arxiv": "2605.12481"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:39Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2605.12481"}], "name": "ToolCUA: Towards Optimal GUI-Tool Path Orchestration for Computer Use Agents", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-05-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "ToolCUA synthesizes interleaved GUI/tool trajectories and trains agents to choose between graphical actions and higher-level tools. Staged supervised and reinforcement learning culminate in online hybrid-environment training with a reward for efficient execution paths.", "topics": ["tool-integration", "data-generation", "imitation-learning", "reinforcement-learning", "planning", "efficiency"], "updated_at": "2026-09-19", "url": "https://arxiv.org/abs/2605.12481"}
{"aliases": [], "authors": ["Miaosen Zhang", "Xiaohan Zhao", "Zhihong Tan", "Zhou Huoshen", "Yijia Fan", "Yifan Yang", "Kai Qiu", "Bei Liu", "Justin Wagle", "Chenzhong Yin", "Mingxi Cheng", "Ji Li", "Qi Dai", "Chong Luo", "Xu Yang", "Xin Geng", "Baining Guo"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "1fb57ba3cd858513d1da58d64a7c85388f12e331c9ea9f89204be5f43e6fec38", "submitted_at": "2026-05-12T17:59:58Z", "updated_at": "2026-05-12T17:59:58Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2605.12501v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2605.12501", "identifiers": {"arxiv": "2605.12501"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:39Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2605.12501"}], "name": "Covering Human Action Space for Computer Use: Data Synthesis and Benchmark", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-05-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces CUActSpot to evaluate clicks, drags, drawing and other screen interactions across GUI, text, tables, canvases and natural images. A renderer-based pipeline records geometry and screenshots, then generates matching instructions and action traces for training a grounding model.", "topics": ["grounding", "data-generation", "evaluation", "action-prediction"], "updated_at": "2026-05-12", "url": "https://arxiv.org/abs/2605.12501"}
{"aliases": [], "authors": ["Haoren Zhao", "Tianyi Chen", "Zhen Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "671f33dd87a7ebf9233494ef6fa0efd9c9772a3951f3c218bf4de822825f7af2", "submitted_at": "2026-05-13T02:48:52Z", "updated_at": "2026-05-13T02:48:52Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2605.16402v1"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2605.16402", "identifiers": {"arxiv": "2605.16402"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:39Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2605.16402"}], "name": "WinDeskGround: A Benchmark for Robust GUI Grounding in Complex Multi-Window Desktop Environments", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-05-13", "relations": [], "review_status": "curated", "scope": "core", "summary": "WinDeskGround synthesizes multi-window desktop layouts with controlled occlusion, density and semantic similarity to test GUI grounding robustness. Its instruction-target benchmark measures how localization changes when interfaces depart from simple single-layer scenes.", "topics": ["grounding", "evaluation", "robustness", "data-generation"], "updated_at": "2026-05-13", "url": "https://arxiv.org/abs/2605.16402"}
{"aliases": [], "authors": ["Athar Parvez", "Muhammad Jawad Mufti", "Muqaddas Gull", "Omar Hammad"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "ff95e0b071508592190ba236741c1c01200dd38fdc74c31b2cc4d1bf8143ae9c", "submitted_at": "2026-05-17T21:14:32Z", "updated_at": "2026-05-17T21:14:32Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2605.17656v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T12:31:23Z", "role": "primary", "sha256": "9a71e95803ce58c462c84e2b45a891555e31d1fb409e643bb3ca4675a44c96e6", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2605.12233%2C2605.12493%2C2605.12978%2C2605.13119%2C2605.14133%2C2605.16909%2C2605.17260%2C2605.17324%2C2605.17448%2C2605.17656%2C2605.17721%2C2605.19447%2C2605.19587%2C2605.20894%2C2605.22642%2C2605.23204%2C2605.25233%2C2605.25901%2C2605.26086%2C2605.26144%2C2605.26252%2C2605.26646%2C2605.27209%2C2605.27705%2C2605.27898%2C2605.28158%2C2605.28255%2C2605.29341%2C2605.29463%2C2605.29801%2C2605.30010%2C2605.30052%2C2605.30126%2C2605.30434%2C2605.30907%2C2605.30914%2C2605.31264%2C2605.31557%2C2606.00530%2C2606.00543%2C2606.00856%2C2606.01057%2C2606.01961%2C2606.02540%2C2606.03103%2C2606.03889%2C2606.03895%2C2606.04315%2C2606.04455%2C2606.05525&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:31:23Z", "role": "primary", "sha256": "9a71e95803ce58c462c84e2b45a891555e31d1fb409e643bb3ca4675a44c96e6", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2605.12233%2C2605.12493%2C2605.12978%2C2605.13119%2C2605.14133%2C2605.16909%2C2605.17260%2C2605.17324%2C2605.17448%2C2605.17656%2C2605.17721%2C2605.19447%2C2605.19587%2C2605.20894%2C2605.22642%2C2605.23204%2C2605.25233%2C2605.25901%2C2605.26086%2C2605.26144%2C2605.26252%2C2605.26646%2C2605.27209%2C2605.27705%2C2605.27898%2C2605.28158%2C2605.28255%2C2605.29341%2C2605.29463%2C2605.29801%2C2605.30010%2C2605.30052%2C2605.30126%2C2605.30434%2C2605.30907%2C2605.30914%2C2605.31264%2C2605.31557%2C2606.00530%2C2606.00543%2C2606.00856%2C2606.01057%2C2606.01961%2C2606.02540%2C2606.03103%2C2606.03889%2C2606.03895%2C2606.04315%2C2606.04455%2C2606.05525&max_results=50"}], "id": "arxiv:2605.17656", "identifiers": {"arxiv": "2605.17656"}, "kind": "paper", "last_verified_at": "2026-10-17T12:31:23Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2605.17656"}], "name": "MUIAnno: An Expert-Annotated Dataset and Evaluation Benchmark for Mobile UI Understanding", "notes": "Reviewed against the complete retained official abstract. Experimental findings and release availability have not been independently reproduced or verified.", "published_at": "2026-05-17", "relations": [], "review_status": "curated", "scope": "core", "summary": "MUIAnno provides expert labels for elements in manually explored mobile-app screens, with a web annotation tool that emits structured JSON. It also supplies UI-element detection baselines for the dataset.", "topics": ["screen-understanding", "grounding", "evaluation", "data-generation"], "updated_at": "2026-05-17", "url": "https://arxiv.org/abs/2605.17656"}
{"aliases": [], "authors": ["Ziyun Zeng", "Hang Hua", "Bocheng Zou", "Mu Cai", "Rogerio Feris", "Jiebo Luo"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d96c8844274f79adea167169a0853e3a7cd82fee4a8a666913f97898a0836e7c", "submitted_at": "2026-05-18T16:57:36Z", "updated_at": "2026-05-18T16:57:36Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2605.18652v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2605.18652", "identifiers": {"arxiv": "2605.18652"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2605.18652"}], "name": "MementoGUI: Learning Agentic Multimodal Memory Control for Long-Horizon GUI Agents", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-05-18", "relations": [], "review_status": "curated", "scope": "core", "summary": "MementoGUI learns a controller for selecting, compressing and retrieving textual and region-level visual memories while leaving the GUI-agent backbone fixed. It also converts computer-use traces into controller training data and evaluates semantic actions, progress and memory consistency on long-horizon tasks.", "topics": ["memory", "interaction-history", "long-horizon", "data-generation", "evaluation"], "updated_at": "2026-05-18", "url": "https://arxiv.org/abs/2605.18652"}
{"aliases": [], "authors": ["Jinbiao Wei", "Qianran Ma", "Yilun Zhao", "Xiao Zhou", "Kangqi Ni", "Guo Gan", "Arman Cohan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "0a93d880793018d1b47a39e0487e3be76550fd2b49de9535b2913e9f08159074", "submitted_at": "2026-05-19T12:40:29Z", "updated_at": "2026-05-19T12:40:29Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2605.19769v1"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2605.19769", "identifiers": {"arxiv": "2605.19769"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2605.19769"}], "name": "OpenComputer: Verifiable Software Worlds for Computer-Use Agents", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2026-05-19", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces OpenComputer, which builds desktop tasks around application-state verifiers and refines those verifiers using execution feedback. An evaluation harness records trajectories and assigns auditable partial credit across professional software workflows.", "topics": ["evaluation", "data-generation", "observability"], "updated_at": "2026-05-19", "url": "https://arxiv.org/abs/2605.19769"}
{"aliases": [], "authors": ["Fatemeh Pesaran Zadeh", "Seyeon Choi", "Xing Han Lù", "Siva Reddy", "Gunhee Kim"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "aa6e11bae4d952d79276543ced6f81193b36868a6222fe51ee7b0816aa72fbab", "submitted_at": "2026-05-19T09:19:01Z", "updated_at": "2026-05-26T02:53:26Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2605.20291v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2605.20291", "identifiers": {"arxiv": "2605.20291"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2605.20291"}], "name": "Weasel: Out-of-Domain Generalization for Web Agents via Importance-Diversity Data Selection", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-05-19", "relations": [], "review_status": "curated", "scope": "core", "summary": "Weasel selects web-agent training steps by balancing importance and diversity across states, sites and interaction patterns. It also prunes training accessibility trees around known action targets and regenerates rationales to match the model’s reasoning style.", "topics": ["web-navigation", "imitation-learning", "efficiency", "robustness", "data-generation"], "updated_at": "2026-05-26", "url": "https://arxiv.org/abs/2605.20291"}
{"aliases": [], "authors": ["Guohong Liu", "Jialei Ye", "Pengzhi Gao", "Wei Liu", "Jian Luan", "Yunxin Liu", "Yuanchun Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "eb40c8bf005d56139d263e1b5dc4553ddcd68522124686b7f0865e6d8d1725ad", "submitted_at": "2026-05-24T16:33:14Z", "updated_at": "2026-06-18T02:24:07Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2605.25160v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2605.25160", "identifiers": {"arxiv": "2605.25160"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2605.25160"}], "name": "ScaleWoB: Guiding GUI Agents with Coding Agents via Large-Scale Environmental Synthesis", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced. Mobile, desktop and automotive interfaces are simulated as webpages; this does not assert native deployment on those platforms.", "published_at": "2026-05-24", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces ScaleWoB, which synthesizes interactive GUI environments as backend-free webpages with verifiable rewards. Its mobile-oriented benchmark evaluates agent execution and compares selected synthetic tasks with real-app behavior.", "topics": ["data-generation", "evaluation", "sandbox"], "updated_at": "2026-06-18", "url": "https://arxiv.org/abs/2605.25160"}
{"aliases": [], "authors": ["Bowen Wang", "Dunjie Lu", "Junli Wang", "Tianyi Bai", "Shixuan Liu", "Zhipeng Zhang", "Haiquan Wang", "Hao Hu", "Tianbao Xie", "Shuai Bai", "Dayiheng Liu", "Que Shen", "Junyang Lin", "Tao Yu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "ecc0712f55ee90c237f9a48e013e396b1082f932a485c22d8249d1da7acbf307", "submitted_at": "2026-05-25T09:28:03Z", "updated_at": "2026-06-08T08:32:35Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2605.25624v2"}}, "environments": ["web", "desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2605.25624", "identifiers": {"arxiv": "2605.25624"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2605.25624"}], "name": "CUA-Gym: Scaling Verifiable Training Environments and Tasks for Computer-Use Agents", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced. The abstract promises future open sourcing; this annotation does not independently verify artifact release or licensing.", "published_at": "2026-05-25", "relations": [], "review_status": "curated", "scope": "core", "summary": "Presents CUA-Gym, which co-generates instructions, initial and target states, and executable reward checks through generator-discriminator iterations. Mock web applications expand training environments, with learned agents evaluated on OSWorld-Verified and WebArena.", "topics": ["data-generation", "reinforcement-learning", "sandbox"], "updated_at": "2026-06-08", "url": "https://arxiv.org/abs/2605.25624"}
{"aliases": [], "authors": ["Yusong Lin", "Xinyuan Liang", "Haiyang Wang", "Qipeng Gu", "Siqi Cheng", "Jiangui Chen", "Shuzhe Wu", "Feiyang Pan", "Lue Fan", "Sanyuan Zhao", "Dandan Tu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "27999553928305cfa4223011feb3f9c2decda8a861e5800865537a81c7c08832", "submitted_at": "2026-05-25T17:50:04Z", "updated_at": "2026-05-25T17:50:04Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2605.26086v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T12:31:23Z", "role": "primary", "sha256": "9a71e95803ce58c462c84e2b45a891555e31d1fb409e643bb3ca4675a44c96e6", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2605.12233%2C2605.12493%2C2605.12978%2C2605.13119%2C2605.14133%2C2605.16909%2C2605.17260%2C2605.17324%2C2605.17448%2C2605.17656%2C2605.17721%2C2605.19447%2C2605.19587%2C2605.20894%2C2605.22642%2C2605.23204%2C2605.25233%2C2605.25901%2C2605.26086%2C2605.26144%2C2605.26252%2C2605.26646%2C2605.27209%2C2605.27705%2C2605.27898%2C2605.28158%2C2605.28255%2C2605.29341%2C2605.29463%2C2605.29801%2C2605.30010%2C2605.30052%2C2605.30126%2C2605.30434%2C2605.30907%2C2605.30914%2C2605.31264%2C2605.31557%2C2606.00530%2C2606.00543%2C2606.00856%2C2606.01057%2C2606.01961%2C2606.02540%2C2606.03103%2C2606.03889%2C2606.03895%2C2606.04315%2C2606.04455%2C2606.05525&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:31:23Z", "role": "primary", "sha256": "9a71e95803ce58c462c84e2b45a891555e31d1fb409e643bb3ca4675a44c96e6", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2605.12233%2C2605.12493%2C2605.12978%2C2605.13119%2C2605.14133%2C2605.16909%2C2605.17260%2C2605.17324%2C2605.17448%2C2605.17656%2C2605.17721%2C2605.19447%2C2605.19587%2C2605.20894%2C2605.22642%2C2605.23204%2C2605.25233%2C2605.25901%2C2605.26086%2C2605.26144%2C2605.26252%2C2605.26646%2C2605.27209%2C2605.27705%2C2605.27898%2C2605.28158%2C2605.28255%2C2605.29341%2C2605.29463%2C2605.29801%2C2605.30010%2C2605.30052%2C2605.30126%2C2605.30434%2C2605.30907%2C2605.30914%2C2605.31264%2C2605.31557%2C2606.00530%2C2606.00543%2C2606.00856%2C2606.01057%2C2606.01961%2C2606.02540%2C2606.03103%2C2606.03889%2C2606.03895%2C2606.04315%2C2606.04455%2C2606.05525&max_results=50"}], "id": "arxiv:2605.26086", "identifiers": {"arxiv": "2605.26086"}, "kind": "paper", "last_verified_at": "2026-10-17T12:31:23Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2605.26086"}], "name": "Claw-Anything: Benchmarking Always-On Personal Assistants with Broader Access to User's Digital World", "notes": "Reviewed against the complete retained official abstract. Experimental findings and release availability have not been independently reproduced or verified.", "published_at": "2026-05-25", "relations": [], "review_status": "curated", "scope": "core", "summary": "Claw-Anything benchmarks persistent personal assistance across activity histories, backend services and integrated GUI/CLI interaction. Its simulated event histories introduce irrelevant or conflicting context, and a separate generation pipeline supplies training environments.", "topics": ["evaluation", "interaction-history", "long-horizon", "robustness", "data-generation", "workflow"], "updated_at": "2026-05-25", "url": "https://arxiv.org/abs/2605.26086"}
{"aliases": [], "authors": ["Ruoqi Guo", "Yi Liu", "Gelei Deng", "Yiheng Xiong", "Yuekang Li", "Ying Zhang", "Leo Yu Zhang", "Lida Zhao", "Ji Jie", "Yuxiao Lu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "4f39433695b9574b275fc844ca6b10a43baf66aed57048bf841f41bf08d4ac9a", "submitted_at": "2026-05-27T08:09:53Z", "updated_at": "2026-05-27T08:09:53Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2605.28116v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2605.28116", "identifiers": {"arxiv": "2605.28116"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2605.28116"}], "name": "MIRAGE: Context-Aware Prompt Injection against Mobile GUI Agents via User-Generated Content", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-05-27", "relations": [], "review_status": "curated", "scope": "core", "summary": "MIRAGE constructs mobile screenshot injections in ordinary user-content regions through localization, contextual payload generation and realism curation. Its evaluation separates visual realism from attack effectiveness against pixel-observing GUI agents.", "topics": ["safety-security", "data-generation", "evaluation", "robustness"], "updated_at": "2026-05-27", "url": "https://arxiv.org/abs/2605.28116"}
{"aliases": [], "authors": ["Zheng Wu", "Chengcheng Han", "Zhengxi Lu", "Tianjie Ju", "Yanyu Chen", "Qi Gu", "Xunliang Cai", "Zhuosheng Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "78a5f5d7df9a8b6f1dbdb5693849c728a4b97e1f25c6e2e24d0c2418628c64e7", "submitted_at": "2026-05-27T14:29:41Z", "updated_at": "2026-05-27T14:29:41Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2605.28534v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T17:11:36Z", "role": "primary", "sha256": "235f1fff1385719d603377b680d74f6803bbe46d112dd6c008bee6b42f105429", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2605.28534", "identifiers": {"arxiv": "2605.28534"}, "kind": "paper", "last_verified_at": "2026-10-17T17:11:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2605.28534"}], "name": "GUI-CIDER: Mid-training GUI Agents via Causal Internalization and Density-aware Exemplar Reselection", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2026-05-27", "relations": [], "review_status": "curated", "scope": "core", "summary": "GUI-CIDER turns GUI trajectories into textual planning and causal interface knowledge, reselects exemplars for causal structure and low redundancy, then mid-trains GUI agents on that corpus.", "topics": ["data-generation", "planning"], "updated_at": "2026-05-27", "url": "https://arxiv.org/abs/2605.28534"}
{"aliases": [], "authors": ["Yuxuan Liu", "Xin Lai", "Junyi Li", "Pengyuan Lyu", "Jason", "Yiduo Guo", "Zhengyao Fang", "Yang Ding", "Yi Zhang", "Weinong Wang", "Huawen Shen", "Xingran Zhou", "Liang Wu", "Fei Tang", "Sunqi Fan", "Shangpin Peng", "Zheng Ruan", "Anran Zhang", "Chengquan Zhang", "Han Hu", "Benyou Wang", "Ji-Rong Wen", "Rui Yan", "Zhengyang Tang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a41cb1a93e6aa6ba8fbc7dca55415577ab5880a6592e9b0bee5b12b5a7732cbf", "submitted_at": "2026-05-28T07:14:15Z", "updated_at": "2026-09-28T15:27:12Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2605.29486v3"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:32:03Z", "role": "primary", "sha256": "2adad07a9ec7189aea25e48c76fa28206eda97f8974d5745cadda9529103bb1a", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2605.06761%2C2605.07110%2C2605.07134%2C2605.09497%2C2605.11212%2C2605.12481%2C2605.12501%2C2605.12549%2C2605.13631%2C2605.14290%2C2605.14747%2C2605.15030%2C2605.15542%2C2605.15777%2C2605.16024%2C2605.16116%2C2605.16402%2C2605.17320%2C2605.18048%2C2605.18652%2C2605.18758%2C2605.19260%2C2605.19769%2C2605.21470%2C2605.25160%2C2605.25624%2C2605.25707%2C2605.26114%2C2605.27761%2C2605.28116%2C2605.28534%2C2605.28646%2C2605.28775%2C2605.29119%2C2605.29324%2C2605.29447%2C2605.29486%2C2605.29534%2C2605.29615%2C2606.06322%2C2606.09426%2C2606.11042%2C2606.22557%2C2606.23189%2C2606.24551%2C2608.17659%2C2609.22000%2C2609.34139&max_results=48"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:32:03Z", "role": "primary", "sha256": "2adad07a9ec7189aea25e48c76fa28206eda97f8974d5745cadda9529103bb1a", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?id_list=2605.06761%2C2605.07110%2C2605.07134%2C2605.09497%2C2605.11212%2C2605.12481%2C2605.12501%2C2605.12549%2C2605.13631%2C2605.14290%2C2605.14747%2C2605.15030%2C2605.15542%2C2605.15777%2C2605.16024%2C2605.16116%2C2605.16402%2C2605.17320%2C2605.18048%2C2605.18652%2C2605.18758%2C2605.19260%2C2605.19769%2C2605.21470%2C2605.25160%2C2605.25624%2C2605.25707%2C2605.26114%2C2605.27761%2C2605.28116%2C2605.28534%2C2605.28646%2C2605.28775%2C2605.29119%2C2605.29324%2C2605.29447%2C2605.29486%2C2605.29534%2C2605.29615%2C2606.06322%2C2606.09426%2C2606.11042%2C2606.22557%2C2606.23189%2C2606.24551%2C2608.17659%2C2609.22000%2C2609.34139&max_results=48"}], "id": "arxiv:2605.29486", "identifiers": {"arxiv": "2605.29486"}, "kind": "paper", "last_verified_at": "2026-10-16T06:32:03Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2605.29486"}], "name": "PhoneWorld: From Real-App Trajectories to Dynamic and Verifiable Environments for Phone-Use Agents", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2026-05-28", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces PhoneWorld, which converts observed app pages, transitions and state changes into executable Android environments through a build-inspect-repair process. Generated tasks and programmatic verifiers support benchmarking, trajectory collection and online training.", "topics": ["data-generation", "sandbox", "reinforcement-learning", "evaluation"], "updated_at": "2026-09-28", "url": "https://arxiv.org/abs/2605.29486"}
{"aliases": [], "authors": ["Nathan Bout", "Maxime Langevin", "Ronan Riochet"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "27ffcb10d966946de9fa2353c4b96798e5f090bdfdc5610f5e8a5c6233cf70ea", "submitted_at": "2026-06-04T15:57:29Z", "updated_at": "2026-06-04T15:57:29Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.06322v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:32:03Z", "role": "primary", "sha256": "2adad07a9ec7189aea25e48c76fa28206eda97f8974d5745cadda9529103bb1a", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2605.06761%2C2605.07110%2C2605.07134%2C2605.09497%2C2605.11212%2C2605.12481%2C2605.12501%2C2605.12549%2C2605.13631%2C2605.14290%2C2605.14747%2C2605.15030%2C2605.15542%2C2605.15777%2C2605.16024%2C2605.16116%2C2605.16402%2C2605.17320%2C2605.18048%2C2605.18652%2C2605.18758%2C2605.19260%2C2605.19769%2C2605.21470%2C2605.25160%2C2605.25624%2C2605.25707%2C2605.26114%2C2605.27761%2C2605.28116%2C2605.28534%2C2605.28646%2C2605.28775%2C2605.29119%2C2605.29324%2C2605.29447%2C2605.29486%2C2605.29534%2C2605.29615%2C2606.06322%2C2606.09426%2C2606.11042%2C2606.22557%2C2606.23189%2C2606.24551%2C2608.17659%2C2609.22000%2C2609.34139&max_results=48"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:32:03Z", "role": "primary", "sha256": "2adad07a9ec7189aea25e48c76fa28206eda97f8974d5745cadda9529103bb1a", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?id_list=2605.06761%2C2605.07110%2C2605.07134%2C2605.09497%2C2605.11212%2C2605.12481%2C2605.12501%2C2605.12549%2C2605.13631%2C2605.14290%2C2605.14747%2C2605.15030%2C2605.15542%2C2605.15777%2C2605.16024%2C2605.16116%2C2605.16402%2C2605.17320%2C2605.18048%2C2605.18652%2C2605.18758%2C2605.19260%2C2605.19769%2C2605.21470%2C2605.25160%2C2605.25624%2C2605.25707%2C2605.26114%2C2605.27761%2C2605.28116%2C2605.28534%2C2605.28646%2C2605.28775%2C2605.29119%2C2605.29324%2C2605.29447%2C2605.29486%2C2605.29534%2C2605.29615%2C2606.06322%2C2606.09426%2C2606.11042%2C2606.22557%2C2606.23189%2C2606.24551%2C2608.17659%2C2609.22000%2C2609.34139&max_results=48"}], "id": "arxiv:2606.06322", "identifiers": {"arxiv": "2606.06322"}, "kind": "paper", "last_verified_at": "2026-10-16T06:32:03Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2606.06322"}], "name": "DragOn: A Benchmark and Dataset for Drag-Based GUI Interactions", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-06-04", "relations": [], "review_status": "curated", "scope": "core", "summary": "DragOn supplies drag-grounding training data and a held-out benchmark for highlighting text, selecting cells, resizing elements and moving sliders. It evaluates existing vision-language models and a model fine-tuned on the drag tasks.", "topics": ["grounding", "data-generation", "evaluation", "screen-understanding"], "updated_at": "2026-06-04", "url": "https://arxiv.org/abs/2606.06322"}
{"aliases": [], "authors": ["Longkun Hao", "Hongyu Lin", "Hao Li", "Zhuowen Liu", "Zhichao Yang", "Haojie Hao", "Dongshuo Huang", "Haitao Yang", "Hongyu Ge", "Ming jie Xie", "Yanjun Wu", "Zi Hao Yin", "Yan Bai", "Yihang Lou"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "408c8e4bc46072641819f7ab1d97ba98604c06ba7e91049b8d0e29e1a8ff2db1", "submitted_at": "2026-06-10T08:56:27Z", "updated_at": "2026-08-17T02:54:44Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2606.12485v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2606.12485", "identifiers": {"arxiv": "2606.12485"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2606.12485"}], "name": "Speculative Rollback Correction for Quality-Diverse Web Agent Imitation", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-06-10", "relations": [], "review_status": "curated", "scope": "core", "summary": "Speculative Rollback Correction lets a student web agent execute short branches before teacher review localizes harmful deviations. Verified continuations and localized corrections feed supervised training while an archive retains diverse successful solutions.", "topics": ["imitation-learning", "data-generation", "robustness"], "updated_at": "2026-08-17", "url": "https://arxiv.org/abs/2606.12485"}
{"aliases": [], "authors": ["Xinyu Qiu", "Yunzhu Zhang", "Heng Jia", "Shuheng Shen", "Changhua Meng", "Linchao Zhu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "1a5e949a760f236c08ae3cebc32d8a52220b1101f5324770d40bd7d054351f6a", "submitted_at": "2026-06-12T15:58:18Z", "updated_at": "2026-06-12T15:58:18Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.14579v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2606.14579", "identifiers": {"arxiv": "2606.14579"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2606.14579"}], "name": "VISTA: View-Consistent Self-Verified Training for GUI Grounding", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-06-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "VISTA forms GRPO comparison groups from target-preserving crops of the same GUI instance with exactly remapped target boxes. A conditional, advantage-weighted cross-view anchor supplies additional supervision without entering the group baseline, and evaluation includes view-consistency robustness.", "topics": ["grounding", "reinforcement-learning", "robustness", "data-generation"], "updated_at": "2026-06-12", "url": "https://arxiv.org/abs/2606.14579"}
{"aliases": [], "authors": ["Jaehun Jung", "Ximing Lu", "Brandon Cui", "Muhammad Khalifa", "Shaokun Zhang", "Hao Zhang", "Jin Xu", "Amala Sanjay Deshmukh", "Karan Sapra", "Andrew Tao", "Yejin Choi", "Jan Kautz", "Mingjie Liu", "Yi Dong"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d7e6e0b66f2923cb66ab7aa404bedf9f4e1c73ae00df2e9fbfa7a5c3795cac4c", "submitted_at": "2026-06-15T22:04:11Z", "updated_at": "2026-06-15T22:04:11Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.17321v1"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:05Z", "role": "primary", "sha256": "2e5722527f8da4b2cf06372d1e2daccd33ef7fdc7151704b411a0bddd734beae", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2606.10423%2C2606.10522%2C2606.10742%2C2606.11078%2C2606.11520%2C2606.12485%2C2606.12666%2C2606.13239%2C2606.13385%2C2606.14106%2C2606.14832%2C2606.15017%2C2606.15034%2C2606.16659%2C2606.16748%2C2606.16802%2C2606.17321%2C2606.17645%2C2606.17929%2C2606.18448%2C2606.18671%2C2606.18890%2C2606.19388%2C2606.19926%2C2606.19930%2C2606.20363%2C2606.20636%2C2606.20717%2C2606.20724%2C2606.20785%2C2606.21654%2C2606.22864%2C2606.22948%2C2606.23449%2C2606.24515%2C2606.24525%2C2606.25705%2C2606.25760%2C2606.27330%2C2606.27595%2C2606.29445%2C2606.29537%2C2606.29705%2C2606.30697%2C2606.31154%2C2606.31270%2C2606.31410%2C2606.31612%2C2607.00502%2C2607.04425&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:05Z", "role": "primary", "sha256": "2e5722527f8da4b2cf06372d1e2daccd33ef7fdc7151704b411a0bddd734beae", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?id_list=2606.10423%2C2606.10522%2C2606.10742%2C2606.11078%2C2606.11520%2C2606.12485%2C2606.12666%2C2606.13239%2C2606.13385%2C2606.14106%2C2606.14832%2C2606.15017%2C2606.15034%2C2606.16659%2C2606.16748%2C2606.16802%2C2606.17321%2C2606.17645%2C2606.17929%2C2606.18448%2C2606.18671%2C2606.18890%2C2606.19388%2C2606.19926%2C2606.19930%2C2606.20363%2C2606.20636%2C2606.20717%2C2606.20724%2C2606.20785%2C2606.21654%2C2606.22864%2C2606.22948%2C2606.23449%2C2606.24515%2C2606.24525%2C2606.25705%2C2606.25760%2C2606.27330%2C2606.27595%2C2606.29445%2C2606.29537%2C2606.29705%2C2606.30697%2C2606.31154%2C2606.31270%2C2606.31410%2C2606.31612%2C2607.00502%2C2607.04425&max_results=50"}], "id": "arxiv:2606.17321", "identifiers": {"arxiv": "2606.17321"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:05Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2606.17321"}], "name": "ProCUA-SFT Technical Report", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-06-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "ProCUA-SFT generates tasks and synthetic trajectories on live desktops seeded with document content, checking task preconditions before rollout. The resulting step-prefix examples are formatted to match inference context and used for supervised GUI-agent fine-tuning.", "topics": ["data-generation", "imitation-learning", "evaluation"], "updated_at": "2026-06-15", "url": "https://arxiv.org/abs/2606.17321"}
{"aliases": [], "authors": ["Zhimin Fan", "Hongwei Yu", "Yeqing Shen", "Haolong Yan", "Guozhen Peng", "Tianhao Peng", "Yudong Zhang", "Xiaowen Zhang", "Kaijun Tan", "Zheng Ge", "Xiangyu Zhang", "Daxin Jiang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "1929dfb5074e159d83006c7bbb470237a0c557e0723ac631116aa5e0369f9247", "submitted_at": "2026-06-17T10:07:51Z", "updated_at": "2026-06-17T10:07:51Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.18890v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2606.18890", "identifiers": {"arxiv": "2606.18890"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2606.18890"}], "name": "Skill-Guided Continuation Distillation for GUI Agents", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-06-17", "relations": [], "review_status": "curated", "scope": "core", "summary": "Skill-Guided Continuation Distillation first lets a policy reach its own off-trajectory states, then uses skill guidance to collect successful continuations. Mixing those continuations with expert traces supplies supervision for recovery states absent from demonstrations.", "topics": ["imitation-learning", "data-generation", "robustness", "planning"], "updated_at": "2026-06-17", "url": "https://arxiv.org/abs/2606.18890"}
{"aliases": [], "authors": ["Guangyi Liu", "Gao Wu", "Congxiao Liu", "Pengxiang Zhao", "Liang Liu", "Mading Li", "Qi Zhang", "Mengyan Wang", "Liang Guo", "Yong Liu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "cbbb1bd57057e465ce9d991d3f6e2bb52112fc133a5846fc8cb6ce51eddc2121", "submitted_at": "2026-06-18T08:26:09Z", "updated_at": "2026-06-18T08:26:09Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.19926v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2606.19926", "identifiers": {"arxiv": "2606.19926"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2606.19926"}], "name": "MemGUI-Agent: An End-to-End Long-Horizon Mobile GUI Agent with Proactive Context Management", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-06-18", "relations": [], "review_status": "curated", "scope": "core", "summary": "MemGUI-Agent makes context management an action chosen by the same policy that controls the mobile UI. Structured history, UI-state and recent-step fields are trained from annotated trajectories to retain cross-app facts while limiting context growth.", "topics": ["memory", "interaction-history", "long-horizon", "imitation-learning", "data-generation"], "updated_at": "2026-06-18", "url": "https://arxiv.org/abs/2606.19926"}
{"aliases": [], "authors": ["Guangyi Liu", "Pengxiang Zhao", "Gao Wu", "Yiwen Yin", "Mading Li", "Liang Liu", "Congxiao Liu", "Zhang Qi", "Mengyan Wang", "Liang Guo", "Jiangning Zhang", "Yong Liu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "cab809618ea4e218da26b4ae11bf937859c3a4bc240c91dd27b3a0d333737069", "submitted_at": "2026-06-18T08:29:33Z", "updated_at": "2026-08-07T03:54:59Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2606.19930v2"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2606.19930", "identifiers": {"arxiv": "2606.19930"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2606.19930"}], "name": "MobileForge: Annotation-Free Adaptation for Mobile GUI Agents with Hierarchical Feedback-Guided Policy Optimization", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-06-18", "relations": [], "review_status": "curated", "scope": "core", "summary": "MobileForge couples real-app exploration and generated tasks with hierarchical feedback for mobile-agent adaptation. Its optimizer turns trajectory outcomes, step feedback and corrective hints into contextualized GRPO updates, with evaluation on AndroidWorld and a GUI-only MobileWorld split.", "topics": ["mobile-navigation", "reinforcement-learning", "data-generation", "evaluation"], "updated_at": "2026-08-07", "url": "https://arxiv.org/abs/2606.19930"}
{"aliases": [], "authors": ["Yuexing Hao", "Xiaomin Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "b370491b37d5bf25c624f2bac654e438459c8084b5da292fb18a0bf9142adaa2", "submitted_at": "2026-06-18T15:25:42Z", "updated_at": "2026-06-18T15:25:42Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.20363v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2606.20363", "identifiers": {"arxiv": "2606.20363"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2606.20363"}], "name": "Automating SKILL.md Generation for Computer-Using Agents via Interaction Trajectory Mining", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-06-18", "relations": [], "review_status": "curated", "scope": "core", "summary": "Segments and clusters GUI trajectories into inspectable skills, then trains a skill-aware policy from those annotations. Its diagnostic results distinguish readable source-domain clusters from reliable transfer and report limited or absent downstream gains.", "topics": ["memory", "data-generation", "reinforcement-learning", "evaluation"], "updated_at": "2026-06-18", "url": "https://arxiv.org/abs/2606.20363"}
{"aliases": [], "authors": ["Ahmed Awadallah", "Sahil Gupta", "Yash Lara", "Yadong Lu", "Hussein Mozannar", "Akshay Nambi", "Zach Nussbaum", "Yash Pandya", "Aravind Rajeswaran", "Corby Rosset", "Alexey Taymanov", "Luiz do Valle", "Vibhav Vineet", "Spencer Whitehead", "Andrew Zhao"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "414eced85da9f76828e4398b977f9b8dac77236c97d1594108c8f5dfc3be7cbb", "submitted_at": "2026-06-18T17:53:03Z", "updated_at": "2026-07-22T17:47:01Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2606.20785v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2606.20785", "identifiers": {"arxiv": "2606.20785"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2606.20785"}], "name": "Fara-1.5: Scalable Learning Environments for Computer Use Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-06-18", "relations": [], "review_status": "curated", "scope": "core", "summary": "Builds browser-agent training trajectories with live and synthetic environments, solvers, user simulation, and complementary correctness and efficiency verifiers.", "topics": ["data-generation", "web-navigation", "evaluation"], "updated_at": "2026-07-22", "url": "https://arxiv.org/abs/2606.20785"}
{"aliases": [], "authors": ["Yincheng Zhou", "Athena Zhuoming Zhong", "Shijie Zhang", "Kevin Zhang", "Teresa Xiaotao Shang", "Shanghang Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "fbd8892e06a5e847c16aa3101b98b3d669a783b84ee8e3d4d4d810754914028b", "submitted_at": "2026-06-22T07:28:55Z", "updated_at": "2026-06-22T07:28:55Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.22948v1"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2606.22948", "identifiers": {"arxiv": "2606.22948"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2606.22948"}], "name": "ENVS: Environment-Native Verified Search for Long-Horizon GUI Agents", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-06-22", "relations": [], "review_status": "curated", "scope": "core", "summary": "ENVS branches over GUI actions in live desktop VMs, verifies successful leaves, and converts them into balanced step-level training supervision. It also introduces OSWorld-Noisy to test recovery from dynamic interruptions while retaining the underlying tasks.", "topics": ["data-generation", "reinforcement-learning", "long-horizon", "robustness", "evaluation"], "updated_at": "2026-06-22", "url": "https://arxiv.org/abs/2606.22948"}
{"aliases": [], "authors": ["Jiachen Qian"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a3d0341f41a8ecba57fb7ed9e4b266bb4a176639c742be2b4fc98555754f5744", "submitted_at": "2026-06-23T12:57:02Z", "updated_at": "2026-06-23T12:57:02Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.24525v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2606.24525", "identifiers": {"arxiv": "2606.24525"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2606.24525"}], "name": "VisCritic: Visual State Comparison as Process Reward for GUI Agents", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-06-23", "relations": [], "review_status": "curated", "scope": "core", "summary": "VisCritic compares before/after screenshots through change-aware visual features and an action-conditioned critic. Weakly supervised samples from existing traces train judgments of action success, progress and error type for GUI-agent verification.", "topics": ["reward-models", "state-tracking", "data-generation", "evaluation"], "updated_at": "2026-06-23", "url": "https://arxiv.org/abs/2606.24525"}
{"aliases": [], "authors": ["Aradhana Nayak", "Mussadiq Nazeer", "Wang Peng", "Feng Liu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "3c5a2d0e461b011e2d5b9dbd73012d007261d8df37bd1d687fa3bd86dc6f2246", "submitted_at": "2026-06-24T11:18:01Z", "updated_at": "2026-06-24T11:18:01Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.25705v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2606.25705", "identifiers": {"arxiv": "2606.25705"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2606.25705"}], "name": "GUI agent: Guided Exploration of User-Sensitive Screens", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-06-24", "relations": [], "review_status": "curated", "scope": "core", "summary": "This short paper develops an explorer agent that expands from a demonstrated GUI task to discover requests likely to reach user-sensitive screens. The proposed exploration supports identifying situations where task execution should be handed back to the user.", "topics": ["safety-security", "human-in-the-loop", "data-generation"], "updated_at": "2026-06-24", "url": "https://arxiv.org/abs/2606.25705"}
{"aliases": [], "authors": ["Chengheng Li-Chen", "Zhiqian Zhou", "Hao Chen", "Nicolas Chauvin"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "7a8efe4b9476d7f55ca4da17621ece653be2e33bf02be0183a339a12f5fac3ca", "submitted_at": "2026-06-24T15:35:29Z", "updated_at": "2026-06-24T15:35:29Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.25964v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2606.25964", "identifiers": {"arxiv": "2606.25964"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2606.25964"}], "name": "WinDOM: Self-Family Distillation for Small-Model GUI Grounding", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-06-24", "relations": [], "review_status": "curated", "scope": "core", "summary": "WinDOM obtains bounding-box supervision from a Windows 11 web reimplementation driven through Playwright, and studies rejection-sampling distillation for small GUI grounders. It treats the depth of the supervised cold start as a GRPO choice, comparing same-size EMA and larger same-family teachers.", "topics": ["grounding", "data-generation", "imitation-learning", "reinforcement-learning"], "updated_at": "2026-06-24", "url": "https://arxiv.org/abs/2606.25964"}
{"aliases": [], "authors": ["Tianyi Men", "Zhuoran Jin", "Pengfei Cao", "Yubo Chen", "Kang Liu", "Jun Zhao"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "42166ac3980296a53b56bc4a076634beec36ba885b3260b5d36115b5a5261cd0", "submitted_at": "2026-06-25T17:44:48Z", "updated_at": "2026-06-25T17:44:48Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.27330v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2606.27330", "identifiers": {"arxiv": "2606.27330"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2606.27330"}], "name": "Empowering GUI Agents via Autonomous Experience Exploration and Hindsight Experience Utilization for Task Planning", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-06-25", "relations": [], "review_status": "curated", "scope": "core", "summary": "PEEU explores web environments and turns hindsight experience into aligned high-level planning supervision. A hierarchical task-decomposition analysis examines why atomic-skill competence alone does not ensure planning generalization across websites.", "topics": ["planning", "data-generation", "imitation-learning", "web-navigation"], "updated_at": "2026-06-25", "url": "https://arxiv.org/abs/2606.27330"}
{"aliases": [], "authors": ["Sunqi Fan", "Lingshan Chen", "Runqi Yin", "Qingle Liu", "Yongming Rao", "Meng-Hao Guo", "Shi-Min Hu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "7beb70ac91463b77a8babaf2b077c46c5072e8f86d22f4d9afc9c1e77680f309", "submitted_at": "2026-06-29T02:16:21Z", "updated_at": "2026-06-29T02:16:21Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.29705v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2606.29705", "identifiers": {"arxiv": "2606.29705"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2606.29705"}], "name": "GUICrafter: Weakly-Supervised GUI Agent Leveraging Massive Unannotated Screenshots", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-06-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "GUICrafter first learns grounding from large collections of unannotated screenshots and webpages, then uses a smaller high-quality dataset for reinforcement-learning calibration. The curriculum targets reduced dependence on manually annotated GUI training data.", "topics": ["grounding", "data-generation", "reinforcement-learning", "screen-understanding"], "updated_at": "2026-06-29", "url": "https://arxiv.org/abs/2606.29705"}
{"aliases": [], "authors": ["Niu Lian", "Tongbo Chen", "Zhehao Yu", "Chengzhen Duan", "Fazhan Liu", "Hui Liu", "Pei Fu", "Jian Luan", "Heng Qu", "Shu-Tao Xia", "Jinpeng Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "32cb3020ccf25c31b805dc44218f5f9496cbc310e1df2906d8b15e5b60c31278", "submitted_at": "2026-07-05T17:37:12Z", "updated_at": "2026-08-10T13:11:37Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2607.04425v2"}}, "environments": ["desktop", "cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2607.04425", "identifiers": {"arxiv": "2607.04425"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2607.04425"}], "name": "UI-MOPD: Multi-Platform On-Policy Distillation for Unified GUI Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-07-05", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces Uni-GUI, a collection of executable desktop and mobile trajectories, and UI-MOPD, which trains a shared policy from its own rollouts using platform-specific teachers. The distillation procedure routes each visited state to the appropriate teacher to preserve distinct interaction conventions.", "topics": ["data-generation", "imitation-learning", "action-prediction"], "updated_at": "2026-08-10", "url": "https://arxiv.org/abs/2607.04425"}
{"aliases": [], "authors": ["Mianqiu Huang", "Taofeng Xue", "Chong Peng", "Jinrui Ding", "Jie Yang", "Sicheng Fan", "Jiale Hong", "Yufei Gao", "Xiaocheng Zhang", "Linsen Guo", "Xin Yang", "Dengchang Zhao", "Yuchen Xie", "Peng Pei", "Xunliang Xie", "Xipeng Qiu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "53c254c53a834d4f28bb4b65317121f5c46f5dd6985d03a30f1cc1f0fa78707d", "submitted_at": "2026-07-07T16:36:25Z", "updated_at": "2026-09-04T15:47:10Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2607.09773v2"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2607.09773", "identifiers": {"arxiv": "2607.09773"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2607.09773"}], "name": "EvoCUA-1.5: Online Reinforcement Learning for Multi-turn Computer-Use Agents", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-07-07", "relations": [], "review_status": "curated", "scope": "core", "summary": "EvoCUA-1.5 trains multi-turn desktop agents through online reinforcement learning in executable sandboxes. It combines step-level policy optimization, task filtering and calibration, an adaptive curriculum, and asynchronous rollout handling with staleness control.", "topics": ["reinforcement-learning", "long-horizon", "sandbox", "data-generation"], "updated_at": "2026-09-04", "url": "https://arxiv.org/abs/2607.09773"}
{"aliases": [], "authors": ["Bowen Lv", "Xiao Liu", "Yanyu Ren", "Hanyu Lai", "Bohao Jing", "Hanchen Zhang", "Yanxiao Zhao", "Shuntian Yao", "Jie Tang", "Yuxiao Dong"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "554844f23cddbdaadc3c52e93d07d06c756faadbd0136ffef0984b25423f84f8", "submitted_at": "2026-07-13T07:32:35Z", "updated_at": "2026-07-13T07:32:35Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2607.11185v1"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2607.11185", "identifiers": {"arxiv": "2607.11185"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2607.11185"}], "name": "SCALECUA: Scaling Computer Use Agents with Verifiable Task Synthesis and Efficient Online RL", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-07-13", "relations": [], "review_status": "curated", "scope": "core", "summary": "Scales computer-agent reinforcement learning through verifiable task generation, capability-aware rollout allocation, and segmented visual context.", "topics": ["data-generation", "reinforcement-learning", "efficiency"], "updated_at": "2026-07-13", "url": "https://arxiv.org/abs/2607.11185"}
{"aliases": [], "authors": ["Hy Vision Team", "Huawen Shen", "Zhengyang Tang", "Shangpin Peng", "Liang Wu", "Anran Zhang", "Weinong Wang", "Yiduo Guo", "Chenxin Li", "Zhengyao Fang", "Yang Ding", "Junyi Li", "Fei Tang", "Zheng Ruan", "Yi Zhang", "Xingran Zhou", "Dingchen Yang", "Sunqi Fan", "Zhiyi Wan", "Han Hu", "Xin Lai", "Pengyuan Lyu", "Chengquan Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "151d4882fecc835dfffb9651bd860ebe8c85919a63ff5730e191f9146fbba272", "submitted_at": "2026-07-16T04:12:42Z", "updated_at": "2026-07-16T04:12:42Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2607.14548v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2607.14548", "identifiers": {"arxiv": "2607.14548"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2607.14548"}], "name": "HyMobileAgent: Data-Environment Co-Scaling for Efficient GUI Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-07-16", "relations": [], "review_status": "curated", "scope": "core", "summary": "Combines GUI perception data, instructional videos, real and simulated devices, and planning with reflection to develop a mobile agent.", "topics": ["mobile-navigation", "data-generation", "planning"], "updated_at": "2026-07-16", "url": "https://arxiv.org/abs/2607.14548"}
{"aliases": [], "authors": ["Zhuohang Fan", "Beichen Zhang", "Yuanfa Li", "Changqiao Wu", "Wei Liu", "Jian Luan", "Weigang Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "b85b453a688543477d31a940f9d1da2857b2cb7a8cb9eea91a39a3a2b8a1e1c7", "submitted_at": "2026-07-20T15:16:59Z", "updated_at": "2026-08-18T02:32:56Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2607.18046v2"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2607.18046", "identifiers": {"arxiv": "2607.18046"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2607.18046"}], "name": "SEE: Structure-aware Exploring & Exploiting for Long-horizon GUI Agent Trajectory Synthesis", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-07-20", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces SEE, which first explores mobile interfaces to build a screen-and-element transition graph, then composes task trajectories through graph planning and controlled sampling. It studies coverage, loop avoidance, and transfer after training agents on the generated sequences.", "topics": ["data-generation", "long-horizon", "mobile-navigation"], "updated_at": "2026-08-18", "url": "https://arxiv.org/abs/2607.18046"}
{"aliases": [], "authors": ["Yang Wan", "Zhenhao Zhang", "Jierui Wang", "Xiangxin Zhou", "Linchao Zhu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2de35aa3c7f4457d43d485eb136d8b380ea61870cc7253e807880ae731c51316", "submitted_at": "2026-07-25T16:00:45Z", "updated_at": "2026-09-26T12:15:44Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2607.23263v2"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2607.23263", "identifiers": {"arxiv": "2607.23263"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2607.23263"}], "name": "SeekJudge: A Practical Reward Framework for Reinforcement Learning in Computer-Use Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-07-25", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces SeekJudge, which iteratively searches and analyzes GUI trajectories to assess instruction fulfillment. The work combines dense judgment-data synthesis, a step-label benchmark, and reward-server scheduling to study accurate and practical reward feedback for agent training.", "topics": ["reward-models", "evaluation", "data-generation"], "updated_at": "2026-09-26", "url": "https://arxiv.org/abs/2607.23263"}
{"aliases": [], "authors": ["Xiangyan Liu", "Kaixin Li", "Haonan Wang", "Biao Wu", "Meng Fang", "Longxu Dou", "Chao Du", "Michael Qizhe Shieh", "Tianyu Pang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "4651964cbe45e4d740ed8f789753f82ae808e442d82c6598e7b3c997c09453c3", "submitted_at": "2026-07-27T07:54:08Z", "updated_at": "2026-07-27T07:54:08Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2607.24112v1"}}, "environments": ["desktop", "android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2607.24112", "identifiers": {"arxiv": "2607.24112"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2607.24112"}], "name": "Scaling GUI Agents with Visual State Transitions", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-07-27", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces State Transition Pretraining, which jointly learns to infer actions from GUI changes and predict future states from current screens and actions. Subsequent instruction-trajectory fine-tuning tests whether these representations transfer to desktop and mobile agent tasks.", "topics": ["world-models", "action-prediction", "data-generation"], "updated_at": "2026-07-27", "url": "https://arxiv.org/abs/2607.24112"}
{"aliases": [], "authors": ["Yash Pandya", "Sahil Gupta", "Sarthak Harne", "Archana Yadav", "Kavyansh Chourasia", "Hussein Mozannar", "Vibhav Vineet", "Sara Abdali", "Corby Rosset", "Yash Lara", "Ahmed Awadallah", "Ece Kamar", "Akshay Nambi"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6f28e1ee8e101000f3cfe9215555251656fcb40b921b5c41c895fbc958f867e3", "submitted_at": "2026-07-30T11:48:22Z", "updated_at": "2026-07-30T11:48:22Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2607.28074v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2607.28074", "identifiers": {"arxiv": "2607.28074"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2607.28074"}], "name": "Echoverse: Deep, Evolving Environments for Training Computer-Use Agents at Scale", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-07-30", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces Echoverse, which compiles specifications into stateful applications with database-grounded graders and uses rollouts to improve both the environments and agent policies. It examines behavioral depth, targeted interface practice, and environment repair as sources of training value.", "topics": ["sandbox", "data-generation", "reinforcement-learning"], "updated_at": "2026-07-30", "url": "https://arxiv.org/abs/2607.28074"}
{"aliases": [], "authors": ["Qiushi Sun", "Kanzhi Cheng", "Yian Wang", "Bowen Yang", "Hang Yan", "Liheng Chen", "Fangzhi Xu", "Zichen Ding", "Nuo Chen", "Jialin Cao", "Xingdong Gong", "Zehao Li", "Kaiming Jin", "Xinfeng Yuan", "Zhoumianze Liu", "Jingyang Gong", "Zhangyue Yin", "Jiahui Gao", "Zhiyong Wu", "Tianbao Xie", "Jianbing Zhang", "Ben Kao", "Lingpeng Kong"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e3e82df7abffbd0c8f7e969ed8edb77423263a2b1d826a3e72529b0a51903d58", "submitted_at": "2026-07-30T17:57:41Z", "updated_at": "2026-08-06T17:55:59Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2607.28609v2"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2607.28609", "identifiers": {"arxiv": "2607.28609"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2607.28609"}], "name": "OSReward: Instituting Standardized Evaluation for Cross-Platform Computer-Use Reward Models", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-07-30", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces OSReward and its harder and multidimensional variants for assessing judgments of computer-use trajectories. It also releases a reasoning-annotated judgment corpus and trains reward models, examining reliability, leniency bias, and evaluation cost.", "topics": ["reward-models", "evaluation", "data-generation"], "updated_at": "2026-08-06", "url": "https://arxiv.org/abs/2607.28609"}
{"aliases": [], "authors": ["Dunjie Lu", "Shuai Bai", "Tianyi Bai", "Sicheng Fan", "Chang Gao", "Jian Guan", "Feng Hu", "Mianqiu Huang", "Xingyang Huang", "Yizhen Jiang", "Yuheng Jing", "Dehui Kong", "Ning Li", "Dayiheng Liu", "Shixuan Liu", "Zheng Liu", "Que Shen", "Bowen Wang", "Junli Wang", "Chencan Wu", "Rui Xie", "Tianbao Xie", "Zhihui Xie", "Haiyang Xu", "An Yang", "Tao Yu", "Wenzhen Yuan", "Xi Zhang", "Zhenru Zhang", "Mingkang Zhu", "Zhaoqing Zhu", "Yizhong Cao", "Kai Dang", "Binyuan Hui", "Kaixin Li", "Junyang Lin", "Haiquan Wang", "Zekun Wang", "Yiheng Xu", "Fan Yan", "Mengqi Yuan", "Danyang Zhang", "Jiajun Zhang", "Zhipeng Zhang", "Fan Zhou", "Fan Zhou"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "26a61ccfdddb5cdc2da2b14d29ae8d4098544fea4e9458126f2a49c28418c4f0", "submitted_at": "2026-08-03T15:04:20Z", "updated_at": "2026-08-03T15:04:20Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2608.02352v1"}}, "environments": ["desktop", "web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2608.02352", "identifiers": {"arxiv": "2608.02352"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2608.02352"}], "name": "Qwen-CUA: Native Computer Use for (almost) Everything", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-08-03", "relations": [], "review_status": "curated", "scope": "core", "summary": "Qwen-CUA operates through screenshots and keyboard/mouse events while compressing older visual history into reusable context blocks. Its report combines verifiable task generation, trajectory-based training and separate browser and Bash-augmented studies.", "topics": ["reinforcement-learning", "imitation-learning", "data-generation", "interaction-history", "long-horizon", "evaluation"], "updated_at": "2026-08-03", "url": "https://arxiv.org/abs/2608.02352"}
{"aliases": [], "authors": ["Weikai Xu", "Yunren Feng", "Haoxiang Lei", "Kun Huang", "Yuxuan Liu", "Kang Zhao", "Xiaolin Hu", "Shuo Shang", "Bo An"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "ce37dd8217afbaa07ddea25c0daf78cdbfb449851ca6ab17b7f3ecacd451eb2f", "submitted_at": "2026-08-06T11:15:41Z", "updated_at": "2026-08-06T11:15:41Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2608.05891v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2608.05891", "identifiers": {"arxiv": "2608.05891"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2608.05891"}], "name": "AppDeltaWorld: Transition-Grounded Delta Code World Model for Mobile GUI Agents", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-08-06", "relations": [], "review_status": "curated", "scope": "core", "summary": "AppDeltaWorld predicts executable code updates for the next mobile UI state, using transition-constrained retrieval and rendered visual assets. It evaluates reconstruction fidelity separately from policy training and adaptation using generated environments.", "topics": ["world-models", "data-generation", "imitation-learning", "reinforcement-learning", "mobile-navigation"], "updated_at": "2026-08-06", "url": "https://arxiv.org/abs/2608.05891"}
{"aliases": [], "authors": ["Ruitao Wang", "Yuwen Hao", "Menglin Yang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "bf543ef0b77eb6fe430c8cef5382932f13e4e4136aaba9c14c44b5a591d6612a", "submitted_at": "2026-08-12T12:03:20Z", "updated_at": "2026-08-12T12:03:20Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2608.12429v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2608.12429", "identifiers": {"arxiv": "2608.12429"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2608.12429"}], "name": "SynWeaver: Website-Prior Task and Trajectory Co-Synthesis for Web Agents", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-08-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "SynWeaver explores a website to build a map of page states and executable interactions, then learns website priors for task proposals. Joint task/trajectory synthesis revises inconsistent pairs and verifies or repairs the resulting supervision.", "topics": ["data-generation", "web-navigation", "state-tracking", "robustness"], "updated_at": "2026-08-12", "url": "https://arxiv.org/abs/2608.12429"}
{"aliases": [], "authors": ["Zihan Ding", "Longxu Dou", "Qi Gao", "Xiangwu Guo", "Shengchao Hu", "Zilong Huang", "Zihang Jiang", "Lei Ke", "Mengcheng Lan", "Weixian Lei", "Hanxuan Li", "Honglin Li", "Xiyun Li", "Zaitang Li", "Leowei Liang", "Xin Luo", "Haozhe Ma", "Jiayi Mao", "Zhoujie Pan", "Can Qin", "Tianyuan Qu", "Weiqi Wang", "Wenkai Wang", "Yonglin Wang", "Yuxin Wang", "Chenxu Wu", "Yingchen Yu", "Chenyu Zhang", "Yuhao Zheng"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "efe635792c696bca73c9a436be67d8528582fb239d4d4f8d3de531219b1f4f8e", "submitted_at": "2026-08-16T20:59:59Z", "updated_at": "2026-08-16T20:59:59Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2608.15930v1"}}, "environments": ["desktop", "windows"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2608.15930", "identifiers": {"arxiv": "2608.15930"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2608.15930"}], "name": "UI-Mate: Advancing Open-Weight Foundation GUI Agents with In-Context Demonstrations", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-08-16", "relations": [], "review_status": "curated", "scope": "core", "summary": "Combines environment-grounded training with demonstration-conditioned workflows, and introduces OSWorkerBench for extended office-task evaluation.", "topics": ["workflow", "data-generation", "long-horizon"], "updated_at": "2026-08-16", "url": "https://arxiv.org/abs/2608.15930"}
{"aliases": [], "authors": ["Chenghao Zhang", "Yuxi Cheng", "Saisai Hu", "Canran Xiao", "Yikai Mao", "Dan Roth"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "7d862afa969c0b4d1c6edc8d126017749844158f786b29c3b0710afecdec2008", "submitted_at": "2026-08-22T10:18:45Z", "updated_at": "2026-09-28T15:54:31Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2608.21898v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:13Z", "role": "primary", "sha256": "ca895b2f0dcd74816fb5e775a7161a0932ff7120a72477791f5e4c0bb1ead88e", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2608.17319%2C2608.18307%2C2608.20319%2C2608.20707%2C2608.20797%2C2608.21794%2C2608.21830%2C2608.21832%2C2608.21898%2C2608.22077%2C2608.22533%2C2608.22577%2C2608.22847%2C2608.24015%2C2608.24017%2C2608.24099%2C2608.24174%2C2608.24680%2C2608.24848%2C2608.25777%2C2608.26200%2C2608.26204%2C2608.27477%2C2608.27508%2C2608.27808%2C2608.27866%2C2608.29305%2C2608.30038%2C2608.30207%2C2609.00028%2C2609.00048%2C2609.00524%2C2609.02057%2C2609.02149%2C2609.02154%2C2609.02309%2C2609.02885%2C2609.03438%2C2609.04850%2C2609.04869%2C2609.05374%2C2609.05511%2C2609.05513%2C2609.05535%2C2609.06530%2C2609.06792%2C2609.07603%2C2609.07712%2C2609.09212%2C2609.10031&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:13Z", "role": "primary", "sha256": "ca895b2f0dcd74816fb5e775a7161a0932ff7120a72477791f5e4c0bb1ead88e", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2608.17319%2C2608.18307%2C2608.20319%2C2608.20707%2C2608.20797%2C2608.21794%2C2608.21830%2C2608.21832%2C2608.21898%2C2608.22077%2C2608.22533%2C2608.22577%2C2608.22847%2C2608.24015%2C2608.24017%2C2608.24099%2C2608.24174%2C2608.24680%2C2608.24848%2C2608.25777%2C2608.26200%2C2608.26204%2C2608.27477%2C2608.27508%2C2608.27808%2C2608.27866%2C2608.29305%2C2608.30038%2C2608.30207%2C2609.00028%2C2609.00048%2C2609.00524%2C2609.02057%2C2609.02149%2C2609.02154%2C2609.02309%2C2609.02885%2C2609.03438%2C2609.04850%2C2609.04869%2C2609.05374%2C2609.05511%2C2609.05513%2C2609.05535%2C2609.06530%2C2609.06792%2C2609.07603%2C2609.07712%2C2609.09212%2C2609.10031&max_results=50"}], "id": "arxiv:2608.21898", "identifiers": {"arxiv": "2608.21898"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:13Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2608.21898"}], "name": "Training Needs Trustworthy Worlds: Verified Synthetic Web Environments for Agent Learning", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-08-22", "relations": [], "review_status": "curated", "scope": "core", "summary": "Builds synthetic websites with verified navigation, backend state, and feasible tasks to provide reliable supervision for web-agent training.", "topics": ["data-generation", "sandbox", "reinforcement-learning"], "updated_at": "2026-09-28", "url": "https://arxiv.org/abs/2608.21898"}
{"aliases": [], "authors": ["Long Zhang", "Yuhan Chen", "Chaoran Zhang", "Wanxia Cao", "Kun Huang", "Pengzhi Gao", "Wei Liu", "Jian Luan", "Chenliang Li", "Lixin Zou"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "4c1198ce65d5a2fcffc51b37bc4553b35b9e0d809757dc4181747c57c7ae443d", "submitted_at": "2026-08-24T06:27:47Z", "updated_at": "2026-08-24T06:27:47Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2608.22847v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2608.22847", "identifiers": {"arxiv": "2608.22847"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2608.22847"}], "name": "GSAR: Goal-State-Anchor Rewards for Mobile GUI Agents with Self-Evolving Data Synthesis", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-08-24", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces GSAR, which synthesizes GUI tasks and successful goal states, then annotates task-relevant interface elements as reward anchors. These anchors support offline trajectory verification and online policy training.", "topics": ["data-generation", "reward-models", "reinforcement-learning"], "updated_at": "2026-08-24", "url": "https://arxiv.org/abs/2608.22847"}
{"aliases": [], "authors": ["Fei Tang", "Huawen Shen", "Zhiqiong Lu", "Zhengxi Lu", "Pengyuan Lyu", "Chengquan Zhang", "Weiming Lu", "Jun Xiao", "Yueting Zhuang", "Yongliang Shen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "0d2180e9b1b0c3ae20e32b0b597dc209e363551aaa5031c98504ef08314979a6", "submitted_at": "2026-08-25T17:35:42Z", "updated_at": "2026-08-25T17:35:42Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2608.24848v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:13Z", "role": "primary", "sha256": "ca895b2f0dcd74816fb5e775a7161a0932ff7120a72477791f5e4c0bb1ead88e", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2608.17319%2C2608.18307%2C2608.20319%2C2608.20707%2C2608.20797%2C2608.21794%2C2608.21830%2C2608.21832%2C2608.21898%2C2608.22077%2C2608.22533%2C2608.22577%2C2608.22847%2C2608.24015%2C2608.24017%2C2608.24099%2C2608.24174%2C2608.24680%2C2608.24848%2C2608.25777%2C2608.26200%2C2608.26204%2C2608.27477%2C2608.27508%2C2608.27808%2C2608.27866%2C2608.29305%2C2608.30038%2C2608.30207%2C2609.00028%2C2609.00048%2C2609.00524%2C2609.02057%2C2609.02149%2C2609.02154%2C2609.02309%2C2609.02885%2C2609.03438%2C2609.04850%2C2609.04869%2C2609.05374%2C2609.05511%2C2609.05513%2C2609.05535%2C2609.06530%2C2609.06792%2C2609.07603%2C2609.07712%2C2609.09212%2C2609.10031&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:13Z", "role": "primary", "sha256": "ca895b2f0dcd74816fb5e775a7161a0932ff7120a72477791f5e4c0bb1ead88e", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2608.17319%2C2608.18307%2C2608.20319%2C2608.20707%2C2608.20797%2C2608.21794%2C2608.21830%2C2608.21832%2C2608.21898%2C2608.22077%2C2608.22533%2C2608.22577%2C2608.22847%2C2608.24015%2C2608.24017%2C2608.24099%2C2608.24174%2C2608.24680%2C2608.24848%2C2608.25777%2C2608.26200%2C2608.26204%2C2608.27477%2C2608.27508%2C2608.27808%2C2608.27866%2C2608.29305%2C2608.30038%2C2608.30207%2C2609.00028%2C2609.00048%2C2609.00524%2C2609.02057%2C2609.02149%2C2609.02154%2C2609.02309%2C2609.02885%2C2609.03438%2C2609.04850%2C2609.04869%2C2609.05374%2C2609.05511%2C2609.05513%2C2609.05535%2C2609.06530%2C2609.06792%2C2609.07603%2C2609.07712%2C2609.09212%2C2609.10031&max_results=50"}], "id": "arxiv:2608.24848", "identifiers": {"arxiv": "2608.24848"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:13Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2608.24848"}], "name": "BrowserForge: Scaling Web Episode via Parallel Browser Sandboxes", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-08-25", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces BrowserForge, which combines broad website sourcing, parallel browser sandboxes, and task-proposer/solver interaction to collect verified web trajectories. The pipeline uses page structure during synthesis but trains a policy that acts from screenshots.", "topics": ["data-generation", "sandbox", "web-navigation"], "updated_at": "2026-08-25", "url": "https://arxiv.org/abs/2608.24848"}
{"aliases": [], "authors": ["Jiahe Ying", "Wendong Bu", "Kaihang Pan", "Bingchen Miao", "Siyu Chen", "Wen Wang", "Xueming Jiang", "Juncheng Li", "Siliang Tang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "352699b04f3a9406d7cab95ce9b5fb000e3e98a28996777e75b7399a6290357a", "submitted_at": "2026-08-28T03:15:11Z", "updated_at": "2026-08-28T03:15:11Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2608.27866v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:13Z", "role": "primary", "sha256": "ca895b2f0dcd74816fb5e775a7161a0932ff7120a72477791f5e4c0bb1ead88e", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2608.17319%2C2608.18307%2C2608.20319%2C2608.20707%2C2608.20797%2C2608.21794%2C2608.21830%2C2608.21832%2C2608.21898%2C2608.22077%2C2608.22533%2C2608.22577%2C2608.22847%2C2608.24015%2C2608.24017%2C2608.24099%2C2608.24174%2C2608.24680%2C2608.24848%2C2608.25777%2C2608.26200%2C2608.26204%2C2608.27477%2C2608.27508%2C2608.27808%2C2608.27866%2C2608.29305%2C2608.30038%2C2608.30207%2C2609.00028%2C2609.00048%2C2609.00524%2C2609.02057%2C2609.02149%2C2609.02154%2C2609.02309%2C2609.02885%2C2609.03438%2C2609.04850%2C2609.04869%2C2609.05374%2C2609.05511%2C2609.05513%2C2609.05535%2C2609.06530%2C2609.06792%2C2609.07603%2C2609.07712%2C2609.09212%2C2609.10031&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:13Z", "role": "primary", "sha256": "ca895b2f0dcd74816fb5e775a7161a0932ff7120a72477791f5e4c0bb1ead88e", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2608.17319%2C2608.18307%2C2608.20319%2C2608.20707%2C2608.20797%2C2608.21794%2C2608.21830%2C2608.21832%2C2608.21898%2C2608.22077%2C2608.22533%2C2608.22577%2C2608.22847%2C2608.24015%2C2608.24017%2C2608.24099%2C2608.24174%2C2608.24680%2C2608.24848%2C2608.25777%2C2608.26200%2C2608.26204%2C2608.27477%2C2608.27508%2C2608.27808%2C2608.27866%2C2608.29305%2C2608.30038%2C2608.30207%2C2609.00028%2C2609.00048%2C2609.00524%2C2609.02057%2C2609.02149%2C2609.02154%2C2609.02309%2C2609.02885%2C2609.03438%2C2609.04850%2C2609.04869%2C2609.05374%2C2609.05511%2C2609.05513%2C2609.05535%2C2609.06530%2C2609.06792%2C2609.07603%2C2609.07712%2C2609.09212%2C2609.10031&max_results=50"}], "id": "arxiv:2608.27866", "identifiers": {"arxiv": "2608.27866"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:13Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2608.27866"}], "name": "Iron: Intent-Aligned and Retrospective Dual Learning Framework for Enhancing Generalist Virtual Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-08-28", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces Iron, a GUI-agent training framework that aligns low-level actions with high-level intent using cycle-consistency rewards. It also repurposes failed trajectories through hindsight reconstruction and evaluates transfer to other digital tasks.", "topics": ["reinforcement-learning", "data-generation", "grounding"], "updated_at": "2026-08-28", "url": "https://arxiv.org/abs/2608.27866"}
{"aliases": [], "authors": ["Chen Xiong", "Zhiyuan He", "Pin-Yu Chen", "Stjepan Picek", "Tsung-Yi Ho"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "87c3f58eebe2db437b1166ff8e83b336db4cde37684a7cd86f61c4b34fca7237", "submitted_at": "2026-08-31T03:39:43Z", "updated_at": "2026-10-03T10:41:41Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2608.30207v2"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2608.30207", "identifiers": {"arxiv": "2608.30207"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2608.30207"}], "name": "SIR: Self-improving Red-teaming for Compute Use Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-08-31", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces SIR, which iteratively refines a reusable library of prompt-injection strategies from failed computer-use attack trajectories. Its evaluation uses deterministic system-state checks and counts attacks only when the adversarial objective and legitimate task both complete.", "topics": ["safety-security", "evaluation", "data-generation"], "updated_at": "2026-10-03", "url": "https://arxiv.org/abs/2608.30207"}
{"aliases": [], "authors": ["Kelvin Li", "Dhruv Pendharkar", "Anish Pahilajani", "Chuyi Shang", "Leon Oks", "Leonid Karlinsky", "Rogerio Feris", "Trevor Darrell", "Roei Herzig"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "8703474cb4f577b307731cf1a986f76f800470fde625b6ca4b3569d30e74f903", "submitted_at": "2026-09-02T17:59:40Z", "updated_at": "2026-09-02T17:59:40Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.02885v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2609.02885", "identifiers": {"arxiv": "2609.02885"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.02885"}], "name": "Discriminative World Models for Web Agents", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-09-02", "relations": [], "review_status": "curated", "scope": "core", "summary": "Trains discriminative web world models to distinguish the true next state from outcomes of alternative actions. Branching web trajectories support training, followed by tests of predicted-state matching, process-reward action ranking and end-to-end WebArena-Lite selection.", "topics": ["world-models", "reward-models", "planning", "data-generation", "evaluation"], "updated_at": "2026-09-02", "url": "https://arxiv.org/abs/2609.02885"}
{"aliases": [], "authors": ["Haoting Shi", "Wenhao Wang", "Weicheng Fang", "Yaozhong Liang", "Tian Jin", "Pengxiang Zhao", "Guangyi Liu", "Siheng Chen", "Yanfeng Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "69eab19b6e45e1b2ddf9f29f6f0937d51e06a5bb7dea34dadb761361c82f08d4", "submitted_at": "2026-09-04T17:26:18Z", "updated_at": "2026-09-04T17:26:18Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.05374v1"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:13Z", "role": "primary", "sha256": "ca895b2f0dcd74816fb5e775a7161a0932ff7120a72477791f5e4c0bb1ead88e", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2608.17319%2C2608.18307%2C2608.20319%2C2608.20707%2C2608.20797%2C2608.21794%2C2608.21830%2C2608.21832%2C2608.21898%2C2608.22077%2C2608.22533%2C2608.22577%2C2608.22847%2C2608.24015%2C2608.24017%2C2608.24099%2C2608.24174%2C2608.24680%2C2608.24848%2C2608.25777%2C2608.26200%2C2608.26204%2C2608.27477%2C2608.27508%2C2608.27808%2C2608.27866%2C2608.29305%2C2608.30038%2C2608.30207%2C2609.00028%2C2609.00048%2C2609.00524%2C2609.02057%2C2609.02149%2C2609.02154%2C2609.02309%2C2609.02885%2C2609.03438%2C2609.04850%2C2609.04869%2C2609.05374%2C2609.05511%2C2609.05513%2C2609.05535%2C2609.06530%2C2609.06792%2C2609.07603%2C2609.07712%2C2609.09212%2C2609.10031&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:13Z", "role": "primary", "sha256": "ca895b2f0dcd74816fb5e775a7161a0932ff7120a72477791f5e4c0bb1ead88e", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2608.17319%2C2608.18307%2C2608.20319%2C2608.20707%2C2608.20797%2C2608.21794%2C2608.21830%2C2608.21832%2C2608.21898%2C2608.22077%2C2608.22533%2C2608.22577%2C2608.22847%2C2608.24015%2C2608.24017%2C2608.24099%2C2608.24174%2C2608.24680%2C2608.24848%2C2608.25777%2C2608.26200%2C2608.26204%2C2608.27477%2C2608.27508%2C2608.27808%2C2608.27866%2C2608.29305%2C2608.30038%2C2608.30207%2C2609.00028%2C2609.00048%2C2609.00524%2C2609.02057%2C2609.02149%2C2609.02154%2C2609.02309%2C2609.02885%2C2609.03438%2C2609.04850%2C2609.04869%2C2609.05374%2C2609.05511%2C2609.05513%2C2609.05535%2C2609.06530%2C2609.06792%2C2609.07603%2C2609.07712%2C2609.09212%2C2609.10031&max_results=50"}], "id": "arxiv:2609.05374", "identifiers": {"arxiv": "2609.05374"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:13Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.05374"}], "name": "CUA-Universe: A Scalable and Dynamic Environment for Hybrid GUI+CLI Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-09-04", "relations": [], "review_status": "curated", "scope": "core", "summary": "Turns desktop applications into reproducible GUI-and-CLI environments and generates verified hybrid trajectories for agent post-training.", "topics": ["tool-integration", "data-generation", "sandbox"], "updated_at": "2026-09-04", "url": "https://arxiv.org/abs/2609.05374"}
{"aliases": [], "authors": ["Jianwei Zhang", "Sihan Cao", "Pengcheng Zheng", "Ya Wen", "Pei Ke", "Kuien Liu", "Shen Gao", "Wei Dong", "Yang Yang", "Chaoning Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "aaeb2a57b2bb92ed926c9b65fb40b6bbf272e0ad9bb7415716d24eab0aeb4402", "submitted_at": "2026-08-31T08:40:34Z", "updated_at": "2026-08-31T08:40:34Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.05513v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2609.05513", "identifiers": {"arxiv": "2609.05513"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.05513"}], "name": "When and What to Teach: Budget-Aware Online Adaptation for Web Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-08-31", "relations": [], "review_status": "curated", "scope": "core", "summary": "Proposes an online teaching framework that decides which web-agent episodes warrant a stronger teacher and which trajectory turns merit training. A solvability gate and score-guided trimming aim to preserve first-attempt performance while reducing teaching and training costs.", "topics": ["efficiency", "data-generation", "web-navigation"], "updated_at": "2026-08-31", "url": "https://arxiv.org/abs/2609.05513"}
{"aliases": [], "authors": ["Tong Ye", "Kunyang Han", "Guozhi Wang", "Longqiang Luo", "Zhifeng Ding", "Yongxiang Zhang", "Xiaolei Shen", "Yuxuan Zhang", "Zhuping Zhang", "Tao Xu", "Yue Pan", "Yucheng Zhao", "Yupei Hu", "Yuanjiang Ouyang", "Danfeng Shen", "Runqi Lin", "Hongda Cai", "Zhaoxiong Wang", "Mengjia Yan", "Yingjie Zhong", "Chen Zhou", "Zeyu Zhang", "Xuwen Zhu", "Penggang Shi", "Mingcheng Luo", "Ziyang Wu", "Min Jin", "Mingfu Shen", "Zairong Xu", "Fan Zhang", "Hao Wang", "Liang Liu", "Zhulin Xie", "Lijun Yao", "Xiao Liang", "Liangmin Wen", "Liqiang Feng", "Feilong Wu", "Min Hu", "Min Chen", "Guanjing Xiong", "Xiaohu Ruan", "Xiaoxin Chen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "cc15573bca49d453001ec22061f190e429704a5318751b041550e3610c469f4e", "submitted_at": "2026-09-11T03:36:35Z", "updated_at": "2026-09-15T06:36:04Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2609.12394v3"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2609.12394", "identifiers": {"arxiv": "2609.12394"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:29Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.12394"}], "name": "BlueLM-GUI Technical Report: A Real-Device-Centric Flywheel for Self-Improving Mobile GUI Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-09-11", "relations": [], "review_status": "curated", "scope": "core", "summary": "Describes a mobile-agent training pipeline centered on real-device rollouts, recovery of useful supervision from failures, and evolving evaluation tasks.", "topics": ["mobile-navigation", "reinforcement-learning", "data-generation"], "updated_at": "2026-09-15", "url": "https://arxiv.org/abs/2609.12394"}
{"aliases": [], "authors": ["Liyang Fan", "Xinping Bi", "Yitai Li", "Shuaimin Li", "Hui Li", "Min Yang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "ccbc85f4d2598b4cb1e6e1a96b3b9184ddeba838cb610658070fd2d1ac9c8d06", "submitted_at": "2026-09-16T14:06:38Z", "updated_at": "2026-09-16T14:06:38Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.18690v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2609.18690", "identifiers": {"arxiv": "2609.18690"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:29Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.18690"}], "name": "RankGround: Efficient High-Resolution GUI Grounding via Lightweight Reranker-Guided Crop Selection", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-09-16", "relations": [], "review_status": "curated", "scope": "core", "summary": "RankGround trains a lightweight reranker to select a promising screenshot crop before one VLM localization call. Ranking supervision uses containment and boundary-aware augmentation, with pointwise and listwise training stages.", "topics": ["grounding", "efficiency", "data-generation", "screen-understanding"], "updated_at": "2026-09-16", "url": "https://arxiv.org/abs/2609.18690"}
{"aliases": [], "authors": ["Shuai Bai", "Jiayong Deng", "Sicheng Fan", "Yikun Fu", "Chang Gao", "Xuhao Hu", "Mianqiu Huang", "Yizhen Jiang", "Yuheng Jing", "Dehui Kong", "Keliang Li", "Ning Li", "Wanli Li", "Dayiheng Liu", "Dunjie Lu", "Changwei Luo", "Que Shen", "Zheyuan Wang", "Zijian Wang", "Jie Wu", "Gao Wu", "Zhihui Xie", "Rui Xie", "Haiyang Xu", "An Yang", "Jiakang Yuan", "Yanming Zhang", "Jiajun Zhang", "Xi Zhang", "Zhenru Zhang", "Zhuo Zhen", "Mingkang Zhu", "Bowen Zhou"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "3b312c5a4d638ec9f9a7408fd9ce2d8cbf74fa462d48c8aa37f59e4ad3bd8edc", "submitted_at": "2026-09-18T17:00:56Z", "updated_at": "2026-09-21T07:09:10Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2609.22000v2"}}, "environments": ["desktop", "linux", "macos", "windows", "android", "web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2609.22000", "identifiers": {"arxiv": "2609.22000"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:29Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.22000"}], "name": "RecreationWorld: Scalable and Verifiable Environments for Hybrid Computer-Use Agents", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-09-18", "relations": [], "review_status": "curated", "scope": "core", "summary": "RecreationWorld gives hybrid agents running reference applications to explore and recreate using native GUI controls and coding tools. Its multi-platform benchmark checks interaction-dependent behavior with programmatic and visual assertions, while trajectory-training transfer is evaluated separately.", "topics": ["evaluation", "data-generation", "tool-integration", "workflow"], "updated_at": "2026-09-21", "url": "https://arxiv.org/abs/2609.22000"}
{"aliases": [], "authors": ["Mingfei Gao", "Rui Tian", "Haiming Gang", "Bohan Zhai", "Le Zhang", "Yuanzheng Gong", "Di Feng", "Ege Özsoy", "Kaixin Ma", "Vishwesh Kirthivasan", "Oğuzhan Fatih Kar", "Roman Bachmann", "Anders Boesen Lindbo Larsen", "Afshin Dehghan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "4dfb0703133032426ff8f40ed6a2c8131025b07658c0b643603f8d50ba163d16", "submitted_at": "2026-09-18T17:59:34Z", "updated_at": "2026-09-18T17:59:34Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.22083v1"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:17Z", "role": "primary", "sha256": "5d7c9ff24833b254d33a41a7c4247e830e4d9c371881a1e16151bad5c71cc8b3", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2609.10297%2C2609.10451%2C2609.12394%2C2609.12404%2C2609.13287%2C2609.13491%2C2609.13548%2C2609.13873%2C2609.14239%2C2609.15134%2C2609.15307%2C2609.15364%2C2609.15457%2C2609.16251%2C2609.16635%2C2609.16732%2C2609.16900%2C2609.17653%2C2609.17885%2C2609.18411%2C2609.18690%2C2609.19125%2C2609.19523%2C2609.19843%2C2609.22083%2C2609.22724%2C2609.24277%2C2609.24890%2C2609.25769%2C2609.27273%2C2609.27307%2C2609.27353%2C2609.27461%2C2609.29892%2C2609.30186%2C2609.30604%2C2609.30887%2C2609.31957%2C2609.32036%2C2609.32130%2C2609.32166%2C2609.32351%2C2609.32385%2C2609.32600%2C2609.32679%2C2609.32750%2C2609.32915%2C2609.33304%2C2609.33646%2C2609.34113&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:17Z", "role": "primary", "sha256": "5d7c9ff24833b254d33a41a7c4247e830e4d9c371881a1e16151bad5c71cc8b3", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?id_list=2609.10297%2C2609.10451%2C2609.12394%2C2609.12404%2C2609.13287%2C2609.13491%2C2609.13548%2C2609.13873%2C2609.14239%2C2609.15134%2C2609.15307%2C2609.15364%2C2609.15457%2C2609.16251%2C2609.16635%2C2609.16732%2C2609.16900%2C2609.17653%2C2609.17885%2C2609.18411%2C2609.18690%2C2609.19125%2C2609.19523%2C2609.19843%2C2609.22083%2C2609.22724%2C2609.24277%2C2609.24890%2C2609.25769%2C2609.27273%2C2609.27307%2C2609.27353%2C2609.27461%2C2609.29892%2C2609.30186%2C2609.30604%2C2609.30887%2C2609.31957%2C2609.32036%2C2609.32130%2C2609.32166%2C2609.32351%2C2609.32385%2C2609.32600%2C2609.32679%2C2609.32750%2C2609.32915%2C2609.33304%2C2609.33646%2C2609.34113&max_results=50"}], "id": "arxiv:2609.22083", "identifiers": {"arxiv": "2609.22083"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:17Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.22083"}], "name": "MintAct: A Unified Visual Agent for Digital Environments", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-09-18", "relations": [], "review_status": "curated", "scope": "core", "summary": "MintAct trains a visual-agent family for grounding, navigation and tool use across mobile, desktop and web environments. Its shared infrastructure supports trajectory collection and asynchronous online RL while controlling cross-domain sampling and off-policy drift.", "topics": ["grounding", "reinforcement-learning", "data-generation", "tool-integration"], "updated_at": "2026-09-18", "url": "https://arxiv.org/abs/2609.22083"}
{"aliases": [], "authors": ["Zeyu Kang", "Zhenyun Yin", "Yang Zhang", "Shan He", "Shanzhe Lei", "Yanjiu Zhong", "Xinquan Chen", "Xuhong Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a88d287d5433009962a905dbdbf03359e1183e66e4a6015441ec28405f64a6ee", "submitted_at": "2026-08-27T05:43:50Z", "updated_at": "2026-09-22T06:39:18Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2609.22178v2"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2609.22178", "identifiers": {"arxiv": "2609.22178"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.22178"}], "name": "Beyond Task Completion: Training Capable and Safe Computer-Use Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-08-27", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces SCOPE, which trains computer-use agents on ordinary task demonstrations, safe continuations around hazards, and explicit refusals. Its SCOPE-Gen pipeline pairs verifiable tasks with environmental-risk variants to support joint capability and safety training.", "topics": ["safety-security", "reinforcement-learning", "data-generation"], "updated_at": "2026-09-22", "url": "https://arxiv.org/abs/2609.22178"}
{"aliases": [], "authors": ["Tingyu Qu", "Weigao Sun", "Yuecheng Liu", "Yucheng Zhao", "Yi Zhu", "Yifeng Ding", "Qiyi Wang", "Sihan Cao", "Pengkun Jiao", "Hanlei Xie", "Xiongwei Wu", "Qichao Wang", "Haodong Zhang", "Jiajun Liu", "Yuhao Wang", "Yuqing Xie", "Junpeng Zhao", "Long Chen", "Ming Ma", "Sihan Yang", "Ziwang Zhao", "Yanhao Jia", "Liangquan Gong", "Feida Zhu", "Yiran Zhong", "Steven Hoi"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "572ba5397fc1e4bb138febc3e7004e6e4f244e0ba8b628fb91debef4a854561f", "submitted_at": "2026-09-24T14:38:08Z", "updated_at": "2026-09-24T14:38:08Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.29892v1"}, "license_review": {"document": {"evidence_urls": ["https://arxiv.org/html/2609.29892v1"], "reason": "Record the paper-specific arXiv document declaration; it does not establish code, data or model-weight permissions."}}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:17Z", "role": "primary", "sha256": "5d7c9ff24833b254d33a41a7c4247e830e4d9c371881a1e16151bad5c71cc8b3", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2609.10297%2C2609.10451%2C2609.12394%2C2609.12404%2C2609.13287%2C2609.13491%2C2609.13548%2C2609.13873%2C2609.14239%2C2609.15134%2C2609.15307%2C2609.15364%2C2609.15457%2C2609.16251%2C2609.16635%2C2609.16732%2C2609.16900%2C2609.17653%2C2609.17885%2C2609.18411%2C2609.18690%2C2609.19125%2C2609.19523%2C2609.19843%2C2609.22083%2C2609.22724%2C2609.24277%2C2609.24890%2C2609.25769%2C2609.27273%2C2609.27307%2C2609.27353%2C2609.27461%2C2609.29892%2C2609.30186%2C2609.30604%2C2609.30887%2C2609.31957%2C2609.32036%2C2609.32130%2C2609.32166%2C2609.32351%2C2609.32385%2C2609.32600%2C2609.32679%2C2609.32750%2C2609.32915%2C2609.33304%2C2609.33646%2C2609.34113&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T15:43:20.402537Z", "role": "primary", "sha256": "d4385231cea2965a48895eeebd2dd6000265dd7b43731a0ad88b0373a21d0308", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "licenses.document"], "url": "https://arxiv.org/html/2609.29892v1"}], "id": "arxiv:2609.29892", "identifiers": {"arxiv": "2609.29892"}, "kind": "paper", "last_verified_at": "2026-10-20T15:43:20.402537Z", "licenses": {"code": null, "data": null, "document": {"basis": "The exact-v1 arXiv HTML infobox (#license-tr) declares License: CC BY 4.0 for this paper. This declaration is document-only.", "declared": "CC-BY-4.0", "retrieved_at": "2026-10-20T15:43:20.402537Z", "url": "https://arxiv.org/html/2609.29892v1"}, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.29892"}], "name": "Qwen-Planner-Agent: A Closed-Loop AI-for-AI Framework for Real-World Mobile Planner Agents", "notes": "Primarily typed tool actions and structured feedback, not pixel-coordinate GUI control; backends combine programmatic sandboxes, LLM simulation and selected real-device sessions. No specific OS is established. Model/harness updates are reviewed offline, not autonomous serving-time changes. Co-evolution scores use an evolve/development set; output-cost estimates exclude input, tools, devices and harness processing. Memory comparisons differ in history access and processing cost; user authorization relies on trusted upstream services. No independent reproduction.", "published_at": "2026-09-24", "relations": [], "review_status": "curated", "scope": "core", "summary": "Trains a mobile planner with supervised trajectories and competence-adaptive RL, then couples typed tool execution with memory, skills and offline model-harness refinement.", "topics": ["planning", "data-generation", "reinforcement-learning", "memory", "mobile-navigation", "efficiency"], "updated_at": "2026-09-24", "url": "https://arxiv.org/abs/2609.29892"}
{"aliases": [], "authors": ["Yuchen Sun", "Chenglin Cai", "Gongjie Zhang", "Tianyu Xia", "Quyu Kong", "Panrong Tong", "Zhengwen Zeng", "Long Chen", "Steven Hoi", "Chongyang Zhang", "Yue Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "838635f067b706c7b36a948c8100b30bf4cacd218470aaed6970e5186fa44fb9", "submitted_at": "2026-09-25T06:50:04Z", "updated_at": "2026-09-25T06:50:04Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.30887v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T08:27:30.162100+00:00", "role": "primary", "sha256": "096098b62f82867ee39dfd7cf44095edc5c6a9307351265592e485d82eadf111", "supports": ["name", "kind", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agent%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22+OR+ti%3A%22browser+agents%22+OR+ti%3A%22mobile+agent%22+OR+ti%3A%22mobile+agents%22+OR+ti%3A%22desktop+agent%22%29+AND+submittedDate%3A%5B202609200000+TO+202610192359%5D&start=0&max_results=200&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T10:53:44.035673+00:00", "role": "primary", "sha256": "2b6b8147cf7b0e1fcded85f397c272bcf535ed179789e76c3ff4f15aa3fe53a7", "supports": ["summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/html/2609.30887v1"}], "id": "arxiv:2609.30887", "identifiers": {"arxiv": "2609.30887"}, "kind": "paper", "last_verified_at": "2026-10-20T08:27:30.162100+00:00", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.30887"}], "name": "From Tapping to Hopping: Augmenting Mobile GUI Agents with App-Native Deeplinks", "notes": "Reviewed the complete retained Atom abstract and selected exact-v1 full text. The Atom has incomplete clauses; its text/hash remain unchanged and this annotation uses the explicit methods. Evaluation is Android-only, with a selected real-app subset. Landing-Activity verification is not task completion. Model/catalog release is prospective in the paper. No independent reproduction.", "published_at": "2026-09-25", "relations": [], "review_status": "curated", "scope": "core", "summary": "GUI-Hopper combines verified Android deeplinks with GUI actions, using landing-screen descriptions, hybrid trajectory training and reinforcement learning to select navigation shortcuts and recover through GUI control.", "topics": ["mobile-navigation", "tool-integration", "data-generation", "imitation-learning", "reinforcement-learning", "efficiency"], "updated_at": "2026-09-25", "url": "https://arxiv.org/abs/2609.30887"}
{"aliases": [], "authors": ["Kiana Hooshanfar", "Amirhossein Kazerouni", "Alireza Hosseini", "Michael Brudno", "Babak Taati"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "9ce76f34abe391289643321027a33ac3cac8b5bd64eaa901f9a5fbf7c5561362", "submitted_at": "2026-09-25T14:59:14Z", "updated_at": "2026-09-25T14:59:14Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.31364v1"}, "license_review": {"document": {"evidence_urls": ["https://arxiv.org/html/2609.31364v1"], "reason": "Document declaration only. The paper does not establish code, model-weight or underlying dataset permissions through this license."}}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-20T15:06:37Z", "role": "primary", "sha256": "a3d753815e82190ee2de6f322cd816c32f2dd391f862845305fc7e57300303a6", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28abs%3A%22desktop+agent%22+OR+abs%3A%22computer+interaction%22+OR+abs%3A%22GUI+automation%22+OR+abs%3A%22web+navigation%22+OR+abs%3A%22browser+automation%22+OR+abs%3A%22visual+grounding%22+OR+abs%3A%22screen+understanding%22+OR+abs%3A%22mobile+agents%22%29+AND+submittedDate%3A%5B202609200000+TO+202610200000%5D&start=0&max_results=200&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T19:01:29.156954Z", "role": "primary", "sha256": "cd4ba28778b725c63f4a7329f683b2a287db2e46ed228a0d47006fcd95bbce80", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "licenses.document"], "url": "https://arxiv.org/html/2609.31364v1"}], "id": "arxiv:2609.31364", "identifiers": {"arxiv": "2609.31364"}, "kind": "paper", "last_verified_at": "2026-10-20T19:01:29.156954Z", "licenses": {"code": null, "data": null, "document": {"basis": "The exact-v1 arXiv HTML infobox (#license-tr) declares License: CC BY 4.0 for this paper document.", "declared": "CC-BY-4.0", "retrieved_at": "2026-10-20T19:01:29.156954Z", "url": "https://arxiv.org/html/2609.31364v1"}, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.31364"}], "name": "OpenVAM: Open-World Visual Attention Modeling with VLMs", "notes": "Static human-attention modeling on U-EYE webpage screenshots, with FIWI held out from training; no live-browser, agent-action or task-success evaluation. Gemini-generated reference rationales are expert-checked, and JScore compares generated text with references; human validation covers a separate 100-image subset. Reference text and heatmap alignment do not establish causal explanations of gaze or temporal scanpaths. Several baseline scores come from prior papers. Code/data/weight releases and licenses were not verified; no independent reproduction.", "published_at": "2026-09-25", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "Combines dense saliency prediction with explanations of attended elements, including webpage layouts; evaluates human-attention agreement and textual rationales rather than agent actions.", "topics": ["screen-understanding", "ui-design", "evaluation", "data-generation"], "updated_at": "2026-09-25", "url": "https://arxiv.org/abs/2609.31364"}
{"aliases": [], "authors": ["Zhenhao Zhang", "Zhaoyu Fan", "Haohan Ying", "Jingwen Hu", "Hancen Fan", "Junhao Zhou", "Zitian Chen", "Linchao Zhu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "fa2bcfa05fcbd6937af6ba3d23e892e5a105248939538dcdc9d7b015daf38593", "submitted_at": "2026-09-25T20:04:24Z", "updated_at": "2026-09-25T20:04:24Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.31957v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2609.31957", "identifiers": {"arxiv": "2609.31957"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:29Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.31957"}], "name": "CaptchaArena: A Large-Scale, Fine-Grained Dataset for Training Computer-Use Agents on Interactive CAPTCHAs", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-09-25", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces CaptchaArena, a dataset of execution-verified interactive CAPTCHA solutions with screenshot–action trajectories, reasoning annotations, and pixel masks for irregular targets. The work trains a single policy across CAPTCHA types using supervised learning followed by verifier-reward reinforcement learning.", "topics": ["data-generation", "grounding", "reinforcement-learning"], "updated_at": "2026-09-25", "url": "https://arxiv.org/abs/2609.31957"}
{"aliases": [], "authors": ["Yangzhe Peng", "Xiaoyang Wang", "Yiyang Zhao", "Lijun Wu", "Kun He"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "7cc6fd87d2299e2881eab9a2db96e4022663e2d94da5549dcde166c0c8b8be11", "submitted_at": "2026-09-26T08:14:50Z", "updated_at": "2026-09-26T08:14:50Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.32351v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:17Z", "role": "primary", "sha256": "5d7c9ff24833b254d33a41a7c4247e830e4d9c371881a1e16151bad5c71cc8b3", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2609.10297%2C2609.10451%2C2609.12394%2C2609.12404%2C2609.13287%2C2609.13491%2C2609.13548%2C2609.13873%2C2609.14239%2C2609.15134%2C2609.15307%2C2609.15364%2C2609.15457%2C2609.16251%2C2609.16635%2C2609.16732%2C2609.16900%2C2609.17653%2C2609.17885%2C2609.18411%2C2609.18690%2C2609.19125%2C2609.19523%2C2609.19843%2C2609.22083%2C2609.22724%2C2609.24277%2C2609.24890%2C2609.25769%2C2609.27273%2C2609.27307%2C2609.27353%2C2609.27461%2C2609.29892%2C2609.30186%2C2609.30604%2C2609.30887%2C2609.31957%2C2609.32036%2C2609.32130%2C2609.32166%2C2609.32351%2C2609.32385%2C2609.32600%2C2609.32679%2C2609.32750%2C2609.32915%2C2609.33304%2C2609.33646%2C2609.34113&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:17Z", "role": "primary", "sha256": "5d7c9ff24833b254d33a41a7c4247e830e4d9c371881a1e16151bad5c71cc8b3", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2609.10297%2C2609.10451%2C2609.12394%2C2609.12404%2C2609.13287%2C2609.13491%2C2609.13548%2C2609.13873%2C2609.14239%2C2609.15134%2C2609.15307%2C2609.15364%2C2609.15457%2C2609.16251%2C2609.16635%2C2609.16732%2C2609.16900%2C2609.17653%2C2609.17885%2C2609.18411%2C2609.18690%2C2609.19125%2C2609.19523%2C2609.19843%2C2609.22083%2C2609.22724%2C2609.24277%2C2609.24890%2C2609.25769%2C2609.27273%2C2609.27307%2C2609.27353%2C2609.27461%2C2609.29892%2C2609.30186%2C2609.30604%2C2609.30887%2C2609.31957%2C2609.32036%2C2609.32130%2C2609.32166%2C2609.32351%2C2609.32385%2C2609.32600%2C2609.32679%2C2609.32750%2C2609.32915%2C2609.33304%2C2609.33646%2C2609.34113&max_results=50"}], "id": "arxiv:2609.32351", "identifiers": {"arxiv": "2609.32351"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:17Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.32351"}], "name": "From Trajectories to Grounded Preferences: Process Preference Synthesis via Interaction Element Graphs for Web PRMs", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-09-26", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces SURFPRM, which turns web demonstrations into an interaction-element graph for synthesizing grounded negative actions. Its contrastive preference data trains process reward models to distinguish plausible on-page alternatives and guide downstream trajectory search.", "topics": ["data-generation", "reward-models", "web-navigation"], "updated_at": "2026-09-26", "url": "https://arxiv.org/abs/2609.32351"}
{"aliases": [], "authors": ["Xunjian Yin", "Tianchen Guan", "Jinao Wang", "Weili Cao", "Daisy Xinlei Lin", "Royce Cheng-Yue", "Keagan Long", "Kyle Wong", "Bhuwan Dhingra", "Xiangjun Wang", "Shuyan Zhou"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "197cfb8424584955bc85602d6559037ad31ee815f89b45dc8f9f18c8636552ad", "submitted_at": "2026-09-20T16:16:23Z", "updated_at": "2026-09-20T16:16:23Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.35814v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:20Z", "role": "primary", "sha256": "c420ee40cc87a574ae2554556843f27f2eda43af553f88e95affd6319b1dae33", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2609.34321%2C2609.34769%2C2609.34974%2C2609.35026%2C2609.35671%2C2609.35814%2C2609.35904%2C2609.36201%2C2609.36488%2C2609.36923%2C2609.36927%2C2609.36939%2C2609.37053%2C2609.37176%2C2609.37686%2C2609.38008%2C2609.38397%2C2609.38778%2C2609.38867%2C2609.39306%2C2609.39547%2C2609.39903%2C2609.40253%2C2609.40284%2C2610.00948%2C2610.01215%2C2610.01491%2C2610.01787%2C2610.02320%2C2610.02932%2C2610.03036%2C2610.03089%2C2610.03574%2C2610.03818%2C2610.04116%2C2610.04800%2C2610.04906%2C2610.05185%2C2610.05437%2C2610.05861%2C2610.06829%2C2610.07118%2C2610.07127%2C2610.07444%2C2610.07972%2C2610.08699%2C2610.08773%2C2610.09240%2C2610.09469%2C2610.11050&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:20Z", "role": "primary", "sha256": "c420ee40cc87a574ae2554556843f27f2eda43af553f88e95affd6319b1dae33", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?id_list=2609.34321%2C2609.34769%2C2609.34974%2C2609.35026%2C2609.35671%2C2609.35814%2C2609.35904%2C2609.36201%2C2609.36488%2C2609.36923%2C2609.36927%2C2609.36939%2C2609.37053%2C2609.37176%2C2609.37686%2C2609.38008%2C2609.38397%2C2609.38778%2C2609.38867%2C2609.39306%2C2609.39547%2C2609.39903%2C2609.40253%2C2609.40284%2C2610.00948%2C2610.01215%2C2610.01491%2C2610.01787%2C2610.02320%2C2610.02932%2C2610.03036%2C2610.03089%2C2610.03574%2C2610.03818%2C2610.04116%2C2610.04800%2C2610.04906%2C2610.05185%2C2610.05437%2C2610.05861%2C2610.06829%2C2610.07118%2C2610.07127%2C2610.07444%2C2610.07972%2C2610.08699%2C2610.08773%2C2610.09240%2C2610.09469%2C2610.11050&max_results=50"}], "id": "arxiv:2609.35814", "identifiers": {"arxiv": "2609.35814"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:20Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.35814"}], "name": "Constructing Challenging Browser-Use Tasks by Controlled Environment Interventions", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-09-20", "relations": [], "review_status": "curated", "scope": "core", "summary": "BreakingWeb pairs solvable browser tasks with controlled environment interventions that preserve the goal, target and backend success criterion. The paired evaluations isolate robustness failures, including agents declaring completion without the required state change.", "topics": ["evaluation", "robustness", "data-generation", "state-tracking"], "updated_at": "2026-09-20", "url": "https://arxiv.org/abs/2609.35814"}
{"aliases": [], "authors": ["Tongbo Chen", "Junbo Niu", "Zhengxi Lu", "Niu Lian", "Fei Tang", "Yuchen Yan", "Yike Hong", "Yong Du", "Yizhou Liu", "Bofan Chen", "Yongliang Shen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "b394677ebe37782950cfc56c0be65ca7a93f34e5550ec9425d91921bf1799847", "submitted_at": "2026-09-29T16:57:25Z", "updated_at": "2026-09-29T16:57:25Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.38008v1"}}, "environments": ["desktop", "windows"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2609.38008", "identifiers": {"arxiv": "2609.38008"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:29Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.38008"}], "name": "HybridCUA: Learning to Orchestrate GUI and CLI for Computer-Use Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-09-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "Trains computer agents on verified GUI, command-line, and interleaved trajectories to learn when and how to combine interaction modes.", "topics": ["tool-integration", "reinforcement-learning", "data-generation"], "updated_at": "2026-09-29", "url": "https://arxiv.org/abs/2609.38008"}
{"aliases": [], "authors": ["Bo Han", "Qianyi Wang", "Shuai Liu", "Xiong Zifan", "Changqiao Wu", "Yuanfa Li", "Pengzhi Gao", "Wei Liu", "Jian Luan", "Heng Qu", "Yunpeng Song", "Zhongmin Cai"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "3594620e97b41f3a7481871588b63ffa4aa8822e8beb42334b2cd888bcfcfaaf", "submitted_at": "2026-09-30T11:45:51Z", "updated_at": "2026-09-30T11:45:51Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.39547v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2609.39547", "identifiers": {"arxiv": "2609.39547"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:29Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.39547"}], "name": "Learning Reliable GUI Agents under Imperfect Priors", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-09-30", "relations": [], "review_status": "curated", "scope": "core", "summary": "Combines structured GUI exploration with training designed to tolerate imperfect retrieved experience. It builds state-transition graphs and synthetic task trajectories, then injects realistic forms of interface drift so agents learn when to rely on or reject prior guidance.", "topics": ["data-generation", "memory", "robustness"], "updated_at": "2026-09-30", "url": "https://arxiv.org/abs/2609.39547"}
{"aliases": [], "authors": ["Cheng Yang", "Yifan Wu", "Yutao Huang", "Zhaohua Zhang", "Beiduo Chen", "Muxi Chen", "Chenchen Zhao", "Hexuan Deng", "Haolin Yang", "Geyuan Zhu", "Sa Zhu", "Jianhuan Zhuo", "Qiuyong Xiao", "Jianhao Ruan", "Yiran Peng", "Jiayi Zhang", "Tian Ye", "Xinlei Yu", "Tianwen Jiang", "Jihong Zhang", "Yuyu Luo"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "ecd40b78c6de41abb442705b993aad8ebf7b35741536a11e00ebbd60140d1f18", "submitted_at": "2026-10-01T07:22:53Z", "updated_at": "2026-10-01T07:22:53Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.01215v1"}}, "environments": ["desktop", "linux", "windows", "macos", "web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2610.01215", "identifiers": {"arxiv": "2610.01215"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:29Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2610.01215"}], "name": "AutoGUIWorld: Image Generators as Visual World Models for GUI Agent", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-10-01", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces AutoGUIWorld, which combines task planning and image editing to synthesize action-grounded GUI trajectories without running the depicted software. It filters generated transitions and tests whether training on them improves agents operating real desktop and scientific applications.", "topics": ["world-models", "data-generation", "grounding"], "updated_at": "2026-10-01", "url": "https://arxiv.org/abs/2610.01215"}
{"aliases": [], "authors": ["A. Said Gurbuz", "Ahmed Nassar", "Sunghwan Hong", "Marc Pollefeys", "Peter W. J. Staar"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "bd66468c2670ec22216e1592719a05e0b56af69606237a50f2067a340172553e", "submitted_at": "2026-10-01T18:00:05Z", "updated_at": "2026-10-01T18:00:05Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.02320v1"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2610.02320", "identifiers": {"arxiv": "2610.02320"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:29Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2610.02320"}], "name": "DeskForge: Dense Supervision from Desktop Environments for Computer-Use Agents", "notes": "Editorial summary based on the primary paper abstract; experiments and released artifacts have not been independently reproduced.", "published_at": "2026-10-01", "relations": [], "review_status": "curated", "scope": "core", "summary": "Generates dense desktop supervision by varying real application scenes and combining screenshots, accessibility trees, and window geometry.", "topics": ["data-generation", "grounding"], "updated_at": "2026-10-01", "url": "https://arxiv.org/abs/2610.02320"}
{"aliases": [], "authors": ["Yongxin Ning", "Runliang Niu", "Qianli Xing", "Zhiyi Duan", "Qingzu He", "Pan Wang", "Qi Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "43845ad64843a0b019c1988d4ba861eccee3c48a7b2c09d0bae7794e5e86347a", "submitted_at": "2026-10-05T06:28:58Z", "updated_at": "2026-10-05T06:28:58Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.05861v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2610.05861", "identifiers": {"arxiv": "2610.05861"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:29Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2610.05861"}], "name": "Imagine to Act: High-Fidelity Data Synthesis via Image Editing World Model for Scalable GUI Agent Training", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-10-05", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces Infinite-Dreamer, which learns action-conditioned screenshot editing from structured descriptions of GUI changes and uses it to synthesize robustness examples and multi-step trajectories. The synthetic data trains a mobile GUI agent evaluated on AndroidWorld, MobileWorld, and AndroidControl-Curated.", "topics": ["world-models", "data-generation", "mobile-navigation"], "updated_at": "2026-10-05", "url": "https://arxiv.org/abs/2610.05861"}
{"aliases": [], "authors": ["Tian Jin", "Wenhao Wang", "Jing Kang", "Haoting Shi", "Rui Ye", "Yijun Lu", "Siheng Chen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "06b25b25e804683e11305e21dd04feb9131aad27f901998eda70ac91e527c6b9", "submitted_at": "2026-10-08T18:31:16Z", "updated_at": "2026-10-14T08:40:10Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2610.12611v2"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2610.12611", "identifiers": {"arxiv": "2610.12611"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:29Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2610.12611"}], "name": "SoftwareGym: Scaling Computer-Use Agent Training with Diverse Synthesized Software Environments", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-10-08", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces SoftwareGym, a pipeline that synthesizes runnable GUI applications and iteratively repairs them using execution checks. It also provides a task benchmark and studies whether training on synthesized application trajectories transfers to other computer-use environments.", "topics": ["data-generation", "evaluation", "sandbox"], "updated_at": "2026-10-14", "url": "https://arxiv.org/abs/2610.12611"}
{"aliases": [], "authors": ["Yuqi Huang", "Hui Li", "Yifei Cao", "Zikun Cai", "Tianci Xie", "Yangfan Gao", "Pinxue Guo", "Chaoyun Zhang", "Bingxuan Li", "Rui Zheng", "Hang Yan", "Ning Liao"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "0e90bb713e7034a64821502eb6fc792fa91a5d1e420aff9ff658506d39269f03", "submitted_at": "2026-10-09T04:32:02Z", "updated_at": "2026-10-09T04:32:02Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.12984v1"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:24Z", "role": "primary", "sha256": "9641755401e0e8ec9613f7c4509029d968ab4f10751ba4d43cab57e7d27c1da1", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2610.11655%2C2610.11707%2C2610.11942%2C2610.12611%2C2610.12823%2C2610.12866%2C2610.12984%2C2610.13268%2C2610.13416%2C2610.13703%2C2610.14157%2C2610.14228%2C2610.15138%2C2610.15765%2C2610.16234%2C2610.16319%2C2610.16720%2C2610.16778%2C2610.17457%2C2610.17710%2C2610.17792%2C2610.17895%2C2610.17928%2C2610.18508%2C2610.18784%2C2610.18868%2C2610.19100%2C2610.19128%2C2610.20584&max_results=29"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:24Z", "role": "primary", "sha256": "9641755401e0e8ec9613f7c4509029d968ab4f10751ba4d43cab57e7d27c1da1", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2610.11655%2C2610.11707%2C2610.11942%2C2610.12611%2C2610.12823%2C2610.12866%2C2610.12984%2C2610.13268%2C2610.13416%2C2610.13703%2C2610.14157%2C2610.14228%2C2610.15138%2C2610.15765%2C2610.16234%2C2610.16319%2C2610.16720%2C2610.16778%2C2610.17457%2C2610.17710%2C2610.17792%2C2610.17895%2C2610.17928%2C2610.18508%2C2610.18784%2C2610.18868%2C2610.19100%2C2610.19128%2C2610.20584&max_results=29"}], "id": "arxiv:2610.12984", "identifiers": {"arxiv": "2610.12984"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:24Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2610.12984"}], "name": "Work2Skill: Capturing Human GUI Experience for Reusable Agent Skill Induction", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-10-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "Aligns human screen recordings, input events, and accessibility information to derive reusable workflow skills from desktop demonstrations.", "topics": ["recording", "workflow", "data-generation"], "updated_at": "2026-10-09", "url": "https://arxiv.org/abs/2610.12984"}
{"aliases": [], "authors": ["Ruotian Zhang", "Yuhui Zhang", "Junzhe Huang", "Yu Xin", "Hehe Fan", "Xiaoxiao Sun"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "010e8ebc657dccad670ea6d17985e9c3c1680017f2516eaa86b7ea9db5660990", "submitted_at": "2026-10-09T09:20:02Z", "updated_at": "2026-10-09T09:20:02Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.13268v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2610.13268", "identifiers": {"arxiv": "2610.13268"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:29Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2610.13268"}], "name": "ConClick: Element-Wise Contrastive Learning Improves GUI Grounding", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-10-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "Analyzes recurring distractor clicks in GUI grounding and introduces ConClick, a contrastive training method built from target–distractor pairs. Its ReClick inference procedure revisits candidate regions and compares competing predictions.", "topics": ["grounding", "data-generation", "robustness"], "updated_at": "2026-10-09", "url": "https://arxiv.org/abs/2610.13268"}
{"aliases": [], "authors": ["Chenyang Yan", "Yingying Zhang", "Kanzhi Cheng", "Qiushi Sun", "Zhengyuan Pan", "Zheng Ma", "Hang Yan", "Yian Wang", "Nuo Chen", "Jialin Cao", "Xingdong Gong", "Wenpo Song", "Han Chen", "Zichen Ding", "Fangzhi Xu", "Shujian Huang", "Xinyu Dai", "Yichen Liu", "Tiankuo Yao", "Bo Wang", "Ben Kao", "Jianbing Zhang", "Lewei Lu", "Dahua Lin"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6f7d940d3e341fd8c9f23a4829a5ab1150fc07ee231ccda9b108a3f0d785efb9", "submitted_at": "2026-10-09T17:26:38Z", "updated_at": "2026-10-09T17:26:38Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.14157v1"}}, "environments": ["android", "web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:24Z", "role": "primary", "sha256": "9641755401e0e8ec9613f7c4509029d968ab4f10751ba4d43cab57e7d27c1da1", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2610.11655%2C2610.11707%2C2610.11942%2C2610.12611%2C2610.12823%2C2610.12866%2C2610.12984%2C2610.13268%2C2610.13416%2C2610.13703%2C2610.14157%2C2610.14228%2C2610.15138%2C2610.15765%2C2610.16234%2C2610.16319%2C2610.16720%2C2610.16778%2C2610.17457%2C2610.17710%2C2610.17792%2C2610.17895%2C2610.17928%2C2610.18508%2C2610.18784%2C2610.18868%2C2610.19100%2C2610.19128%2C2610.20584&max_results=29"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:24Z", "role": "primary", "sha256": "9641755401e0e8ec9613f7c4509029d968ab4f10751ba4d43cab57e7d27c1da1", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2610.11655%2C2610.11707%2C2610.11942%2C2610.12611%2C2610.12823%2C2610.12866%2C2610.12984%2C2610.13268%2C2610.13416%2C2610.13703%2C2610.14157%2C2610.14228%2C2610.15138%2C2610.15765%2C2610.16234%2C2610.16319%2C2610.16720%2C2610.16778%2C2610.17457%2C2610.17710%2C2610.17792%2C2610.17895%2C2610.17928%2C2610.18508%2C2610.18784%2C2610.18868%2C2610.19100%2C2610.19128%2C2610.20584&max_results=29"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T01:46:06Z", "role": "primary", "sha256": "9b385e9f6aa2f7169f616a76887df3b76df605cb2bea071d99ee7ca3f9b659cc", "supports": ["environments"], "url": "https://os-copilot.github.io/OpenMobile2-Home/"}], "id": "arxiv:2610.14157", "identifiers": {"arxiv": "2610.14157"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:24Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2610.14157"}], "name": "OpenMobile-2: Building Versatile Mobile Agents with Scalable Environments and App-Native Tools", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-10-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces OpenMobile-2, combining simulated mobile apps, emulator tasks, released training data, and policies that can alternate GUI interaction with app-native tools.", "topics": ["mobile-navigation", "tool-integration", "data-generation", "reinforcement-learning"], "updated_at": "2026-10-09", "url": "https://arxiv.org/abs/2610.14157"}
{"aliases": [], "authors": ["Zijian Cai", "Haowei Liu", "Kailai Chen", "Lin Peng", "Cong Wan", "Ziqi Zhang", "Chunfeng Yuan", "Bing Li", "Weiming Hu", "Haiyang Xu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "4596cc60436c4fa6c2aa0951b4699b804edb73aa1b1a46900b1d14f901d20f3b", "submitted_at": "2026-10-12T07:42:11Z", "updated_at": "2026-10-12T07:42:11Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.16319v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:24Z", "role": "primary", "sha256": "9641755401e0e8ec9613f7c4509029d968ab4f10751ba4d43cab57e7d27c1da1", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2610.11655%2C2610.11707%2C2610.11942%2C2610.12611%2C2610.12823%2C2610.12866%2C2610.12984%2C2610.13268%2C2610.13416%2C2610.13703%2C2610.14157%2C2610.14228%2C2610.15138%2C2610.15765%2C2610.16234%2C2610.16319%2C2610.16720%2C2610.16778%2C2610.17457%2C2610.17710%2C2610.17792%2C2610.17895%2C2610.17928%2C2610.18508%2C2610.18784%2C2610.18868%2C2610.19100%2C2610.19128%2C2610.20584&max_results=29"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:24Z", "role": "primary", "sha256": "9641755401e0e8ec9613f7c4509029d968ab4f10751ba4d43cab57e7d27c1da1", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2610.11655%2C2610.11707%2C2610.11942%2C2610.12611%2C2610.12823%2C2610.12866%2C2610.12984%2C2610.13268%2C2610.13416%2C2610.13703%2C2610.14157%2C2610.14228%2C2610.15138%2C2610.15765%2C2610.16234%2C2610.16319%2C2610.16720%2C2610.16778%2C2610.17457%2C2610.17710%2C2610.17792%2C2610.17895%2C2610.17928%2C2610.18508%2C2610.18784%2C2610.18868%2C2610.19100%2C2610.19128%2C2610.20584&max_results=29"}], "id": "arxiv:2610.16319", "identifiers": {"arxiv": "2610.16319"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:24Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2610.16319"}], "name": "HyWe: Unifying Search and Operation for Real-World Hybrid Web Tasks", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-10-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces HyWe, which switches between information search and browser operation within one trajectory, alongside a demonstration pipeline and a benchmark of hybrid web tasks.", "topics": ["web-navigation", "planning", "data-generation"], "updated_at": "2026-10-12", "url": "https://arxiv.org/abs/2610.16319"}
{"aliases": [], "authors": ["Wayne Chi", "I-Hung Hsu", "Li Chen", "Wei-Lin Chiang", "Ion Stoica", "Anastasios Nikolas Angelopoulos", "Ameet Talwalkar", "Chris Donahue", "Cho-Jui Hsieh"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "14d2128155d1540955e54c3bdf9010724f27cf2e101079650315602856c98efa", "submitted_at": "2026-10-13T07:49:50Z", "updated_at": "2026-10-14T06:37:47Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2610.17895v2"}}, "environments": ["web", "desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:24Z", "role": "primary", "sha256": "9641755401e0e8ec9613f7c4509029d968ab4f10751ba4d43cab57e7d27c1da1", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2610.11655%2C2610.11707%2C2610.11942%2C2610.12611%2C2610.12823%2C2610.12866%2C2610.12984%2C2610.13268%2C2610.13416%2C2610.13703%2C2610.14157%2C2610.14228%2C2610.15138%2C2610.15765%2C2610.16234%2C2610.16319%2C2610.16720%2C2610.16778%2C2610.17457%2C2610.17710%2C2610.17792%2C2610.17895%2C2610.17928%2C2610.18508%2C2610.18784%2C2610.18868%2C2610.19100%2C2610.19128%2C2610.20584&max_results=29"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:24Z", "role": "primary", "sha256": "9641755401e0e8ec9613f7c4509029d968ab4f10751ba4d43cab57e7d27c1da1", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2610.11655%2C2610.11707%2C2610.11942%2C2610.12611%2C2610.12823%2C2610.12866%2C2610.12984%2C2610.13268%2C2610.13416%2C2610.13703%2C2610.14157%2C2610.14228%2C2610.15138%2C2610.15765%2C2610.16234%2C2610.16319%2C2610.16720%2C2610.16778%2C2610.17457%2C2610.17710%2C2610.17792%2C2610.17895%2C2610.17928%2C2610.18508%2C2610.18784%2C2610.18868%2C2610.19100%2C2610.19128%2C2610.20584&max_results=29"}], "id": "arxiv:2610.17895", "identifiers": {"arxiv": "2610.17895"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:24Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2610.17895"}], "name": "Interaction-as-a-Judge", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-10-13", "relations": [], "review_status": "curated", "scope": "core", "summary": "Uses computer-use agents to interact with generated applications, studying whether interaction improves preference judging, requirement checks, and feedback for application generation.", "topics": ["evaluation", "data-generation", "automation"], "updated_at": "2026-10-14", "url": "https://arxiv.org/abs/2610.17895"}
{"aliases": [], "authors": ["Hang He", "Linting Wang", "Guanming Liu", "Bo Mao", "Jinxiu Liu", "Yiran Suo", "Maosen Zhou", "Zhiheng Xi", "Guanting Dong", "Zhou Shao", "Rui Zheng", "Jie Zhou", "Chengcheng Wan", "Tao Gui", "Qi Zhang", "Xipeng Qiu", "Xuanjing Huang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6add363f25f87ea1814c7b79ebf6b51f7431b3745a7d2d910960b78e2d97d13f", "submitted_at": "2026-10-13T09:52:02Z", "updated_at": "2026-10-13T09:52:02Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.18067v1"}, "license_review": {"document": {"evidence_urls": ["https://arxiv.org/html/2610.18067v1"], "reason": "Document declaration only; code, data and weights permissions are not inferred."}}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T12:31:39Z", "role": "primary", "sha256": "c52e4e283e0dae88ef8b3ce67b7424b1d4b48e8f41015d872f4da69f98782eaa", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2609.39924%2C2610.00346%2C2610.00583%2C2610.01939%2C2610.05099%2C2610.09127%2C2610.09484%2C2610.09633%2C2610.11251%2C2610.11922%2C2610.13098%2C2610.13166%2C2610.13361%2C2610.13623%2C2610.16117%2C2610.17953%2C2610.18067%2C2610.19366%2C2610.20424&max_results=19"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T19:02:20Z", "role": "primary", "sha256": "f1fbdd2ecb2a54b7723372118e2b0ac818420f3a3dd280c2486c8e064cf486d6", "supports": ["name", "kind", "summary", "scope", "environments", "topics", "notes", "licenses.document"], "url": "https://arxiv.org/html/2610.18067v1"}], "id": "arxiv:2610.18067", "identifiers": {"arxiv": "2610.18067"}, "kind": "paper", "last_verified_at": "2026-10-20T19:02:20Z", "licenses": {"code": null, "data": null, "document": {"basis": "The exact-v1 arXiv HTML infobox explicitly declares CC BY 4.0 for this paper.", "declared": "CC-BY-4.0", "retrieved_at": "2026-10-20T19:02:20Z", "url": "https://arxiv.org/html/2610.18067v1"}, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2610.18067"}], "name": "WorkRover: Scaling Long-Horizon Cowork Agent Training Via Recursive Environment Evolution", "notes": "App WorkEnvs expose UI observations and interaction interfaces alongside file and tool workflows; no specific operating system or pixel-control success is established. Verifier shaping depends on deterministic-check and dependency coverage. External reference scores use differing harnesses or subsets; GDPval Elo is not comparable to percentage metrics. A successful-trace analysis subset is separate from online RL outcomes. No implementation release or independent reproduction was verified.", "published_at": "2026-10-13", "relations": [], "review_status": "curated", "scope": "core", "summary": "Synthesizes resettable workspaces with files, application observations and skills; repairs environment incompatibilities and trains long-task agents using verifier-linked turn credit.", "topics": ["data-generation", "long-horizon", "evaluation", "reward-models", "reinforcement-learning", "workflow"], "updated_at": "2026-10-13", "url": "https://arxiv.org/abs/2610.18067"}
{"aliases": [], "authors": ["Qingsong Xie", "Jiahao Tian", "Jiacheng Tang", "Zhifeng Li", "Madalina Ciobanu", "Navan Preet Singh", "Qingqing Mao", "Ritankar Das"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "08d90bcec690b238be96207602ca694759c8672fecbed85e30372dd314e4006e", "submitted_at": "2026-10-13T19:37:49Z", "updated_at": "2026-10-13T19:37:49Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.19100v1"}}, "environments": ["desktop", "web", "cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2610.19100", "identifiers": {"arxiv": "2610.19100"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:29Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2610.19100"}], "name": "GroundGUI: GUI Grounding by High-Quality Data and Distillation-to-Reinforcement Learning", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-10-13", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces ScreenRef screenshot grounding data and a training sequence from supervised learning through teacher-guided distillation to reinforcement learning for GroundGUI-8B.", "topics": ["grounding", "data-generation", "reinforcement-learning"], "updated_at": "2026-10-13", "url": "https://arxiv.org/abs/2610.19100"}
{"aliases": [], "authors": ["Md Farhan Ishmam", "Kenneth Marino"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6d154a14c5ac85fb352721c203fe7bd709eaba4287ef7376e6a6f5fc395708f1", "submitted_at": "2026-10-14T09:06:49Z", "updated_at": "2026-10-14T09:06:49Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.19834v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2610.19834", "identifiers": {"arxiv": "2610.19834"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:29Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2610.19834"}], "name": "Teaching Agents in their Own Style: Blackbox Teacher Distillation for Web Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-10-14", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces WebKD, which repairs failed student web-agent trajectories with minimal teacher edits and trains on corrections that preserve the student's reasoning style.", "topics": ["data-generation", "web-navigation", "efficiency"], "updated_at": "2026-10-14", "url": "https://arxiv.org/abs/2610.19834"}
{"aliases": [], "authors": ["Mingyu Ouyang", "Yiran Wang", "Siyuan Hu", "Xingyilang Yin", "Huiqiang Sun", "Guangzhi Wang", "Junfu Pu", "Cheng Cheng", "Honghao Chen", "Ying Shan", "Mike Zheng Shou", "Hwee Tou Ng"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "67b539468bd2bbf0d3b49a5a7eed4791a8ddca4b56ad9515d23a29204d1f069f", "submitted_at": "2026-10-14T17:57:12Z", "updated_at": "2026-10-14T17:57:12Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.20584v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:24Z", "role": "primary", "sha256": "9641755401e0e8ec9613f7c4509029d968ab4f10751ba4d43cab57e7d27c1da1", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2610.11655%2C2610.11707%2C2610.11942%2C2610.12611%2C2610.12823%2C2610.12866%2C2610.12984%2C2610.13268%2C2610.13416%2C2610.13703%2C2610.14157%2C2610.14228%2C2610.15138%2C2610.15765%2C2610.16234%2C2610.16319%2C2610.16720%2C2610.16778%2C2610.17457%2C2610.17710%2C2610.17792%2C2610.17895%2C2610.17928%2C2610.18508%2C2610.18784%2C2610.18868%2C2610.19100%2C2610.19128%2C2610.20584&max_results=29"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:24Z", "role": "primary", "sha256": "9641755401e0e8ec9613f7c4509029d968ab4f10751ba4d43cab57e7d27c1da1", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2610.11655%2C2610.11707%2C2610.11942%2C2610.12611%2C2610.12823%2C2610.12866%2C2610.12984%2C2610.13268%2C2610.13416%2C2610.13703%2C2610.14157%2C2610.14228%2C2610.15138%2C2610.15765%2C2610.16234%2C2610.16319%2C2610.16720%2C2610.16778%2C2610.17457%2C2610.17710%2C2610.17792%2C2610.17895%2C2610.17928%2C2610.18508%2C2610.18784%2C2610.18868%2C2610.19100%2C2610.19128%2C2610.20584&max_results=29"}], "id": "arxiv:2610.20584", "identifiers": {"arxiv": "2610.20584"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:24Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2610.20584"}], "name": "Player-1: Learning to Play from Synthetic Experience", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-10-14", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "Extends computer-use policies to game control through synthetic training environments and richer keyboard and pointer actions. Its evaluation concerns games rather than ordinary productivity workflows.", "topics": ["data-generation", "reinforcement-learning", "evaluation"], "updated_at": "2026-10-14", "url": "https://arxiv.org/abs/2610.20584"}
{"aliases": [], "authors": ["Chulhoon Yoon", "Kyungri Park", "Jinyeong Yun", "Daeyun Jeong", "Sukmin Yun", "Woohwan Jung"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "b9585696d256bacc710dce57bfd7cda6a8037ce5725a30f959028352bc07e97c", "submitted_at": "2026-10-16T06:44:56Z", "updated_at": "2026-10-16T06:44:56Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.23012v1"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-19T04:42:05Z", "role": "primary", "sha256": "ffbb3f28a3a0fc8888f30ea8b2cdaf441239685aaacc2d80f311581f4b5a1d5c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agent%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B202610050000+TO+202610190000%5D&start=0&max_results=200&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T08:27:26.236797+00:00", "role": "primary", "sha256": "80bed80786f939973d0d13b4f892300fee02a6110e326247131fc83d257796de", "supports": ["name", "kind", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agent%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22+OR+ti%3A%22browser+agents%22+OR+ti%3A%22mobile+agent%22+OR+ti%3A%22mobile+agents%22+OR+ti%3A%22desktop+agent%22%29+AND+submittedDate%3A%5B202610130000+TO+202610192359%5D&start=0&max_results=200&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T10:34:04.380045+00:00", "role": "primary", "sha256": "75c025a862a046b8bcbdd84b800a37132edcc1ec271d40c57c18ec984652a0b0", "supports": ["summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/html/2610.23012v1"}], "id": "arxiv:2610.23012", "identifiers": {"arxiv": "2610.23012"}, "kind": "paper", "last_verified_at": "2026-10-20T08:27:26.236797+00:00", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2610.23012"}], "name": "Adaptation of Computer Use Agents to New Software with User Manuals", "notes": "Reviewed the complete retained abstract and selected exact-v1 methods, evaluation and validation sections. Execution plans stay fixed per episode; policy training is separate and application-specific. Screenshot-based training/Slack judgments can miss task conditions, and OSWorld tasks may overlap backbone pretraining. No independent reproduction, safe-deployment guarantee or artifact-release verification.", "published_at": "2026-10-16", "relations": [], "review_status": "curated", "scope": "core", "summary": "MAGE retrieves software-manual sections into fixed execution plans and generates setup/main task-plan pairs whose GUI rollouts support application-specific policy adaptation with round-based GRPO.", "topics": ["planning", "data-generation", "reinforcement-learning"], "updated_at": "2026-10-16", "url": "https://arxiv.org/abs/2610.23012"}
{"aliases": [], "authors": ["Zhanhui Zhou", "Weichen Zhang", "Haoran Liu", "Lingjie Chen", "Tianneng Shi", "Kevin Lin", "Zhengyuan Yang", "Lijuan Wang", "Dawn Song"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "0a297401f7ad4dcd2f485ade6bab295e365280bb803dcecc111aebf0693dd203", "submitted_at": "2026-10-19T09:30:54Z", "updated_at": "2026-10-19T09:30:54Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.26071v1"}}, "environments": ["desktop", "linux", "windows", "web", "android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-20T05:52:00Z", "role": "primary", "sha256": "af310824eea97aac8209023ff53bad7bea77778816586b10bb94110148347af9", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agent%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B202610060000+TO+202610200000%5D&start=0&max_results=200&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T08:27:26.236797+00:00", "role": "primary", "sha256": "80bed80786f939973d0d13b4f892300fee02a6110e326247131fc83d257796de", "supports": ["name", "kind", "summary", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agent%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22+OR+ti%3A%22browser+agents%22+OR+ti%3A%22mobile+agent%22+OR+ti%3A%22mobile+agents%22+OR+ti%3A%22desktop+agent%22%29+AND+submittedDate%3A%5B202610130000+TO+202610192359%5D&start=0&max_results=200&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T08:28:55.967592+00:00", "role": "primary", "sha256": "d4231fa9677e916ca8cc27493dfbbe9fb34c7430207150e6b9be10c2bad43bfb", "supports": ["summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/html/2610.26071v1"}], "id": "arxiv:2610.26071", "identifiers": {"arxiv": "2610.26071"}, "kind": "paper", "last_verified_at": "2026-10-20T08:27:26.236797+00:00", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2610.26071"}], "name": "CUA-Lite: An Open Platform for Computer Use Agents", "notes": "Reviewed the complete official abstract and selected v1 methods, training and limitations sections. WebVoyager uses offline mirrors; VM-free sandboxes are Linux-only, and reported desktop/mobile gains do not establish unseen-app transfer. No results or released artifacts were independently reproduced.", "published_at": "2026-10-19", "relations": [], "review_status": "curated", "scope": "core", "summary": "CUA-Lite unifies environment interfaces, supervised data formats and per-model harnesses for evaluation, reinforcement learning and supervised fine-tuning across desktop, browser and mobile tasks.", "topics": ["tool-integration", "data-generation", "evaluation", "imitation-learning", "reinforcement-learning", "sandbox"], "updated_at": "2026-10-19", "url": "https://arxiv.org/abs/2610.26071"}
{"aliases": [], "authors": ["Zhaoyang Wang", "Qianhui Wu", "Haorui Wang", "Simon Yu", "Baolin Peng", "Canwen Xu", "Hao Cheng", "Wenlin Yao", "Boyi Liu", "Yite Wang", "Jianfeng Gao", "Huaxiu Yao"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "1f1ff33859352b61f017709fff91fd8e32f08636e9423fb044e7ca8e28f06905", "submitted_at": "2026-10-19T17:55:35Z", "updated_at": "2026-10-19T17:55:35Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.26828v1"}}, "environments": ["web", "desktop", "linux"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-20T05:52:00Z", "role": "primary", "sha256": "af310824eea97aac8209023ff53bad7bea77778816586b10bb94110148347af9", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agent%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B202610060000+TO+202610200000%5D&start=0&max_results=200&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T08:27:26.236797+00:00", "role": "primary", "sha256": "80bed80786f939973d0d13b4f892300fee02a6110e326247131fc83d257796de", "supports": ["name", "kind", "summary", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agent%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22+OR+ti%3A%22browser+agents%22+OR+ti%3A%22mobile+agent%22+OR+ti%3A%22mobile+agents%22+OR+ti%3A%22desktop+agent%22%29+AND+submittedDate%3A%5B202610130000+TO+202610192359%5D&start=0&max_results=200&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T08:29:05.337543+00:00", "role": "primary", "sha256": "b680e07a5a883e7bb0210a1734a628be6c47d91b7df0487a2044d56d1b7e227e", "supports": ["summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/html/2610.26828v1"}], "id": "arxiv:2610.26828", "identifiers": {"arxiv": "2610.26828"}, "kind": "paper", "last_verified_at": "2026-10-20T08:27:26.236797+00:00", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2610.26828"}], "name": "Agent World Model in the Wild: Computer-Use Agent Environment Synthesis", "notes": "Reviewed the complete abstract and selected v1 synthesis, training and evaluation sections. The system uses hybrid GUI/MCP/CLI interfaces. Evaluation includes benchmark subsets and model-based judgments, so reported scores are not treated as directly comparable GUI-only results. No independent reproduction.", "published_at": "2026-10-19", "relations": [], "review_status": "curated", "scope": "core", "summary": "WildAWM synthesizes resettable GUI/MCP/CLI environments and cross-interface tasks, using execution-based checks of solvability, difficulty and verifier consistency before supervised and reinforcement learning.", "topics": ["data-generation", "sandbox", "tool-integration", "long-horizon", "imitation-learning", "reinforcement-learning"], "updated_at": "2026-10-19", "url": "https://arxiv.org/abs/2610.26828"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/google-research/google-research/master/android_control/README.md", "population": "AndroidControl collection", "qualifier": "more-than", "unit": "demonstrations", "value": 15000}, {"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/google-research/google-research/master/android_control/README.md", "population": "AndroidControl collection", "qualifier": "exact", "unit": "applications", "value": 833}], "evidence_urls": ["https://raw.githubusercontent.com/google-research/google-research/master/android_control/README.md"], "formats": ["GZIP-compressed TFRecord"], "modalities": ["screenshots", "accessibility trees", "instructions", "JSON actions"], "notes": "Step instructions can be repeated across multiple actions. Each episode has one more screenshot than action; do not count repeated step instructions as independent task goals.", "splits": ["publisher split file"], "supervision": ["interaction trajectories", "high- and low-level action instruction"]}, "modalities": ["screenshot", "accessibility-tree", "instruction", "action"], "reported_scale": {"comparison": ">", "unit": "demonstrations", "value": 15000}, "sample_count": null, "sample_unit": "demonstrations"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:38:29Z", "role": "primary", "sha256": "e4f8baa15441fb63cad07220cc5a4de177a1f785d4fe1fed5cfcbf3c3867faf1", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details.sample_count", "details.sample_unit", "details.reported_scale", "details.modalities", "details.dataset_profile"], "url": "https://raw.githubusercontent.com/google-research/google-research/master/android_control/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T06:42:06Z", "role": "primary", "sha256": "cfc7749b96f63bd31c3c42b5c471bf756814053e847c10f3eb003417bc523d30", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/google-research/google-research/HEAD/LICENSE"}], "id": "component:google-research/android_control", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T06:38:29Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "Apache-2.0", "retrieved_at": "2026-10-16T06:42:06Z", "url": "https://raw.githubusercontent.com/google-research/google-research/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/google-research/google-research/tree/master/android_control"}], "name": "AndroidControl", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/google-research/google-research/master/android_control/README.md", "target_id": "arxiv:2406.03679", "type": "introduced_by"}], "review_status": "curated", "scope": "core", "summary": "Android demonstrations with high-level goals, step instructions, screenshots, and accessibility trees.", "topics": ["data-generation", "mobile-navigation"], "updated_at": null, "url": "https://github.com/google-research/google-research/tree/master/android_control"}
{"aliases": [], "authors": [], "details": {"access": "public download referenced by publisher", "dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/google-research/google-research/master/android_in_the_wild/README.md", "population": "sum of the five published AitW subsets", "qualifier": "exact", "unit": "episodes", "value": 715142}, {"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/google-research/google-research/master/android_in_the_wild/README.md", "population": "sum of the five published AitW subsets", "qualifier": "exact", "unit": "action examples", "value": 5689993}], "evidence_urls": ["https://raw.githubusercontent.com/google-research/google-research/master/android_in_the_wild/README.md"], "formats": ["GZIP-compressed TFRecord"], "modalities": ["screenshots", "instructions", "actions"], "notes": "Episodes and action examples are different units. The collection combines multi-step subsets and hindsight-labeled single-step tasks; preserve subset and split choices.", "splits": [], "supervision": ["interaction trajectories", "action prediction"]}, "download_url": "https://console.cloud.google.com/storage/browser/gresearch/android-in-the-wild", "example_count": 5689993, "modalities": ["screenshot", "action", "instruction"], "sample_count": 715142, "sample_unit": "episodes"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:38:29Z", "role": "primary", "sha256": "6d8964c95db41dada1d5867e50a5ec61f6cb90595965d0e72e010b8e4a12dcb2", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details.sample_count", "details.sample_unit", "details.example_count", "details.modalities", "details.access", "details.download_url", "details.dataset_profile"], "url": "https://raw.githubusercontent.com/google-research/google-research/master/android_in_the_wild/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T06:42:06Z", "role": "primary", "sha256": "cfc7749b96f63bd31c3c42b5c471bf756814053e847c10f3eb003417bc523d30", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/google-research/google-research/HEAD/LICENSE"}], "id": "component:google-research/android_in_the_wild", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T06:38:29Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "Apache-2.0", "retrieved_at": "2026-10-16T06:42:06Z", "url": "https://raw.githubusercontent.com/google-research/google-research/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/google-research/google-research/tree/master/android_in_the_wild"}], "name": "Android in the Wild", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/google-research/google-research/master/android_in_the_wild/README.md", "target_id": "arxiv:2307.10088", "type": "introduced_by"}], "review_status": "curated", "scope": "core", "summary": "Human demonstrations of Android tasks with screenshots and touch actions.", "topics": ["data-generation", "mobile-navigation"], "updated_at": null, "url": "https://github.com/google-research/google-research/tree/master/android_in_the_wild"}
{"aliases": [], "authors": [], "details": {"component_path": "WebEvolver", "license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/Tencent/SelfEvolvingAgent/HEAD/License.txt"], "reason": "Reviewed repository license declaration for code; third-party dependencies, data, weights and executable environment images are separate."}}, "repository_tree_sha": "a82450e1b5e15070bee79ee8b79f8d9a97cb85f7", "repository_url": "https://github.com/Tencent/SelfEvolvingAgent"}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:27:14Z", "role": "primary", "sha256": "447dfa1e52789f7f3f111e9e0d453fb1f9fd8bf2d7b858f28fbe21a2239b4b2e", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations"], "url": "https://raw.githubusercontent.com/Tencent/SelfEvolvingAgent/HEAD/WebEvolver/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T15:37:01Z", "role": "primary", "sha256": "4214af42bcadf1ea52a46c2008f8e741f0aa955a58f346d9880858589be7de1d", "supports": ["identity"], "url": "https://api.github.com/repos/Tencent/SelfEvolvingAgent"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:25:25Z", "role": "primary", "sha256": "ad0f4ae4b58b5a0e84feeb5df717929b97ea37cb945c17c98510ba54509c99e7", "supports": ["identity", "release_status", "details.repository_tree_sha"], "url": "https://api.github.com/repos/Tencent/SelfEvolvingAgent/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:27:13Z", "role": "primary", "sha256": "873a78a46309fa3892d94b72cf73fd55542bae0e0ba8235afb26a848a64a8255", "supports": ["identity"], "url": "https://raw.githubusercontent.com/Tencent/SelfEvolvingAgent/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:35:19Z", "role": "primary", "sha256": "1d7a5369af7ec336bc173dba6cfee9ad0898e527f96c5b7a10cee930806c67a3", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/Tencent/SelfEvolvingAgent/HEAD/License.txt"}], "id": "component:tencent/selfevolvingagent/webevolver", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-18T04:27:14Z", "licenses": {"code": {"basis": "Reviewed repository license declaration for code; third-party dependencies, data, weights and executable environment images are separate.", "declared": "MIT", "retrieved_at": "2026-10-18T04:35:19Z", "url": "https://raw.githubusercontent.com/Tencent/SelfEvolvingAgent/HEAD/License.txt"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/Tencent/SelfEvolvingAgent/tree/main/WebEvolver"}], "name": "WebEvolver code component", "notes": "Scoped to the WebEvolver directory; the repository also hosts WebCoT and VScan, which are not merged into this resource. Training and model services require separate setup. Root License.txt explicitly declares MIT for SelfEvolvingAgent and separately identifies CogKernel’s Apache-2.0 dependency terms.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/Tencent/SelfEvolvingAgent/HEAD/WebEvolver/README.md", "target_id": "arxiv:2504.21024", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Component for web-agent trajectory collection, supervised-data preparation and world-model-based trajectory synthesis.", "topics": ["web-navigation", "world-models", "data-generation", "self-reflection"], "updated_at": null, "url": "https://github.com/Tencent/SelfEvolvingAgent/tree/main/WebEvolver"}
{"aliases": [], "authors": [], "details": {}, "environments": ["cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:02Z", "role": "primary", "sha256": "e06aedcb1760867470afeb27d4eb4cdde9a3b76b9d55af183047df3c0ae2b71c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/alibaba/UI-Ins/HEAD/README.md"}], "id": "github:alibaba/ui-ins", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T06:36:02Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/alibaba/UI-Ins"}], "name": "UI-Ins", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/alibaba/UI-Ins/HEAD/README.md", "target_id": "arxiv:2510.20286", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Treats grounding instructions as multiple reasoning perspectives, releasing a data pipeline, training code and UI-Ins-7B and 32B models trained by SFT on synthesized instructions, then RL.", "topics": ["data-generation"], "updated_at": null, "url": "https://github.com/alibaba/UI-Ins"}
{"aliases": [], "authors": [], "details": {"availability": "announced; data and sampling pipelines in preparation", "dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/AlibabaResearch/RoTS/HEAD/README.md", "population": "paper-described synthesis output; not a released file inventory", "qualifier": "approximately", "unit": "training samples", "value": 800000}], "evidence_urls": ["https://raw.githubusercontent.com/AlibabaResearch/RoTS/HEAD/README.md"], "formats": [], "modalities": ["GUI interaction trajectories"], "notes": "The inspected README marks the dataset and collection pipelines as in preparation. This profile records an announced research asset, not a verified downloadable corpus; file formats, splits, and release-specific data terms remain unknown.", "splits": [], "supervision": ["recovery trajectory synthesis"]}}, "environments": ["cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:02Z", "role": "primary", "sha256": "bf5c1469d33ced951ff0a5de017494b65f654023ceee8e7fbfd8b60afb082244", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://raw.githubusercontent.com/AlibabaResearch/RoTS/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:02Z", "role": "primary", "sha256": "bf5c1469d33ced951ff0a5de017494b65f654023ceee8e7fbfd8b60afb082244", "supports": ["details.dataset_profile"], "url": "https://raw.githubusercontent.com/AlibabaResearch/RoTS/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:02Z", "role": "primary", "sha256": "bf5c1469d33ced951ff0a5de017494b65f654023ceee8e7fbfd8b60afb082244", "supports": ["summary", "notes", "details.availability"], "url": "https://raw.githubusercontent.com/AlibabaResearch/RoTS/HEAD/README.md"}], "id": "github:alibabaresearch/rots", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T06:36:02Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/AlibabaResearch/RoTS"}], "name": "RoTS", "notes": "The 800k figure describes training samples in the study, not an independently counted public download. The repository code-license declaration does not establish terms for an unreleased dataset.", "published_at": null, "relations": [], "review_status": "curated", "scope": "core", "summary": "Announced GUI trajectory synthesis resources for recovery from policy-induced errors; the inspected repository marks code and data releases as in preparation.", "topics": ["data-generation", "robustness"], "updated_at": null, "url": "https://github.com/AlibabaResearch/RoTS"}
{"aliases": [], "authors": [], "details": {}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:06Z", "role": "primary", "sha256": "3e5f24b3175afeba8635dacc7b66eba1ce89fea49996a20281ff2df5a41f1e24", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/amazon-science/PAE/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:05Z", "role": "primary", "sha256": "ae4f966004dd672fd5b994add5a5cba1fee9233d4d81b224e94c339728fafd87", "supports": ["identity"], "url": "https://api.github.com/repos/amazon-science/PAE"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:05Z", "role": "primary", "sha256": "b1e24a3f38fae67f71033a3f920164d282d3f8bb072f438349dbe9eb23730a7a", "supports": ["release_status"], "url": "https://api.github.com/repos/amazon-science/PAE/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:15Z", "role": "primary", "sha256": "db2e35513dbadcdc67f5819a3bfee2777786538dd3531620cd5fbd4b6ed6e538", "supports": ["licenses"], "url": "https://raw.githubusercontent.com/amazon-science/PAE/HEAD/LICENSE.txt"}], "id": "github:amazon-science/pae", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-17T02:02:06Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/amazon-science/PAE"}], "name": "Proposer-Agent-Evaluator (PAE)", "notes": "Source code and browser-environment modules are present. The repository LICENSE is CC-BY-NC-4.0, while its README describes that license as applying to a dataset, leaving code scope ambiguous. The separately linked dataset is gated and was not accessed; model and dataset terms must be checked separately.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/amazon-science/PAE/HEAD/README.md", "target_id": "arxiv:2412.13194", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "A browser-agent training system that proposes tasks, collects visual interaction trajectories, and uses automated evaluation to drive policy improvement.", "topics": ["reinforcement-learning", "web-navigation", "data-generation"], "updated_at": null, "url": "https://github.com/amazon-science/PAE"}
{"aliases": [], "authors": [], "details": {"license_review": {"code": {"evidence_urls": ["https://api.github.com/repos/anikasomaia/local-activity-recorder/git/trees/f7ec5b6ecf0ba1bfe6362d7f2bfb0f1af5ac6bc7?recursive=1"], "reason": "No license declaration was found in the inspected README/C# files, and the complete pinned tree has no LICENSE, LICENCE or COPYING-named file. Public source access does not establish reuse permission."}}, "repository_revision": "967cea087bf6496b3b82743fb96770fa29f8e203"}, "environments": ["desktop", "windows"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-20T11:57:15.716415+00:00", "role": "primary", "sha256": "440e0e43026ba7a20eaca17507cec33bd697d8b1eed6fb056be34136b3f4f76f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://raw.githubusercontent.com/anikasomaia/local-activity-recorder/967cea087bf6496b3b82743fb96770fa29f8e203/README.md"}, {"method": "github_repository_identity", "retrieved_at": "2026-10-20T11:04:41.685729+00:00", "role": "primary", "sha256": "90c30399c8aa5031f82164978bc12eae3bebba172758d20877a87fec53566c12", "supports": ["identity"], "url": "https://api.github.com/search/repositories?q=topic%3Acomputer-use-agents+is%3Apublic+fork%3Afalse+created%3A2008-01-01..2026-10-20&per_page=100&page=1&sort=updated&order=asc"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T11:54:07.061516+00:00", "role": "primary", "sha256": "f5b501fa132163d749c6695d81c76d45fdff2888ea890e31cbaa9bc4172fed3b", "supports": ["identity", "details.repository_revision", "summary", "environments", "topics", "notes"], "url": "https://api.github.com/repos/anikasomaia/local-activity-recorder/commits/master"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T11:56:33.833893+00:00", "role": "primary", "sha256": "ee9e5899d1f150982eb05ca1f8794406cef0ffc80eb8223811b504ff972370c1", "supports": ["identity", "details.repository_revision", "summary", "environments", "notes", "licenses.code"], "url": "https://api.github.com/repos/anikasomaia/local-activity-recorder/git/trees/f7ec5b6ecf0ba1bfe6362d7f2bfb0f1af5ac6bc7?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T12:05:00.168119+00:00", "role": "primary", "sha256": "ce3022b4016d140eac554e392dbe42f54ffb3e8098b03e2bf72e92ff49c5d084", "supports": ["summary", "environments", "topics", "scope", "notes"], "url": "https://raw.githubusercontent.com/anikasomaia/local-activity-recorder/967cea087bf6496b3b82743fb96770fa29f8e203/Program.cs"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T12:05:03.421222+00:00", "role": "primary", "sha256": "e70e2e68b5dd586e5ad0fb0b12c8b937b4614bd5237bac61d03d22144c4751f4", "supports": ["notes"], "url": "https://raw.githubusercontent.com/anikasomaia/local-activity-recorder/967cea087bf6496b3b82743fb96770fa29f8e203/DataUploader.cs"}], "id": "github:anikasomaia/local-activity-recorder", "identifiers": {}, "kind": "tool", "last_verified_at": "2026-10-20T11:57:15.716415+00:00", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/anikasomaia/local-activity-recorder"}], "name": "Activity Recorder", "notes": "Recording is started and stopped from the tray; screenshots use a five-second timer. The manual Upload Data action uses a built-in third-party demonstration endpoint unless changed in code, sending recorded files as base64 JSON. Successful uploads move files into uploaded subfolders rather than delete them; no automatic expiry policy was found in the inspected capture/upload path. Sensitive-application filters are listed as future work. The README's .NET version requirements are inconsistent, and long-session behavior was not established. No license declaration was found in the inspected files or complete tree's license-file names, so code permission remains unknown. No activity was captured and no uploader or executable was run.", "published_at": null, "relations": [], "review_status": "curated", "scope": "core", "summary": "Windows tray utility that records primary-display screenshots, keyboard key codes and mouse events as local JPEG and JSON files for training-data collection.", "topics": ["recording", "data-generation"], "updated_at": null, "url": "https://github.com/anikasomaia/local-activity-recorder"}
{"aliases": [], "authors": [], "details": {}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:44:04Z", "role": "primary", "sha256": "0d4b28c993ccb977c7910ff3fecb5de29db364a7a7ad55523c8e052c9bae2566", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/ApGa/Go-Browse/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:44:04Z", "role": "primary", "sha256": "b02ec75f69ab8bbebe0d59eeccdcdcacf11a0d6b00bdfe803c692a9d587301e2", "supports": ["identity"], "url": "https://api.github.com/repos/ApGa/Go-Browse"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:44:04Z", "role": "primary", "sha256": "cd3a40a596ad9ead856973bb31e5b7f7e88a2f60969799c01a2a8a6d09fe3421", "supports": ["release_status"], "url": "https://api.github.com/repos/ApGa/Go-Browse/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:45:20Z", "role": "primary", "sha256": "e7107f1e1b3a0c65cbbae6c496d9e71b570025bdddebb3d402a6080dd916bcba", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/ApGa/Go-Browse/HEAD/LICENSE"}, {"method": "license_header", "retrieved_at": "2026-10-17T02:45:20Z", "role": "primary", "sha256": "e7107f1e1b3a0c65cbbae6c496d9e71b570025bdddebb3d402a6080dd916bcba", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/ApGa/Go-Browse/HEAD/LICENSE"}], "id": "github:apga/go-browse", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-17T02:44:04Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "MIT", "retrieved_at": "2026-10-17T02:45:20Z", "url": "https://raw.githubusercontent.com/ApGa/Go-Browse/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/ApGa/Go-Browse"}], "name": "Go-Browse", "notes": "The release contains exploration agents, data-processing scripts, fine-tuning and WebArena evaluation entry points under MIT. Checkpoints and processed datasets are separate Hub resources with separate license status. The arXiv badge image has an unrelated number; the linked paper and citation consistently identify 2506.03533.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/ApGa/Go-Browse/HEAD/README.md", "target_id": "arxiv:2506.03533", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Structured website-exploration and data-processing code that revisits discovered pages, proposes tasks and collects trajectories for web-agent fine-tuning.", "topics": ["web-navigation", "data-generation", "imitation-learning"], "updated_at": null, "url": "https://github.com/ApGa/Go-Browse"}
{"aliases": ["github:bytedance/agent-r", "bytedance/Agent-R", "https://github.com/bytedance/Agent-R"], "authors": [], "details": {}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:03:20Z", "role": "primary", "sha256": "8471a0761a94b1af84099a80efb6ac53f17e34491a196e82ca068a5113c0b8be", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/ByteDance-Seed/Agent-R/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:03:19Z", "role": "primary", "sha256": "3fff7137bc4f39a3a757f3b820892fb330449044ef2721e9f9d7719d23ff2b98", "supports": ["identity", "availability"], "url": "https://api.github.com/repos/ByteDance-Seed/Agent-R"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:05:40Z", "role": "primary", "sha256": "e3c70b04094e8e77e249065b2b500ac197fc6fafee6e3275d78289b3790bf7ed", "supports": ["release_status"], "url": "https://api.github.com/repos/ByteDance-Seed/Agent-R/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:05:22Z", "role": "primary", "sha256": "283ea6cc2997a1a70da0049e09adf9317bb60ca1b51279b65196b83a69e1996b", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/ByteDance-Seed/Agent-R/HEAD/LICENSE"}, {"method": "license_header", "retrieved_at": "2026-10-17T02:05:22Z", "role": "primary", "sha256": "283ea6cc2997a1a70da0049e09adf9317bb60ca1b51279b65196b83a69e1996b", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/ByteDance-Seed/Agent-R/HEAD/LICENSE"}, {"method": "github_repository_identity", "retrieved_at": "2026-10-17T18:05:46Z", "role": "primary", "sha256": "3fff7137bc4f39a3a757f3b820892fb330449044ef2721e9f9d7719d23ff2b98", "supports": ["aliases"], "url": "https://api.github.com/repos/bytedance/Agent-R"}, {"method": "github_repository_identity", "retrieved_at": "2026-10-17T18:05:46Z", "role": "primary", "sha256": "3fff7137bc4f39a3a757f3b820892fb330449044ef2721e9f9d7719d23ff2b98", "supports": ["aliases"], "url": "https://api.github.com/repos/ByteDance-Seed/Agent-R"}], "id": "github:bytedance-seed/agent-r", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-17T02:03:20Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "Apache-2.0", "retrieved_at": "2026-10-17T02:05:22Z", "url": "https://raw.githubusercontent.com/ByteDance-Seed/Agent-R/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/ByteDance-Seed/Agent-R"}], "name": "Agent-R", "notes": "The canonical repository is ByteDance-Seed/Agent-R and is archived but readable. MCTS collection, revision and evaluation scripts are released under Apache-2.0. It also supports non-GUI text environments; this entry covers its WebShop interaction component. No separate weight release was verified.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/ByteDance-Seed/Agent-R/HEAD/README.md", "target_id": "arxiv:2501.11425", "type": "documented_by"}], "review_status": "curated", "scope": "adjacent", "summary": "Trajectory collection and revision utilities that use tree search and model-guided critique to train agents to recover from mistakes, including WebShop tasks.", "topics": ["self-reflection", "data-generation", "web-navigation"], "updated_at": null, "url": "https://github.com/ByteDance-Seed/Agent-R"}
{"aliases": [], "authors": [], "details": {"availability": "Public desktop data-generation source; separate training-data and trained-model releases not verified.", "license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/bytedance/GUI-ReWalk/HEAD/LICENSE"], "reason": "Repository code license; no scope transfer to generated trajectories, pretrained detector weights, or GUI application content."}}, "repository_revision": "78a65d003d6f48802fac879508b62f1c89125892"}, "environments": ["desktop", "linux"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:12:15Z", "role": "primary", "sha256": "63708cb2500aefa36916f3fb171b595efaabf614f17b925cfe8b5c8cbf72be10", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations", "details"], "url": "https://raw.githubusercontent.com/bytedance/GUI-ReWalk/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:12:15Z", "role": "primary", "sha256": "34addf6eed41e1e29e51aab8819c5ac316b5699a8c11c4ce29a32c8e38557de8", "supports": ["identity"], "url": "https://api.github.com/repos/bytedance/GUI-ReWalk"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:14:07Z", "role": "primary", "sha256": "a218cb6f8a2c2a243e8207785c98255bf717c00f177521368705d8652e229d7d", "supports": ["release_status"], "url": "https://api.github.com/repos/bytedance/GUI-ReWalk/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:14:08Z", "role": "primary", "sha256": "43070e2d4e532684de521b885f385d0841030efa2b1a20bafb76133a5e1379c1", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/bytedance/GUI-ReWalk/HEAD/LICENSE"}], "id": "github:bytedance/gui-rewalk", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-17T09:12:15Z", "licenses": {"code": {"basis": "Repository code license; no scope transfer to generated trajectories, pretrained detector weights, or GUI application content.", "declared": "Apache-2.0", "retrieved_at": "2026-10-17T09:14:08Z", "url": "https://raw.githubusercontent.com/bytedance/GUI-ReWalk/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/bytedance/GUI-ReWalk"}], "name": "GUI-ReWalk", "notes": "The inspected source tree contains the random walker, desktop environment integration and run_rewalk.sh. The documented runnable setup uses OSWorld and OmniParser. The README also discusses mobile data and a GUI-ReWalk-7B model, but no separately published corpus or model checkpoint was verified; those are not counted as released assets. Apache-2.0 is scoped to the implementation.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/bytedance/GUI-ReWalk/HEAD/README.md", "target_id": "github:xlang-ai/osworld", "type": "uses"}, {"evidence_url": "https://raw.githubusercontent.com/bytedance/GUI-ReWalk/HEAD/README.md", "target_id": "github:microsoft/omniparser", "type": "uses"}, {"evidence_url": "https://raw.githubusercontent.com/bytedance/GUI-ReWalk/HEAD/README.md", "target_id": "arxiv:2509.15738", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "A desktop trajectory-generation framework that moves from stochastic GUI exploration to inferred goals and multi-stage task completion, with retrospective reasoning annotations.", "topics": ["data-generation", "planning", "cross-app"], "updated_at": null, "url": "https://github.com/bytedance/GUI-ReWalk"}
{"aliases": [], "authors": [], "details": {}, "environments": ["desktop", "macos", "linux"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T10:42:12Z", "role": "primary", "sha256": "40a5e9ad3a2169527a9330825bd6e07ac1669293e21c209ed41036987bbf0a8b", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/china-qijizhifeng/Work2Skill/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T10:48:42Z", "role": "primary", "sha256": "70387bc43f567b2ed971c1a789719c8c3567b0d073407c38b5dc6af1eb78012a", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/china-qijizhifeng/Work2Skill/HEAD/LICENSE"}], "id": "github:china-qijizhifeng/work2skill", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-16T10:42:12Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "MIT", "retrieved_at": "2026-10-16T10:48:42Z", "url": "https://raw.githubusercontent.com/china-qijizhifeng/Work2Skill/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/china-qijizhifeng/Work2Skill"}], "name": "Work2Skill", "notes": "The README reports that evaluation integration and live cross-platform acceptance testing remain pending.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/china-qijizhifeng/Work2Skill/HEAD/README.md", "target_id": "arxiv:2610.12984", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Collects human GUI demonstrations and inspects multimodal evidence to derive reusable agent workflow skills.", "topics": ["recording", "workflow", "data-generation"], "updated_at": null, "url": "https://github.com/china-qijizhifeng/Work2Skill"}
{"aliases": [], "authors": [], "details": {"license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/Euphoria16/UI-Genie/HEAD/README.md", "https://api.github.com/repos/Euphoria16/UI-Genie/git/trees/main?recursive=1"], "reason": "The README declares MIT and links a root LICENSE, but the complete retained tree has no root LICENSE; bundled ms-swift has its own license. Preserve this unresolved scope instead of treating a component license as the whole project license."}}}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T03:59:54Z", "role": "primary", "sha256": "e08569bd9d8a6ff44537d4fe69243ddfeb8a41ae5c9f87ef84af50a67c64c726", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "licenses.code"], "url": "https://raw.githubusercontent.com/Euphoria16/UI-Genie/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T03:59:54Z", "role": "primary", "sha256": "a53e4cd3b4391c03c9334745152f8c1f66aff7590051f9bdb7e78b81c6c99a24", "supports": ["identity"], "url": "https://api.github.com/repos/Euphoria16/UI-Genie"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T03:59:55Z", "role": "primary", "sha256": "c27c6d88d043c964776c04a74517d48644418dca2f5ae9c8ba778d082c12a6ae", "supports": ["release_status", "licenses.code"], "url": "https://api.github.com/repos/Euphoria16/UI-Genie/git/trees/main?recursive=1"}], "id": "github:euphoria16/ui-genie", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-17T03:59:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/Euphoria16/UI-Genie"}], "name": "UI-Genie", "notes": "The tree contains ms-swift-based training and AndroidControl evaluation scripts. The README says AndroidLab/Android Arena scripts and the 72B checkpoint are still forthcoming; these are not claimed released. It declares MIT but links a root LICENSE absent from the inspected tree, so the exact code-license scope remains unresolved. Separate Hub weights and data have their own declarations.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/Euphoria16/UI-Genie/HEAD/README.md", "target_id": "arxiv:2505.21496", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Training and evaluation code for a mobile-agent framework that iteratively generates trajectories using agent and reward-model feedback.", "topics": ["data-generation", "reward-models", "imitation-learning"], "updated_at": null, "url": "https://github.com/Euphoria16/UI-Genie"}
{"aliases": ["STEVE", "STEVE-R1"], "authors": [], "details": {}, "environments": ["desktop", "windows", "linux"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T13:14:29Z", "role": "primary", "sha256": "52aa383fdf5d91951f5d9a73f4ebed4454a82b35edae296918116e9f0895dd83", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/FanbinLu/STEVE/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["relations", "environments"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "github:fanbinlu/steve-r1", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-16T13:14:29Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/FanbinLu/STEVE-R1"}], "name": "STEVE / STEVE-R1 research code", "notes": "The repository has moved to STEVE-R1. Its README reports an SFT release, with RL tuning in progress, and describes evaluations using multiple attempts per task. These results are not independently reproduced here.", "published_at": null, "relations": [{"evidence_url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending", "target_id": "arxiv:2503.12532", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Research code and linked models for step-verified computer-use training; the current repository describes STEVE-R1 and its trajectory releases.", "topics": ["data-generation", "automation", "evaluation"], "updated_at": null, "url": "https://github.com/FanbinLu/STEVE-R1"}
{"aliases": [], "authors": [], "details": {"license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/GAIR-NLP/PC-Agent-E/HEAD/LICENSE"], "reason": "Reviewed repository license declaration for code; third-party dependencies, data, weights and executable environment images are separate."}}, "repository_tree_sha": "9b1b00b907b99a4a8dbfc9c01ea49010878dfbd6", "repository_url": "https://github.com/GAIR-NLP/PC-Agent-E"}, "environments": ["desktop", "windows", "linux"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:27:13Z", "role": "primary", "sha256": "3ffb8c843c6acb8c2d5bbba11ba49d6b36cd566553a89974b6252a2bfd289d5d", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations"], "url": "https://raw.githubusercontent.com/GAIR-NLP/PC-Agent-E/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T15:37:01Z", "role": "primary", "sha256": "30761141ea62c5f20183f359bbc3fc8ccc5b458886d38c5426bae4802a23ee0b", "supports": ["identity"], "url": "https://api.github.com/repos/GAIR-NLP/PC-Agent-E"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:25:23Z", "role": "primary", "sha256": "2fb2aa4ee6f02ccadb631dab4972f96c3fdf545cd92c5854fc191e81cf4ae88a", "supports": ["identity", "release_status", "details.repository_tree_sha"], "url": "https://api.github.com/repos/GAIR-NLP/PC-Agent-E/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:35:19Z", "role": "primary", "sha256": "39e37b84c067596361c491e2e9e22af8805386a24b8352b87e9aca3bddf4f92c", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/GAIR-NLP/PC-Agent-E/HEAD/LICENSE"}], "id": "github:gair-nlp/pc-agent-e", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-18T04:27:13Z", "licenses": {"code": {"basis": "Reviewed repository license declaration for code; third-party dependencies, data, weights and executable environment images are separate.", "declared": "MIT", "retrieved_at": "2026-10-18T04:35:19Z", "url": "https://raw.githubusercontent.com/GAIR-NLP/PC-Agent-E/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/GAIR-NLP/PC-Agent-E"}], "name": "PC-Agent-E training and deployment", "notes": "README demonstrates Windows and Linux operation and describes thought completion and trajectory augmentation. The released source tree contains postprocess, train and deploy code. The checkpoint, data archive and WindowsAgentArena-V2 are separate resources; external model APIs and a training framework are required.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/GAIR-NLP/PC-Agent-E/HEAD/README.md", "target_id": "arxiv:2505.13909", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Tools for refining and augmenting human computer-use trajectories, preparing supervised training data, and deploying a native PC-Agent-E scaffold.", "topics": ["data-generation", "imitation-learning", "automation"], "updated_at": null, "url": "https://github.com/GAIR-NLP/PC-Agent-E"}
{"aliases": [], "authors": [], "details": {"license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/kwai/MobileForge/main/LICENSE", "https://raw.githubusercontent.com/kwai/MobileForge/main/README.md"], "reason": "The complete root LICENSE declares Apache-2.0 for repository code. Linked data and model weights, and third-party dependencies, retain separate licensing requirements."}}, "repository_id": 1278021888}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-20T03:55:04.230792+00:00", "role": "primary", "sha256": "f78f58a6e3b8dd2ae3e8c49a15473782ae258ecd70719f7eeddfa43a10441adf", "supports": ["name", "kind", "summary", "topics", "scope", "relations", "notes", "environments", "licenses.code"], "url": "https://raw.githubusercontent.com/kwai/MobileForge/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T03:55:29.533431+00:00", "role": "primary", "sha256": "58187fe25d8398ef2ae31de02debd4bc3b932ff1925db86a87f5de65b016aed1", "supports": ["summary", "environments", "topics", "scope", "notes"], "url": "https://raw.githubusercontent.com/kwai/MobileForge/main/training/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T03:55:07.420029+00:00", "role": "primary", "sha256": "c71d239df91726fc519c6eb72d318ec65820627232b2f796219e87dcf35d0ab4", "supports": ["licenses.code", "notes"], "url": "https://raw.githubusercontent.com/kwai/MobileForge/main/LICENSE"}, {"method": "github_repository_identity", "retrieved_at": "2026-10-19T19:10:26.954787+00:00", "role": "primary", "sha256": "0ae4477b885b0915e43874399c2a6b91f7d07954697370204550df931f7db82c", "supports": ["details.repository_id"], "url": "https://api.github.com/search/repositories?q=topic%3Agui-agent+is%3Apublic+fork%3Afalse+created%3A2008-01-01..2026-10-18&per_page=100&page=1&sort=updated&order=asc"}], "id": "github:kwai/mobileforge", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-20T03:55:04.230792+00:00", "licenses": {"code": {"basis": "Complete repository root Apache License 2.0 text; code declaration only, not linked datasets or weights.", "declared": "Apache-2.0", "retrieved_at": "2026-10-20T03:55:07.420029+00:00", "url": "https://raw.githubusercontent.com/kwai/MobileForge/main/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/kwai/MobileForge"}], "name": "MobileForge", "notes": "The root README describes MobileGym exploration/feedback and HiFPO optimization. The training README limits its public training path to the Qwen3-VL-8B experiment. Linked model and dataset collections, executable source completeness and benchmark results were not independently verified; the repository code license is not a model or dataset license.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/kwai/MobileForge/main/README.md", "target_id": "arxiv:2606.19930", "type": "introduced_by"}], "review_status": "curated", "scope": "core", "summary": "Android GUI-agent adaptation framework combining app exploration, generated curricula and rollout feedback with hint-conditioned step-level GRPO; the documented public training path targets Qwen3-VL-8B.", "topics": ["mobile-navigation", "data-generation", "reinforcement-learning", "evaluation"], "updated_at": null, "url": "https://github.com/kwai/MobileForge"}
{"aliases": [], "authors": [], "details": {"availability": "Public source and minimal-experiment documentation are released; default README path is absent.", "repository_revision": "b1a02a52a6c94b96f63016ac12d3a4ab6a410b8e"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:14:58Z", "role": "primary", "sha256": "74508100843f8db6db6bb151e77ac1cc604a6b7ebd16d3225266d73434465136", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations", "details"], "url": "https://raw.githubusercontent.com/MobileLLM/AutoDroid-V2/HEAD/README_minimal.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:12:10Z", "role": "primary", "sha256": "f349963f2ea3f7b81f1c04ee875654382929b1f93d99d0168d6e086155f41d72", "supports": ["identity"], "url": "https://api.github.com/repos/MobileLLM/AutoDroid-V2"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:14:03Z", "role": "primary", "sha256": "70354b856b75bc4a2aca0a7274321637039ec5d083523ca2f677cd11d475b386", "supports": ["release_status", "licenses.code"], "url": "https://api.github.com/repos/MobileLLM/AutoDroid-V2/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:14:58Z", "role": "primary", "sha256": "8ba37983a2c2da53d96a52696c07e4bdbea4e987a1a07b557935d614c3be46d2", "supports": ["summary", "details", "relations"], "url": "https://raw.githubusercontent.com/MobileLLM/AutoDroid-V2/HEAD/step_3_training/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:14:58Z", "role": "primary", "sha256": "cfed016e038d09ce7c9e829273fffa4a6c618bf8e9043312e4690fbe1f40d247", "supports": ["environments", "details"], "url": "https://raw.githubusercontent.com/MobileLLM/AutoDroid-V2/HEAD/step_0_installation/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:14:58Z", "role": "primary", "sha256": "f5534771f68770ed2e45daf94fbf80304ba2904387571c34c5d80a152bef0a00", "supports": ["summary"], "url": "https://raw.githubusercontent.com/MobileLLM/AutoDroid-V2/HEAD/step_1_doc_generation/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:14:59Z", "role": "primary", "sha256": "0564f566fde97360496bea4a6eb023c4c84e03cdfdf5201c5b27235c36e32b18", "supports": ["summary", "details"], "url": "https://raw.githubusercontent.com/MobileLLM/AutoDroid-V2/HEAD/step_4_accuracy_validation/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "github:mobilellm/autodroid-v2", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-17T09:14:58Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/MobileLLM/AutoDroid-V2"}], "name": "AutoDroid V2", "notes": "The repository uses README_minimal.md rather than README.md and includes a documented App Launcher experiment, training setup and DroidTask/LlamaTouch evaluation scripts. Its supplied emulator setup targets x86_64 Linux and Android API 31. No repository-level code license was found; code, APK, data and weight terms are not inferred from one another. Dependencies, external evaluation files and model service are still required; no software was run.", "published_at": null, "relations": [{"evidence_url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending", "target_id": "arxiv:2412.18116", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "An Android automation implementation that builds app documentation from exploration traces and evaluates language-model-generated execution code in emulator workflows.", "topics": ["mobile-navigation", "tool-integration", "data-generation"], "updated_at": null, "url": "https://github.com/MobileLLM/AutoDroid-V2"}
{"aliases": [], "authors": [], "details": {}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T10:42:34Z", "role": "primary", "sha256": "3095f7262b7d4ff916c6f58c55e6fc9902c08f77d09bf3951199945501148bb3", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/MurtyShikhar/NNetnav/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T10:48:42Z", "role": "primary", "sha256": "c71d239df91726fc519c6eb72d318ec65820627232b2f796219e87dcf35d0ab4", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/MurtyShikhar/NNetnav/HEAD/LICENSE"}], "id": "github:murtyshikhar/nnetnav", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-16T10:42:34Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "Apache-2.0", "retrieved_at": "2026-10-16T10:48:42Z", "url": "https://raw.githubusercontent.com/MurtyShikhar/NNetnav/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/MurtyShikhar/NNetnav"}], "name": "NNetNav", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/MurtyShikhar/NNetnav/HEAD/README.md", "target_id": "arxiv:2410.02907", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Collects and labels browser-exploration trajectories for agent training, with code for data processing, fine-tuning, and web evaluation.", "topics": ["data-generation", "web-navigation"], "updated_at": null, "url": "https://github.com/MurtyShikhar/NNetnav"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/neulab/multiui/HEAD/README.md", "population": "reported MultiUI training collection", "qualifier": "approximately", "unit": "samples", "value": 7300000}], "evidence_urls": ["https://raw.githubusercontent.com/neulab/multiui/HEAD/README.md"], "formats": [], "modalities": ["webpage UI content", "text", "accessibility-derived structure"], "notes": "The README provides a rounded sample count. It does not establish a unique-screenshot inventory or a normalized split breakdown. Synthetic examples and publicly sourced web content have provenance and quality limits; data terms must be checked separately from repository code.", "splits": ["training collection"], "supervision": ["UI understanding", "text-rich visual tasks"]}, "modalities": ["screenshot", "instruction"], "reported_scale": {"comparison": "approximately", "unit": "samples", "value": 7300000}, "sample_count": null, "sample_unit": "instructions"}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:01Z", "role": "primary", "sha256": "fe0c03fe73529a29788d559103c4aed48dca6ed9ba3bfd5f151f902fa8858b38", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://raw.githubusercontent.com/neulab/multiui/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:01Z", "role": "primary", "sha256": "fe0c03fe73529a29788d559103c4aed48dca6ed9ba3bfd5f151f902fa8858b38", "supports": ["details.dataset_profile"], "url": "https://raw.githubusercontent.com/neulab/multiui/HEAD/README.md"}], "id": "github:neulab/multiui", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T06:36:01Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/neulab/multiui"}], "name": "MultiUI", "notes": "", "published_at": null, "relations": [], "review_status": "curated", "scope": "core", "summary": "7.3 million training samples built from webpage UIs and structured with enhanced accessibility trees, for text-rich visual understanding, released with UIX-Qwen2 models.", "topics": ["screen-understanding", "data-generation"], "updated_at": null, "url": "https://github.com/neulab/multiui"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://huggingface.co/datasets/hflqf88888/GUIOdyssey/raw/main/README.md", "population": "recommended hflqf88888/GUIOdyssey release", "qualifier": "exact", "unit": "episodes", "value": 8334}], "evidence_urls": ["https://raw.githubusercontent.com/OpenGVLab/GUI-Odyssey/HEAD/README.md", "https://huggingface.co/datasets/hflqf88888/GUIOdyssey/raw/main/README.md", "https://raw.githubusercontent.com/OpenGVLab/GUI-Odyssey/HEAD/introduction.md"], "formats": ["JSON annotations", "PNG screenshots"], "modalities": ["screenshots", "instructions", "actions", "bounding boxes", "step context"], "notes": "The repository gives conflicting episode totals (8,334 and 8,834). This profile uses 8,334 from the specifically recommended current dataset card. No independent archive recount was performed. Coordinates are normalized to 0–1000.", "splits": ["random", "task", "device", "app"], "supervision": ["cross-app trajectories", "action grounding"]}}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:01Z", "role": "primary", "sha256": "4585b765d7c8b3e9bb6187d36d576edde454d4c0158a6aa8b3fdbedf34c89a15", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details.dataset_profile"], "url": "https://raw.githubusercontent.com/OpenGVLab/GUI-Odyssey/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T11:39:52Z", "role": "primary", "sha256": "fd627ee0a6ee0596e0ee8147996b9a7a3436f754fc95ed63cf5f408d778b2831", "supports": ["details.dataset_profile"], "url": "https://huggingface.co/datasets/hflqf88888/GUIOdyssey/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T11:39:52Z", "role": "primary", "sha256": "78c19ee712487af60df1913f11572958564e59a1a2cc663b1b30d971855d7a28", "supports": ["details.dataset_profile"], "url": "https://raw.githubusercontent.com/OpenGVLab/GUI-Odyssey/HEAD/introduction.md"}], "id": "github:opengvlab/gui-odyssey", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T06:36:01Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/OpenGVLab/GUI-Odyssey"}], "name": "GUI-Odyssey", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/OpenGVLab/GUI-Odyssey/HEAD/README.md", "target_id": "arxiv:2406.08451", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Human-demonstrated Android episodes for cross-app navigation, spanning six task types, 212 apps and about 1,400 app combinations, for training and evaluating agents.", "topics": ["cross-app", "data-generation"], "updated_at": null, "url": "https://github.com/OpenGVLab/GUI-Odyssey"}
{"aliases": [], "authors": [], "details": {}, "environments": ["cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:35:59Z", "role": "primary", "sha256": "0538701c860afedd5d63c3839039699d8e5a14977d5c339b187ed13ad666c8fa", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/OpenGVLab/ScaleCUA/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T06:42:00Z", "role": "primary", "sha256": "c71d239df91726fc519c6eb72d318ec65820627232b2f796219e87dcf35d0ab4", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/OpenGVLab/ScaleCUA/HEAD/LICENSE"}], "id": "github:opengvlab/scalecua", "identifiers": {}, "kind": "model", "last_verified_at": "2026-10-16T06:35:59Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "Apache-2.0", "retrieved_at": "2026-10-16T06:42:00Z", "url": "https://raw.githubusercontent.com/OpenGVLab/ScaleCUA/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/OpenGVLab/ScaleCUA"}], "name": "ScaleCUA", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/OpenGVLab/ScaleCUA/HEAD/README.md", "target_id": "arxiv:2509.15221", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Computer-use model family trained with cross-platform interaction data.", "topics": ["data-generation", "planning"], "updated_at": null, "url": "https://github.com/OpenGVLab/ScaleCUA"}
{"aliases": [], "authors": [], "details": {}, "environments": ["cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:04Z", "role": "primary", "sha256": "033d2de815e11c208862c5c5601b78d58668b79cadd37a750cad9edd4dacabed", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/OpenGVLab/ZeroGUI/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T06:42:05Z", "role": "primary", "sha256": "c71d239df91726fc519c6eb72d318ec65820627232b2f796219e87dcf35d0ab4", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/OpenGVLab/ZeroGUI/HEAD/LICENSE"}], "id": "github:opengvlab/zerogui", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-16T06:36:04Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "Apache-2.0", "retrieved_at": "2026-10-16T06:42:05Z", "url": "https://raw.githubusercontent.com/OpenGVLab/ZeroGUI/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/OpenGVLab/ZeroGUI"}], "name": "ZeroGUI", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/OpenGVLab/ZeroGUI/HEAD/README.md", "target_id": "arxiv:2505.23762", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Online RL framework for GUI agents in which VLMs generate the training tasks and judge success from trajectory screenshots, with OSWorld and AndroidLab checkpoints trained from UI-TARS-7B-DPO.", "topics": ["reinforcement-learning", "data-generation"], "updated_at": null, "url": "https://github.com/OpenGVLab/ZeroGUI"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [], "evidence_urls": ["https://raw.githubusercontent.com/OS-Copilot/OS-Genesis/HEAD/README.md", "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"], "formats": ["JSONL trajectories", "JSON transition triples"], "modalities": ["screenshots", "text", "actions"], "notes": "The repository links distinct AndroidControl-oriented, AndroidWorld-oriented, and web training files plus raw pre-state/action/post-state triples. These are training artifacts, not a documented train/test split, and related representations are not disjoint corpus totals. No numerical inventory is asserted from release names or linked archives.", "splits": [], "supervision": ["reverse task synthesis", "GUI action trajectories"]}}, "environments": ["cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:01Z", "role": "primary", "sha256": "430e495ab250395d666854bd18b0fc29b819cd784cd7730215ef65e461834d0d", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://raw.githubusercontent.com/OS-Copilot/OS-Genesis/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:01Z", "role": "primary", "sha256": "430e495ab250395d666854bd18b0fc29b819cd784cd7730215ef65e461834d0d", "supports": ["details.dataset_profile"], "url": "https://raw.githubusercontent.com/OS-Copilot/OS-Genesis/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["details.dataset_profile"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "github:os-copilot/os-genesis", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T06:36:01Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/OS-Copilot/OS-Genesis"}], "name": "OS-Genesis", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/OS-Copilot/OS-Genesis/HEAD/README.md", "target_id": "arxiv:2412.19723", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Interaction-driven pipeline that synthesizes mobile and web GUI agent trajectories without human supervision or predefined tasks, via reverse task synthesis and a trajectory reward model.", "topics": ["data-generation"], "updated_at": null, "url": "https://github.com/OS-Copilot/OS-Genesis"}
{"aliases": [], "authors": [], "details": {}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:44:04Z", "role": "primary", "sha256": "e06dbc9d58e713a0019d5040aa21220de85266e80aacc7f2d320a0518c781496", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/OSU-NLP-Group/SkillWeaver/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:44:04Z", "role": "primary", "sha256": "4dfa648e22431d1a4903eb1e11f432fbe8a0f39d24e2e3324c300285ffe458fd", "supports": ["identity"], "url": "https://api.github.com/repos/OSU-NLP-Group/SkillWeaver"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:44:04Z", "role": "primary", "sha256": "7a1c94c67131a97516d92632c00a2a20d09226fec57bb09e77ff6d84a4e6e79a", "supports": ["release_status"], "url": "https://api.github.com/repos/OSU-NLP-Group/SkillWeaver/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:45:20Z", "role": "primary", "sha256": "93b43d692b033b76129504448695bfe76ef22d18b11e51352bfab4b5622e5aaa", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/OSU-NLP-Group/SkillWeaver/HEAD/LICENSE"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:45:21Z", "role": "primary", "sha256": "822a60969166874dd8a26273b8d749ef8c69c46137a0915ed7178b3265bd037f", "supports": ["relations"], "url": "https://osu-nlp-group.github.io/SkillWeaver/"}, {"method": "license_header", "retrieved_at": "2026-10-17T02:45:20Z", "role": "primary", "sha256": "93b43d692b033b76129504448695bfe76ef22d18b11e51352bfab4b5622e5aaa", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/OSU-NLP-Group/SkillWeaver/HEAD/LICENSE"}], "id": "github:osu-nlp-group/skillweaver", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-17T02:44:04Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "MIT", "retrieved_at": "2026-10-17T02:45:20Z", "url": "https://raw.githubusercontent.com/OSU-NLP-Group/SkillWeaver/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/OSU-NLP-Group/SkillWeaver"}], "name": "SkillWeaver", "notes": "Exploration, task execution, browser integration and evaluation code are published under MIT. Synthesized skill files are included, but no trained model weights are implied. The source depends on external web environments and model services; no runtime or browser tasks were executed.", "published_at": null, "relations": [{"evidence_url": "https://osu-nlp-group.github.io/SkillWeaver/", "target_id": "arxiv:2504.07079", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "A web-agent framework that explores sites, synthesizes reusable interaction APIs and builds a skill library for later tasks.", "topics": ["web-navigation", "tool-integration", "data-generation"], "updated_at": null, "url": "https://github.com/OSU-NLP-Group/SkillWeaver"}
{"aliases": [], "authors": [], "details": {}, "environments": ["desktop", "web", "cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T01:46:05Z", "role": "primary", "sha256": "a495bb4c86cf54cd5e2276433cfecb401fd133287406b1a7e04c3eaaac6305ac", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/PrentisAI/GroundGUI/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-17T01:52:51Z", "role": "primary", "sha256": "c71d239df91726fc519c6eb72d318ec65820627232b2f796219e87dcf35d0ab4", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/PrentisAI/GroundGUI/HEAD/LICENSE"}], "id": "github:prentisai/groundgui", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-17T01:46:05Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "Apache-2.0", "retrieved_at": "2026-10-17T01:52:51Z", "url": "https://raw.githubusercontent.com/PrentisAI/GroundGUI/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/PrentisAI/GroundGUI"}], "name": "GroundGUI training and evaluation code", "notes": "The repository documents separate training-stage environments. Code, model weights, and ScreenRef data have distinct license declarations; no training or benchmark execution was reproduced here.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/PrentisAI/GroundGUI/HEAD/README.md", "target_id": "arxiv:2610.19100", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Code for screenshot grounding through supervised training, teacher-guided distillation, reinforcement learning, and benchmark evaluation.", "topics": ["grounding", "reinforcement-learning", "data-generation"], "updated_at": null, "url": "https://github.com/PrentisAI/GroundGUI"}
{"aliases": [], "authors": [], "details": {}, "environments": ["android", "ios", "cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:59:55Z", "role": "primary", "sha256": "cb3d2329571a80b6edeab98c04f8b63c7f4e976929788d6a8283238b0b7fdadc", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "notes"], "url": "https://raw.githubusercontent.com/runamu/monday/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-18T11:03:47Z", "role": "primary", "sha256": "c71d239df91726fc519c6eb72d318ec65820627232b2f796219e87dcf35d0ab4", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/runamu/monday/HEAD/LICENSE"}], "id": "github:runamu/monday", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-18T10:59:55Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "Apache-2.0", "retrieved_at": "2026-10-18T11:03:47Z", "url": "https://raw.githubusercontent.com/runamu/monday/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/runamu/monday"}], "name": "MONDAY", "notes": "The retained README marks release of the full data-generation pipeline, training scripts, and model checkpoints as pending. Linked video content and datasets remain subject to their original licenses; the repository code license does not establish a blanket data or weights license.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/runamu/monday/HEAD/README.md", "target_id": "arxiv:2505.12632", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Cross-platform mobile GUI trajectory dataset derived from instructional videos, with released data-processing helpers and example evaluation code for Android and iOS data.", "topics": ["data-generation", "mobile-navigation", "video"], "updated_at": null, "url": "https://github.com/runamu/monday"}
{"aliases": ["GUI-Narrator"], "authors": [], "details": {"dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://export.arxiv.org/api/query?id_list=2406.12814%2C2406.13719%2C2406.14056%2C2406.14250%2C2406.14314%2C2406.18082%2C2406.19263%2C2407.00993%2C2407.01476%2C2407.01489%2C2407.01511%2C2407.01558%2C2407.03037%2C2407.03913%2C2407.04346%2C2407.05291%2C2407.07061%2C2407.08850%2C2407.09018%2C2407.09295%2C2407.10956%2C2407.13032%2C2407.13063%2C2407.15673%2C2407.15711%2C2407.16741%2C2407.17490%2C2407.18901%2C2407.19056%2C2407.20183%2C2408.00203%2C2408.02544%2C2408.06327%2C2408.07199%2C2408.11824%2C2408.15978%2C2409.02453%2C2409.04566%2C2409.07429%2C2409.08264%2C2409.09354%2C2409.10741%2C2409.11295%2C2409.13711%2C2409.14337%2C2409.14818%2C2409.15441%2C2409.15637%2C2409.17140%2C2409.20566&max_results=50", "population": "Act2Cap benchmark described in the paper", "qualifier": "exact", "reference_url": "https://arxiv.org/abs/2406.13719", "unit": "video-captioning samples", "value": 4189}], "evidence_urls": ["https://github.com/showlab/GUI-Action-Narrator", "https://export.arxiv.org/api/query?id_list=2406.12814%2C2406.13719%2C2406.14056%2C2406.14250%2C2406.14314%2C2406.18082%2C2406.19263%2C2407.00993%2C2407.01476%2C2407.01489%2C2407.01511%2C2407.01558%2C2407.03037%2C2407.03913%2C2407.04346%2C2407.05291%2C2407.07061%2C2407.08850%2C2407.09018%2C2407.09295%2C2407.10956%2C2407.13032%2C2407.13063%2C2407.15673%2C2407.15711%2C2407.16741%2C2407.17490%2C2407.18901%2C2407.19056%2C2407.20183%2C2408.00203%2C2408.02544%2C2408.06327%2C2408.07199%2C2408.11824%2C2408.15978%2C2409.02453%2C2409.04566%2C2409.07429%2C2409.08264%2C2409.09354%2C2409.10741%2C2409.11295%2C2409.13711%2C2409.14337%2C2409.14818%2C2409.15441%2C2409.15637%2C2409.17140%2C2409.20566&max_results=50"], "formats": [], "modalities": ["GUI video", "action captions"], "notes": "A captioning sample is not necessarily a unique video or complete task trajectory. Package layout and split sizes were not established by the inspected sources.", "splits": [], "supervision": ["GUI action captioning"]}}, "environments": ["cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:40:52Z", "role": "primary", "sha256": "379aae1a8890a0c3d3b5999589b8338338cf1ea63883c2b77df678fd3bb367a2", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details.dataset_profile"], "url": "https://github.com/showlab/GUI-Action-Narrator"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:32Z", "role": "primary", "sha256": "b9a83994d8ac36ccdb606a5219511f8bb35ad3948ade4f9f3e28d386221aedfb", "supports": ["details.dataset_profile"], "url": "https://export.arxiv.org/api/query?id_list=2406.12814%2C2406.13719%2C2406.14056%2C2406.14250%2C2406.14314%2C2406.18082%2C2406.19263%2C2407.00993%2C2407.01476%2C2407.01489%2C2407.01511%2C2407.01558%2C2407.03037%2C2407.03913%2C2407.04346%2C2407.05291%2C2407.07061%2C2407.08850%2C2407.09018%2C2407.09295%2C2407.10956%2C2407.13032%2C2407.13063%2C2407.15673%2C2407.15711%2C2407.16741%2C2407.17490%2C2407.18901%2C2407.19056%2C2407.20183%2C2408.00203%2C2408.02544%2C2408.06327%2C2408.07199%2C2408.11824%2C2408.15978%2C2409.02453%2C2409.04566%2C2409.07429%2C2409.08264%2C2409.09354%2C2409.10741%2C2409.11295%2C2409.13711%2C2409.14337%2C2409.14818%2C2409.15441%2C2409.15637%2C2409.17140%2C2409.20566&max_results=50"}], "id": "github:showlab/gui-action-narrator", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T06:40:52Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/showlab/GUI-Action-Narrator"}], "name": "GUI-Action Narrator", "notes": "Original repository URL redirects to showlab/GUI-Narrator.", "published_at": null, "relations": [], "review_status": "curated", "scope": "core", "summary": "Action narration resources for GUI interaction trajectories.", "topics": ["data-generation", "video"], "updated_at": null, "url": "https://github.com/showlab/GUI-Action-Narrator"}
{"aliases": [], "authors": [], "details": {}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:06Z", "role": "primary", "sha256": "feee31aa4d0801cb364578099a13e4f9cdcaa4eb5fbf4312b3fc3f8edd479a14", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/SqueezeAILab/plan-and-act/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:05Z", "role": "primary", "sha256": "eada6ee9aa180f55daca059c5e7a4922b320cccb0514864108eba9040b52415c", "supports": ["identity"], "url": "https://api.github.com/repos/SqueezeAILab/plan-and-act"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:05Z", "role": "primary", "sha256": "10ff0a1c1abe25454541949fdaac1869d68b5d4426fbb5ef1d487657afcd26f5", "supports": ["release_status"], "url": "https://api.github.com/repos/SqueezeAILab/plan-and-act/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:15Z", "role": "primary", "sha256": "f86e78202a51ab013e1cc47ba37386134601317f5a0c4e4ed287981b52159122", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/SqueezeAILab/plan-and-act/HEAD/LICENSE"}, {"method": "license_header", "retrieved_at": "2026-10-17T02:02:15Z", "role": "primary", "sha256": "f86e78202a51ab013e1cc47ba37386134601317f5a0c4e4ed287981b52159122", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/SqueezeAILab/plan-and-act/HEAD/LICENSE"}], "id": "github:squeezeailab/plan-and-act", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-17T02:02:06Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "MIT", "retrieved_at": "2026-10-17T02:02:15Z", "url": "https://raw.githubusercontent.com/SqueezeAILab/plan-and-act/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/SqueezeAILab/plan-and-act"}], "name": "Plan-and-Act", "notes": "Runnable entry points, framework modules and synthetic-data tooling are published. The repository code declares MIT; separate model and dataset cards have their own declarations. Its setup targets WebArena-Lite and requires that external environment.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/SqueezeAILab/plan-and-act/HEAD/README.md", "target_id": "arxiv:2503.09572", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "A web-agent framework that separates high-level planning from action execution, with replanning and synthetic planner-data generation components.", "topics": ["planning", "web-navigation", "data-generation"], "updated_at": null, "url": "https://github.com/SqueezeAILab/plan-and-act"}
{"aliases": [], "authors": [], "details": {"license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/sunblaze-ucb/AgentSynth/HEAD/LICENSE"], "reason": "Reviewed repository license declaration for code; third-party dependencies, data, weights and executable environment images are separate."}}, "repository_tree_sha": "cd42a4bd158fe3f0a1b0f376b9428d6a5fcb81b2", "repository_url": "https://github.com/sunblaze-ucb/AgentSynth"}, "environments": ["desktop", "web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:27:12Z", "role": "primary", "sha256": "42e8b8dead7caea3f8b21e93a075e8c895dfc30a25e1470bf97a90940bbf0106", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations"], "url": "https://raw.githubusercontent.com/sunblaze-ucb/AgentSynth/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T15:37:00Z", "role": "primary", "sha256": "e08f08a99b8cb3a4bb857a9fac976b2820f9ad406d66675b0286e880dfeeceaa", "supports": ["identity"], "url": "https://api.github.com/repos/sunblaze-ucb/AgentSynth"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:25:20Z", "role": "primary", "sha256": "b0998652a9b6b465444c2bbeef015bdbef50e83984a3bd15725f0e811645ca62", "supports": ["identity", "release_status", "details.repository_tree_sha"], "url": "https://api.github.com/repos/sunblaze-ucb/AgentSynth/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:35:19Z", "role": "primary", "sha256": "c71d239df91726fc519c6eb72d318ec65820627232b2f796219e87dcf35d0ab4", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/sunblaze-ucb/AgentSynth/HEAD/LICENSE"}], "id": "github:sunblaze-ucb/agentsynth", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-18T04:27:12Z", "licenses": {"code": {"basis": "Reviewed repository license declaration for code; third-party dependencies, data, weights and executable environment images are separate.", "declared": "Apache-2.0", "retrieved_at": "2026-10-18T04:35:19Z", "url": "https://raw.githubusercontent.com/sunblaze-ucb/AgentSynth/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/sunblaze-ucb/AgentSynth"}], "name": "AgentSynth task-generation pipeline", "notes": "Released code includes task-generation and training-format conversion scripts. OSWorld/InSTA environments and model services require separate setup; the repository is not a bundled VM or model checkpoint. Code and the separately published dataset have distinct license declarations.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/sunblaze-ucb/AgentSynth/HEAD/README.md", "target_id": "arxiv:2506.14205", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Generates synthetic computer-use tasks and records trajectories in OSWorld desktop environments and InSTA browser environments.", "topics": ["data-generation", "long-horizon", "automation"], "updated_at": null, "url": "https://github.com/sunblaze-ucb/AgentSynth"}
{"aliases": [], "authors": [], "details": {}, "environments": ["desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T11:09:28Z", "role": "primary", "sha256": "d4a68b00acc4becf2a25688520595babea9b694b56541e709f066413b5b80aea", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/THUDM/SCALE-CUA/HEAD/README.md"}], "id": "github:thudm/scale-cua", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-16T11:09:28Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/THUDM/SCALE-CUA"}], "name": "SCALE-CUA (THUDM)", "notes": "A distinct project from OpenGVLab/ScaleCUA, which documents arXiv:2509.15221. Similar names do not establish a shared project identity.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/THUDM/SCALE-CUA/HEAD/README.md", "target_id": "arxiv:2607.11185", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Computer-use training framework with VeriGen task synthesis, adaptive rollout sampling, and visual-context segmentation.", "topics": ["reinforcement-learning", "data-generation", "efficiency"], "updated_at": null, "url": "https://github.com/THUDM/SCALE-CUA"}
{"aliases": [], "authors": [], "details": {}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:06Z", "role": "primary", "sha256": "eb217b96c54c3c2ccfd3a27255c6fd740df40c2d6743ae811858339bf7f740b0", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/THUDM/WebRL/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:05Z", "role": "primary", "sha256": "42496cf0cff06a7f9e794f4313a10a670bc48df9909ec6d57763cb902bb229ea", "supports": ["identity"], "url": "https://api.github.com/repos/THUDM/WebRL"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:05Z", "role": "primary", "sha256": "bb1683c6fb5ff658c89f0ebf0f8baf8026bb703840e174c9e46b223c85a16856", "supports": ["release_status"], "url": "https://api.github.com/repos/THUDM/WebRL/git/trees/main?recursive=1"}], "id": "github:thudm/webrl", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-17T02:02:06Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/THUDM/WebRL"}], "name": "WebRL", "notes": "Training code and checkpoint links are released. No root code-license declaration was found in the inspected repository metadata; bundled dependency licenses do not establish a license for the whole project. Weights have separate Hub terms.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/THUDM/WebRL/HEAD/README.md", "target_id": "arxiv:2411.02337", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Training pipeline that evolves web tasks from prior attempts and uses an outcome reward model to refine browser policies.", "topics": ["reinforcement-learning", "web-navigation", "data-generation"], "updated_at": null, "url": "https://github.com/THUDM/WebRL"}
{"aliases": [], "authors": [], "details": {"availability": "Public training configurations, preprocessing and inference examples; full experiment-script release incomplete.", "license_review": {"code": {"evidence_urls": ["https://api.github.com/repos/TongUI-agent/TongUI-agent/git/trees/main?recursive=1"], "reason": "No project-level source-license declaration in the inspected tree; upstream dependency licenses do not establish this project’s grant."}}, "repository_revision": "82631b583180cdc870b4774d2495dd94f1558c46"}, "environments": ["cross-platform", "desktop", "web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:30:54Z", "role": "primary", "sha256": "f30a18d1a533b8f71f436dae2ee7369f8971422bd76527e069499c5bba911912", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations"], "url": "https://raw.githubusercontent.com/TongUI-agent/TongUI-agent/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:30:53Z", "role": "primary", "sha256": "90c209a9758f7b527def072cfa78ebc87f1b49b436fad57c4c7bb5a6b66ae790", "supports": ["identity"], "url": "https://api.github.com/repos/TongUI-agent/TongUI-agent"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:32:05Z", "role": "primary", "sha256": "0c955a89dcda18652b09ce836c51c6ccaa007be6f6afeb48d21d01813cef1ec2", "supports": ["release_status", "details.repository_revision", "licenses.code"], "url": "https://api.github.com/repos/TongUI-agent/TongUI-agent/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:36:55Z", "role": "primary", "sha256": "b51acb70c95d079ccbe1de0a685ad7c8a4b3b5bed11e6162e6f7c2f60d78805e", "supports": ["summary", "details", "relations"], "url": "https://raw.githubusercontent.com/TongUI-agent/TongUI-agent/HEAD/docs/train.md"}], "id": "github:tongui-agent/tongui-agent", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-17T12:30:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/TongUI-agent/TongUI-agent"}], "name": "TongUI training and inference framework", "notes": "The public tree includes training configs, preprocessing, inference and serving examples. The project still marks release of all experiment/evaluation scripts as work in progress. This entry covers the implementation rather than counting its 3B/7B/32B model-size variants as separate new projects. The inspected project tree has no root source license; linked frameworks, data and checkpoints retain their own terms.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/TongUI-agent/TongUI-agent/HEAD/docs/train.md", "target_id": "url:8c9aaafb918aaf8b4e92", "type": "uses"}, {"evidence_url": "https://raw.githubusercontent.com/TongUI-agent/TongUI-agent/HEAD/README.md", "target_id": "arxiv:2504.12679", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Training configurations, preprocessing and inference examples for tutorial-trained GUI models, with a LLaMA-Factory fine-tuning workflow and screenshot-grounding interfaces.", "topics": ["grounding", "imitation-learning", "data-generation"], "updated_at": null, "url": "https://github.com/TongUI-agent/TongUI-agent"}
{"aliases": [], "authors": [], "details": {}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:02Z", "role": "primary", "sha256": "2c67c8b181e7196f3ca1e88888ffc5d6b32afca686a0e5c4353ad3f5af51ff9e", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/tsinghua-fib-lab/FingerTip-20K/HEAD/README.md"}], "id": "github:tsinghua-fib-lab/fingertip-20k", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T06:36:02Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/tsinghua-fib-lab/FingerTip-20K"}], "name": "FingerTip-20K", "notes": "", "published_at": null, "relations": [], "review_status": "curated", "scope": "core", "summary": "Over 20,000 human Android episodes across 500+ apps with user context and long-term usage patterns, plus tracks for proactive intent prediction and personalized execution.", "topics": ["data-generation"], "updated_at": null, "url": "https://github.com/tsinghua-fib-lab/FingerTip-20K"}
{"aliases": [], "authors": [], "details": {}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:06Z", "role": "primary", "sha256": "cfe9d8614b0b0558c37ba7c50371f6c47f119b914061ae815a9e3d796dd9ccfa", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/UbiquitousLearning/DroidCall/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:01:50Z", "role": "primary", "sha256": "7a4ca3daa6d011ff2d24d993191882d2b5545d8476c7eddf396581c0fef77496", "supports": ["identity"], "url": "https://api.github.com/repos/UbiquitousLearning/DroidCall"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:06Z", "role": "primary", "sha256": "d0717d706359969376889eac73ea8764b659f41a2b4f3c1a409110267a912526", "supports": ["release_status"], "url": "https://api.github.com/repos/UbiquitousLearning/DroidCall/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:16Z", "role": "primary", "sha256": "2832cc5333f7166924a2823aa38a286d918b39b7fef2ca7af6521766c68a5303", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/UbiquitousLearning/DroidCall/HEAD/LICENSE"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:28Z", "role": "primary", "sha256": "08a6c5cfa655e1ecd774a41a9a815e845ad0afd47b289378bc1d6ee66c81dda9", "supports": ["relations"], "url": "https://huggingface.co/datasets/mllmTeam/DroidCall/raw/main/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-17T02:02:16Z", "role": "primary", "sha256": "2832cc5333f7166924a2823aa38a286d918b39b7fef2ca7af6521766c68a5303", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/UbiquitousLearning/DroidCall/HEAD/LICENSE"}], "id": "github:ubiquitouslearning/droidcall", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-17T02:02:06Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "MIT", "retrieved_at": "2026-10-17T02:02:16Z", "url": "https://raw.githubusercontent.com/UbiquitousLearning/DroidCall/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/UbiquitousLearning/DroidCall"}], "name": "DroidCall generation and evaluation code", "notes": "Source utilities are published under MIT. This component operates through Android intent functions and is marked adjacent to visual GUI agents. Training data is separately released under its own card declaration; no checkpoint release is inferred from the fine-tuning instructions.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/mllmTeam/DroidCall/raw/main/README.md", "target_id": "arxiv:2412.00402", "type": "documented_by"}], "review_status": "curated", "scope": "adjacent", "summary": "Data-generation, fine-tuning and evaluation utilities for teaching small language models to invoke Android intents through predefined functions.", "topics": ["tool-integration", "data-generation", "evaluation"], "updated_at": null, "url": "https://github.com/UbiquitousLearning/DroidCall"}
{"aliases": [], "authors": [], "details": {}, "environments": ["desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:01Z", "role": "primary", "sha256": "91cffd585c667d169486506a785c6cbc3068fe28409efc97abf8911f9d0bd5a5", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/UITron-hub/TreeCUA/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T06:42:01Z", "role": "primary", "sha256": "c71d239df91726fc519c6eb72d318ec65820627232b2f796219e87dcf35d0ab4", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/UITron-hub/TreeCUA/HEAD/LICENSE"}], "id": "github:uitron-hub/treecua", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-16T06:36:01Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "Apache-2.0", "retrieved_at": "2026-10-16T06:42:01Z", "url": "https://raw.githubusercontent.com/UITron-hub/TreeCUA/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/UITron-hub/TreeCUA"}], "name": "TreeCUA", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/UITron-hub/TreeCUA/HEAD/README.md", "target_id": "arxiv:2602.09662", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Framework for generating GUI trajectories through tree-structured exploration and verification.", "topics": ["data-generation"], "updated_at": null, "url": "https://github.com/UITron-hub/TreeCUA"}
{"aliases": [], "authors": [], "details": {"availability": "Public synthesis and evaluation source; model checkpoints not verified.", "repository_revision": "1ec25af942a46ce46f6c00d596332e308198342f"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:18:20Z", "role": "primary", "sha256": "9a46741bbde1373ed4b87887cdadde98e05ea8795d34efe39fc25d69b4d50213", "supports": ["name", "kind", "summary", "topics", "scope", "notes", "relations"], "url": "https://raw.githubusercontent.com/Wuzheng02/GUI-CIDER/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:18:20Z", "role": "primary", "sha256": "ae1d5687a72a39eacbf4fe281b5048914e98adf7cc1517481a5acc8a56a81446", "supports": ["identity"], "url": "https://api.github.com/repos/Wuzheng02/GUI-CIDER"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:22:59Z", "role": "primary", "sha256": "355128b62a281074eebc14a1e3b85a260f06c2bf3c1a6ed140cba3b9aefaa76c", "supports": ["release_status", "details.repository_revision", "licenses.code"], "url": "https://api.github.com/repos/Wuzheng02/GUI-CIDER/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:24:47Z", "role": "primary", "sha256": "bffca456e649a08239f6a1a3be10a7d7c21bcb427e783ba2c304e4b8410d385a", "supports": ["environments", "summary", "notes"], "url": "https://raw.githubusercontent.com/Wuzheng02/GUI-CIDER/HEAD/data/dataset.md"}], "id": "github:wuzheng02/gui-cider", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-17T11:18:20Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/Wuzheng02/GUI-CIDER"}], "name": "GUI-CIDER research code", "notes": "The source tree contains static/causal knowledge synthesis and action/knowledge evaluation scripts. Training procedures are delegated to LLaMA-Factory; no trained checkpoint is verified here. The README displays an MIT badge, but the inspected tree has no license file and no explicit component scope, so the code license remains unrecorded rather than assuming a blanket grant. Evaluation requires separately acquired original datasets and model paths.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/Wuzheng02/GUI-CIDER/HEAD/README.md", "target_id": "arxiv:2605.28534", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Released data-synthesis and evaluation scripts that distill planning and causal GUI knowledge into text for mid-training mobile interface agents.", "topics": ["data-generation", "imitation-learning", "planning"], "updated_at": null, "url": "https://github.com/Wuzheng02/GUI-CIDER"}
{"aliases": [], "authors": [], "details": {}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:01Z", "role": "primary", "sha256": "17fa475b96bdd00270c656072d3f8fc55d8433f0e3a1c2762110160ec21a5c35", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/xlang-ai/AgentTrek/HEAD/README.md"}], "id": "github:xlang-ai/agenttrek", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T06:36:01Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/xlang-ai/AgentTrek"}], "name": "AgentTrek", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/xlang-ai/AgentTrek/HEAD/README.md", "target_id": "arxiv:2412.09605", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Synthesizes browser-agent trajectories by replaying web tutorials as guidance, releasing the trajectory data with multimodal grounding and reasoning and an AgentTrek-32B agent.", "topics": ["data-generation", "video"], "updated_at": null, "url": "https://github.com/xlang-ai/AgentTrek"}
{"aliases": [], "authors": [], "details": {}, "environments": ["desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:35:59Z", "role": "primary", "sha256": "91fed11b234775bde91bb766ee65e1015897ced6668de8e80418fcee09f8f75d", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/xlang-ai/OpenCUA/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T06:42:00Z", "role": "primary", "sha256": "a53fdf6cdedc93c99879b2881d8626d61b09e9a03841b3dc6ebe4e267f4fb2bf", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/xlang-ai/OpenCUA/HEAD/LICENSE"}], "id": "github:xlang-ai/opencua", "identifiers": {}, "kind": "model", "last_verified_at": "2026-10-16T06:35:59Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "MIT", "retrieved_at": "2026-10-16T06:42:00Z", "url": "https://raw.githubusercontent.com/xlang-ai/OpenCUA/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/xlang-ai/OpenCUA"}], "name": "OpenCUA", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/xlang-ai/OpenCUA/HEAD/README.md", "target_id": "arxiv:2508.09123", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Open computer-use research stack including models and training resources.", "topics": ["data-generation", "planning"], "updated_at": null, "url": "https://github.com/xlang-ai/OpenCUA"}
{"aliases": [], "authors": [], "details": {}, "environments": ["cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:01Z", "role": "primary", "sha256": "0968273442bfb9a6478fe78f8b9a6247a3bb8e6d028880939b432fd009961e98", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/yiye3/GUICourse/HEAD/README.md"}], "id": "github:yiye3/guicourse", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T06:36:01Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/yiye3/GUICourse"}], "name": "GUICourse", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/yiye3/GUICourse/HEAD/README.md", "target_id": "arxiv:2406.11317", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Three datasets for turning general VLMs into GUI agents: GUIEnv for OCR and grounding, GUIAct with 67k single-step and 15k multi-step web and Android actions, and GUIChat dialogues.", "topics": ["data-generation"], "updated_at": null, "url": "https://github.com/yiye3/GUICourse"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending", "population": "AMEX collection described in the paper", "qualifier": "more-than", "reference_url": "https://arxiv.org/abs/2407.17490", "unit": "screenshots", "value": 104000}], "evidence_urls": ["https://raw.githubusercontent.com/YuxiangChai/AMEX-codebase/HEAD/README.md", "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"], "formats": [], "modalities": ["screenshots", "element annotations", "functionality descriptions", "action chains"], "notes": "The main repository contains companion training code. The scale and annotation levels here come from the paper, not an independent download or completeness audit.", "splits": [], "supervision": ["element grounding", "UI functionality", "interaction trajectories"]}}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:01Z", "role": "primary", "sha256": "3184bfd612543621e57226e9079132dc8d5695339795a60366d3f6d445e13551", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details.dataset_profile"], "url": "https://raw.githubusercontent.com/YuxiangChai/AMEX-codebase/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["details.dataset_profile", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:27:11Z", "role": "primary", "sha256": "4aab61e6a95ab6b6598fb64b8e55d787cd49b5c4f242d15372aed2dd312f97aa", "supports": ["relations"], "url": "https://yxchai.com/AMEX/"}, {"method": "license_header", "retrieved_at": "2026-10-16T06:42:02Z", "role": "primary", "sha256": "c71d239df91726fc519c6eb72d318ec65820627232b2f796219e87dcf35d0ab4", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/YuxiangChai/AMEX-codebase/HEAD/LICENSE"}], "id": "github:yuxiangchai/amex-codebase", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T06:36:01Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "Apache-2.0", "retrieved_at": "2026-10-16T06:42:02Z", "url": "https://raw.githubusercontent.com/YuxiangChai/AMEX-codebase/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/YuxiangChai/AMEX-codebase"}], "name": "AMEX", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://yxchai.com/AMEX/", "target_id": "arxiv:2407.17490", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Code for AMEX Android GUI data that converts its three annotation levels into question-answer pairs and fine-tunes a LLaMA2-Accessory model, with a released checkpoint.", "topics": ["data-generation"], "updated_at": null, "url": "https://github.com/YuxiangChai/AMEX-codebase"}
{"aliases": [], "authors": [], "details": {}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T13:17:23Z", "role": "primary", "sha256": "fc3009b355e46549c5128c933bb333aa37b1db5ac8ea8f2d0f300848d8a62200", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/ZJU-ACES-ISE/UIDefectMocker/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T13:29:25Z", "role": "primary", "sha256": "2b75b184f28b204d8e2c7b45a341e4583202cf4c9da8964f108770929d1193ba", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/ZJU-ACES-ISE/UIDefectMocker/HEAD/LICENSE"}], "id": "github:zju-aces-ise/uidefectmocker", "identifiers": {}, "kind": "tool", "last_verified_at": "2026-10-16T13:17:23Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "MIT", "retrieved_at": "2026-10-16T13:29:25Z", "url": "https://raw.githubusercontent.com/ZJU-ACES-ISE/UIDefectMocker/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/ZJU-ACES-ISE/UIDefectMocker"}], "name": "UI Defect Mocker", "notes": "This entry describes the released screenshot-defect tool, separately from the under-construction ChatUITest project hub.", "published_at": null, "relations": [], "review_status": "curated", "scope": "core", "summary": "Synthesizes visual UI defects from mobile screenshots and Android layout metadata to support GUI-testing experiments.", "topics": ["data-generation", "evaluation", "screen-understanding"], "updated_at": null, "url": "https://github.com/ZJU-ACES-ISE/UIDefectMocker"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [], "evidence_urls": ["https://huggingface.co/datasets/xlangai/aguvis-stage1/raw/main/README.md", "https://huggingface.co/api/datasets/xlangai/aguvis-stage1", "https://raw.githubusercontent.com/xlang-ai/aguvis/HEAD/README.md", "https://raw.githubusercontent.com/xlang-ai/aguvis/HEAD/data/stage1.yaml"], "formats": ["JSON annotations", "ZIP image packages", "multipart TAR image packages"], "modalities": ["GUI screenshots", "grounding annotations"], "notes": "Stage 1 grounding release, Hub revision 5d11f8f5. JSON annotations and ZIP/multipart TAR image packages are publicly listed. The official training configuration groups web, desktop and mobile inputs and selects all rows of its configured inputs; this is a training recipe, not a release sample count or held-out split. The config names omniact.json while the Hub manifest lists omniact_fix.json, so local paths need reconciliation. No sample total is inferred from names such as webui350k. Apache-2.0 is the release card declaration; upstream source rights remain separate.", "splits": ["training"], "supervision": ["GUI grounding supervised fine-tuning"]}, "disabled": false, "download_url": "https://huggingface.co/datasets/xlangai/aguvis-stage1/tree/main", "gated": false, "hub_id": "xlangai/aguvis-stage1"}, "environments": ["cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:03:33Z", "role": "primary", "sha256": "38bda302cf741b9d2b2394b53d157b3c19d8a73e3704c01ed2f9a27c374feb77", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details", "details.dataset_profile"], "url": "https://huggingface.co/datasets/xlangai/aguvis-stage1/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:03:33Z", "role": "primary", "sha256": "cb3b8ec88d811d946df3cbf38ef555db5a10d1124668765710c380e9b1988b06", "supports": ["identity", "release_status", "details", "licenses.data", "details.dataset_profile"], "url": "https://huggingface.co/api/datasets/xlangai/aguvis-stage1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:00Z", "role": "primary", "sha256": "2dd9f36b818c1b7b6bd772ba8575f6c0ebab6d9e0d307edcc2f196f3e4ba502e", "supports": ["environments", "summary", "details.dataset_profile"], "url": "https://raw.githubusercontent.com/xlang-ai/aguvis/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T14:21:11Z", "role": "primary", "sha256": "f6a9e55c5d48469e2c8f7799d917d25e7a758036f71c7c045fa92c35057d20da", "supports": ["details.dataset_profile"], "url": "https://raw.githubusercontent.com/xlang-ai/aguvis/HEAD/data/stage1.yaml"}, {"method": "huggingface_card", "retrieved_at": "2026-10-17T02:03:33Z", "role": "primary", "sha256": "38bda302cf741b9d2b2394b53d157b3c19d8a73e3704c01ed2f9a27c374feb77", "supports": ["licenses.data"], "url": "https://huggingface.co/datasets/xlangai/aguvis-stage1/raw/main/README.md"}], "id": "url:06dd5ac7917c45d48509", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-17T02:03:33Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card metadata; third-party assets retain their own terms", "declared": "apache-2.0", "retrieved_at": "2026-10-17T02:03:33Z", "url": "https://huggingface.co/datasets/xlangai/aguvis-stage1/raw/main/README.md"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/xlangai/aguvis-stage1"}], "name": "Aguvis stage 1 grounding data", "notes": "The manifest contains grounding JSON annotations and packaged images. The card declares Apache-2.0, which is recorded for this release without overriding source-dataset rights. This first-stage grounding collection is distinct from the project’s second-stage trajectory collection. No archive or training sample was downloaded.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/xlangai/aguvis-stage1/raw/main/README.md", "target_id": "arxiv:2412.04454", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "The first-stage Aguvis training collection, packaging GUI element-grounding annotations and image archives from multiple interface-data sources.", "topics": ["grounding", "data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/xlangai/aguvis-stage1"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/yiye3/GUICourse/HEAD/README.md", "population": "GUIEnv-local SFT data", "qualifier": "approximately", "unit": "region-text QA pairs", "value": 700000}], "evidence_urls": ["https://raw.githubusercontent.com/yiye3/GUICourse/HEAD/README.md"], "formats": ["JSON", "Parquet"], "modalities": ["screenshots", "text", "bounding boxes"], "notes": "Quantities are rounded publisher descriptions, not downloaded sample counts. Data-license declarations conflict across first-party sources; no permissive license is inferred.", "splits": ["train", "test"], "supervision": ["OCR", "region-text grounding"]}, "disabled": false, "download_url": "https://huggingface.co/datasets/yiye2023/GUIEnv/tree/main", "gated": false, "hub_id": "yiye2023/GUIEnv", "license_review": {"data": {"evidence_urls": ["https://raw.githubusercontent.com/yiye3/GUICourse/HEAD/README.md", "https://huggingface.co/datasets/yiye2023/GUIEnv/raw/main/README.md"], "reason": "Conflicting first-party dataset declarations: the GUICourse project README assigns CC-BY-4.0 to the datasets, while this exact Hub card declares Apache-2.0. The code MIT declaration does not resolve the data conflict."}}}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:01Z", "role": "primary", "sha256": "0968273442bfb9a6478fe78f8b9a6247a3bb8e6d028880939b432fd009961e98", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details", "licenses.data"], "url": "https://raw.githubusercontent.com/yiye3/GUICourse/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:01:33Z", "role": "primary", "sha256": "15813be05dc76aed5f667875188bf9612e74ff2f4699d270dd0d1aeda86767f8", "supports": ["identity", "release_status", "details"], "url": "https://huggingface.co/api/datasets/yiye2023/GUIEnv"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:01:33Z", "role": "primary", "sha256": "4bcf87ecfbbb8e07a01b21415a970c8b53a5283bf6872b657040d3f45c9241f7", "supports": ["licenses.data"], "url": "https://huggingface.co/datasets/yiye2023/GUIEnv/raw/main/README.md"}], "id": "url:0af07e00658a430cee6a", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T06:36:01Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/yiye2023/GUIEnv"}], "name": "GUIEnv-local", "notes": "Public JSON annotations and Parquet image packages are present. The project README declares CC-BY-4.0 for datasets, while this Hub card declares Apache-2.0; these conflicting declarations are retained without asserting a single confirmed data license. The project code’s MIT declaration is separate. Only GUIEnv-local is marked released; the advertised 10M-example GUIEnv-global pretraining collection is still unchecked and is not counted here.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/yiye3/GUICourse/HEAD/README.md", "target_id": "arxiv:2406.11317", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "The released local-supervision portion of GUIEnv, pairing webpage regions and text for OCR and GUI grounding.", "topics": ["grounding", "data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/yiye2023/GUIEnv"}
{"aliases": [], "authors": [], "details": {"disabled": false, "download_url": "https://huggingface.co/datasets/vyokky/GUI-360/tree/main", "gated": false, "hub_id": "vyokky/GUI-360", "languages": [], "modalities": [], "size_category_reported": ["1M<n<10M"]}, "environments": ["desktop", "windows"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:38:26Z", "role": "primary", "sha256": "e51af2845c2f31c7af112579b42c3ff9bf5a2c2356e80932e829bc2be1a341bd", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://huggingface.co/api/datasets/vyokky/GUI-360"}, {"method": "huggingface_api", "retrieved_at": "2026-10-16T06:38:26Z", "role": "primary", "sha256": "e51af2845c2f31c7af112579b42c3ff9bf5a2c2356e80932e829bc2be1a341bd", "supports": ["details", "licenses.data"], "url": "https://huggingface.co/api/datasets/vyokky/GUI-360"}], "id": "url:0b590d03a3d5103a79e4", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T06:38:26Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card", "declared": "mit", "retrieved_at": "2026-10-16T06:38:26Z", "url": "https://huggingface.co/datasets/vyokky/GUI-360"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/vyokky/GUI-360"}], "name": "GUI-360", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/api/datasets/vyokky/GUI-360", "target_id": "arxiv:2511.04307", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Desktop GUI trajectory data supporting multiple interface understanding tasks.", "topics": ["grounding", "data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/vyokky/GUI-360"}
{"aliases": [], "authors": [], "details": {"disabled": false, "download_url": "https://huggingface.co/datasets/SenseLLM/GUI-Lasagne-L1/tree/main", "gated": false, "hub_id": "SenseLLM/GUI-Lasagne-L1"}, "environments": ["unspecified"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:28Z", "role": "primary", "sha256": "270b3af0904a2a50d040948100a327458839dffa5083fc74bb030298f5ab3985", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/datasets/SenseLLM/GUI-Lasagne-L1/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:28Z", "role": "primary", "sha256": "224d6487de47cc63251c3b5c5c9f6e75e692b5ef0e12f0a7f71000da0b866e6f", "supports": ["identity", "release_status", "details", "licenses.data"], "url": "https://huggingface.co/api/datasets/SenseLLM/GUI-Lasagne-L1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:28Z", "role": "primary", "sha256": "916bde91ff00aba22ff13ed3d8b4ca679bc8c198632f1d3bc9469660287cb5d3", "supports": ["environments", "summary"], "url": "https://huggingface.co/SenseLLM/SpiritSight-Agent-8B/raw/main/README.md"}, {"method": "huggingface_card", "retrieved_at": "2026-10-17T02:02:28Z", "role": "primary", "sha256": "270b3af0904a2a50d040948100a327458839dffa5083fc74bb030298f5ab3985", "supports": ["licenses.data"], "url": "https://huggingface.co/datasets/SenseLLM/GUI-Lasagne-L1/raw/main/README.md"}], "id": "url:0c5659ca9e59d942aea3", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-17T02:02:28Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card metadata; third-party assets retain their own terms", "declared": "apache-2.0", "retrieved_at": "2026-10-17T02:02:28Z", "url": "https://huggingface.co/datasets/SenseLLM/GUI-Lasagne-L1/raw/main/README.md"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/SenseLLM/GUI-Lasagne-L1"}], "name": "GUI-Lasagne L1", "notes": "Public archive shards are present. The dataset card declares Apache-2.0; it does not resolve rights in third-party interface content. The short card supplies no reliable sample count or explicit train/test partition, so these remain unspecified. The dataset card and archive inventory do not establish its platform distribution; the associated model's platform coverage is not inherited. An associated SpiritSight model card still describes dataset release as forthcoming; the independent dataset card and public archive manifest establish this release's availability.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/SenseLLM/GUI-Lasagne-L1/raw/main/README.md", "target_id": "arxiv:2503.03196", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "The published GUI-Lasagne training-data collection used for SpiritSight, distributed as dated archive shards for visual interface learning.", "topics": ["grounding", "screen-understanding", "data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/SenseLLM/GUI-Lasagne-L1"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://huggingface.co/datasets/OS-Copilot/OS-Atlas-data/raw/main/README.md", "population": "SeeClick-derived web subset", "qualifier": "more-than", "unit": "screenshots", "value": 270000}, {"basis": "reported", "evidence_url": "https://huggingface.co/datasets/OS-Copilot/OS-Atlas-data/raw/main/README.md", "population": "SeeClick-derived web subset", "qualifier": "more-than", "unit": "UI elements", "value": 3000000}, {"basis": "reported", "evidence_url": "https://huggingface.co/datasets/OS-Copilot/OS-Atlas-data/raw/main/README.md", "population": "FineWeb-derived subset", "qualifier": "at-least", "unit": "images", "value": 1600000}], "evidence_urls": ["https://huggingface.co/datasets/OS-Copilot/OS-Atlas-data/raw/main/README.md"], "formats": ["JSON annotations", "image archives"], "modalities": ["screenshots", "referring expressions", "bounding boxes"], "notes": "These are subset-specific counts, not a total for the full OS-Atlas release. Raw boxes use 0–1 coordinates; the documented training wrapper scales them to 0–1000.", "splits": [], "supervision": ["element grounding"]}, "disabled": false, "download_url": "https://huggingface.co/datasets/OS-Copilot/OS-Atlas-data/tree/main", "gated": false, "hub_id": "OS-Copilot/OS-Atlas-data", "languages": [], "modalities": [], "size_category_reported": null}, "environments": ["cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:38:26Z", "role": "primary", "sha256": "c2c5f7b3f07d92349ffd8256792599c856ed77c89108c45dbaef48a4699abe20", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://huggingface.co/api/datasets/OS-Copilot/OS-Atlas-data"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T11:40:46Z", "role": "primary", "sha256": "19f28b4bc0798e28180b968010c0f7c8dfed2dbfb24b122b958476b871d2e4fd", "supports": ["details.dataset_profile"], "url": "https://huggingface.co/datasets/OS-Copilot/OS-Atlas-data/raw/main/README.md"}, {"method": "huggingface_api", "retrieved_at": "2026-10-16T06:38:26Z", "role": "primary", "sha256": "c2c5f7b3f07d92349ffd8256792599c856ed77c89108c45dbaef48a4699abe20", "supports": ["details", "licenses.data"], "url": "https://huggingface.co/api/datasets/OS-Copilot/OS-Atlas-data"}], "id": "url:0fdeb1cf5f49f84820e6", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T06:38:26Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card", "declared": "apache-2.0", "retrieved_at": "2026-10-16T06:38:26Z", "url": "https://huggingface.co/datasets/OS-Copilot/OS-Atlas-data"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/OS-Copilot/OS-Atlas-data"}], "name": "OS-Atlas Data", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/api/datasets/OS-Copilot/OS-Atlas-data", "target_id": "arxiv:2410.23218", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "GUI grounding pre-training data for OS-Atlas in desktop (Linux, macOS, Windows), mobile and web domains, with element boxes stored as relative left, top, right and bottom ratios.", "topics": ["grounding", "data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/OS-Copilot/OS-Atlas-data"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [], "evidence_urls": ["https://raw.githubusercontent.com/GAIR-NLP/PC-Agent-E/HEAD/README.md", "https://huggingface.co/api/datasets/henryhe0123/PC-Agent-E"], "formats": ["ZIP package"], "modalities": ["GUI screenshots", "action trajectories", "generated reasoning"], "notes": "Public manifest exposes data.zip; the internal file schema and number of examples remain unverified. Supervision describes the publisher’s refinement/thought-completion/trajectory-boost pipeline, not an independently sampled data-quality audit.", "splits": ["training"], "supervision": ["human demonstration refinement", "generated action reasoning", "trajectory augmentation"]}, "disabled": false, "download_url": "https://huggingface.co/datasets/henryhe0123/PC-Agent-E/tree/4f64c7b055154bed9ffea677aa86aeb51a01cb73", "gated": false, "hub_id": "henryhe0123/PC-Agent-E", "hub_revision": "4f64c7b055154bed9ffea677aa86aeb51a01cb73", "license_review": {"data": {"evidence_urls": ["https://huggingface.co/datasets/henryhe0123/PC-Agent-E/raw/4f64c7b055154bed9ffea677aa86aeb51a01cb73/README.md"], "reason": "Publisher dataset-card declaration for this release; it does not establish rights to all upstream screenshots, applications or source content."}}, "repository_url": "https://github.com/GAIR-NLP/PC-Agent-E"}, "environments": ["desktop", "windows", "linux"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:27:13Z", "role": "primary", "sha256": "3ffb8c843c6acb8c2d5bbba11ba49d6b36cd566553a89974b6252a2bfd289d5d", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations", "details.dataset_profile"], "url": "https://raw.githubusercontent.com/GAIR-NLP/PC-Agent-E/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:33:21Z", "role": "primary", "sha256": "66f70ee2704cdf2e8847cfe04be05d9e14b3a3c1bee953e744f8209d6c9c2b85", "supports": ["identity", "release_status", "details.hub_id", "details.hub_revision", "details.gated", "details.disabled", "details.download_url", "details.dataset_profile"], "url": "https://huggingface.co/api/datasets/henryhe0123/PC-Agent-E"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:35:17Z", "role": "primary", "sha256": "4dfd1f3ec40f7ff3c94896955216cb60a25114b320caff29e909d296d375d6ef", "supports": ["relations", "licenses.data"], "url": "https://huggingface.co/datasets/henryhe0123/PC-Agent-E/raw/4f64c7b055154bed9ffea677aa86aeb51a01cb73/README.md"}], "id": "url:14449fe321e0a1ca34ea", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-18T04:27:13Z", "licenses": {"code": null, "data": {"basis": "Publisher dataset-card declaration for this release; it does not establish rights to all upstream screenshots, applications or source content.", "declared": "MIT", "retrieved_at": "2026-10-18T04:35:17Z", "url": "https://huggingface.co/datasets/henryhe0123/PC-Agent-E/raw/4f64c7b055154bed9ffea677aa86aeb51a01cb73/README.md"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/henryhe0123/PC-Agent-E"}], "name": "PC-Agent-E training data", "notes": "Hub release contains data.zip. The archive was not downloaded and its internal schema/example total were not independently inspected. Training-purpose and platform descriptions come from the official pipeline README; the brief card establishes the paper association and MIT data declaration.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/GAIR-NLP/PC-Agent-E/HEAD/README.md", "target_id": "arxiv:2505.13909", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Released dataset used to train PC-Agent-E, linked by the official pipeline for refining human trajectories and synthesizing action reasoning.", "topics": ["data-generation", "imitation-learning"], "updated_at": null, "url": "https://huggingface.co/datasets/henryhe0123/PC-Agent-E"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/UI-S1/README.md", "population": "public example subset", "qualifier": "exact", "unit": "training examples", "value": 1000}], "evidence_urls": ["https://huggingface.co/datasets/mPLUG/UI_S1_dataset/raw/main/README.md", "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/UI-S1/README.md"], "formats": ["JSONL", "Parquet"], "modalities": ["images", "text", "actions"], "notes": "The count is explicitly reported by the publisher, not inferred from the filename. Full training data remains outside this release.", "splits": ["training example subset"], "supervision": ["GUI actions"]}, "disabled": false, "download_url": "https://huggingface.co/datasets/mPLUG/UI_S1_dataset/tree/main", "gated": false, "hub_id": "mPLUG/UI_S1_dataset"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:07:19Z", "role": "primary", "sha256": "d826463e2aa159f083dc74869bfabb8abb5b8ea4f2cdf4aeb7676172b18aaaa5", "supports": ["name", "kind", "relations", "licenses.data"], "url": "https://huggingface.co/datasets/mPLUG/UI_S1_dataset/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:07:19Z", "role": "primary", "sha256": "cdfbcbf23a3b1b4a609e1943e80dcf144ca1224a74ebf2308fde4fcb4326ab3c", "supports": ["identity", "release_status", "details", "licenses.data"], "url": "https://huggingface.co/api/datasets/mPLUG/UI_S1_dataset"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:06:57Z", "role": "primary", "sha256": "f668c890a5a87953a243715d0d47367c23f0c99b524d80f48bbdde8163f3e942", "supports": ["summary", "environments", "details.dataset_profile", "topics", "scope"], "url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/UI-S1/README.md"}, {"method": "huggingface_card", "retrieved_at": "2026-10-17T04:07:19Z", "role": "primary", "sha256": "d826463e2aa159f083dc74869bfabb8abb5b8ea4f2cdf4aeb7676172b18aaaa5", "supports": ["licenses.data"], "url": "https://huggingface.co/datasets/mPLUG/UI_S1_dataset/raw/main/README.md"}], "id": "url:16184830c0f4bdb146fa", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-17T04:07:19Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card metadata; third-party assets retain their own terms", "declared": "apache-2.0", "retrieved_at": "2026-10-17T04:07:19Z", "url": "https://huggingface.co/datasets/mPLUG/UI_S1_dataset/raw/main/README.md"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/mPLUG/UI_S1_dataset"}], "name": "UI-S1 released training examples", "notes": "The release contains train_1000.jsonl and images.parquet. The official README explicitly describes 1,000 released examples and calls the release only part of the training dataset. Apache-2.0 is declared by the dataset card; this is not the complete training corpus.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/mPLUG/UI_S1_dataset/raw/main/README.md", "target_id": "arxiv:2509.11543", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "A public example subset of the UI-S1 mobile training data, packaged as JSONL annotations and Parquet images.", "topics": ["reinforcement-learning", "data-generation", "mobile-navigation"], "updated_at": null, "url": "https://huggingface.co/datasets/mPLUG/UI_S1_dataset"}
{"aliases": [], "authors": [], "details": {"disabled": false, "download_url": "https://huggingface.co/datasets/stanfordnlp/nnetnav-wa/tree/main", "gated": false, "hub_id": "stanfordnlp/nnetnav-wa", "languages": [], "modalities": ["text"], "size_category_reported": null}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T11:11:53Z", "role": "primary", "sha256": "3fe40a45aa2bf9d6f0cefffc65ea6382849c3543b44017200a4a8d43ab61e9eb", "supports": ["name", "kind"], "url": "https://huggingface.co/api/datasets/stanfordnlp/nnetnav-wa"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T10:42:34Z", "role": "primary", "sha256": "3095f7262b7d4ff916c6f58c55e6fc9902c08f77d09bf3951199945501148bb3", "supports": ["name", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/MurtyShikhar/NNetnav/HEAD/README.md"}, {"method": "huggingface_api", "retrieved_at": "2026-10-16T11:11:53Z", "role": "primary", "sha256": "3fe40a45aa2bf9d6f0cefffc65ea6382849c3543b44017200a4a8d43ab61e9eb", "supports": ["details", "licenses.data"], "url": "https://huggingface.co/api/datasets/stanfordnlp/nnetnav-wa"}], "id": "url:25a4d0ab29b60250863a", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T11:11:53Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card", "declared": "apache-2.0", "retrieved_at": "2026-10-16T11:11:53Z", "url": "https://huggingface.co/datasets/stanfordnlp/nnetnav-wa"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/stanfordnlp/nnetnav-wa"}], "name": "NNetNav WebArena data", "notes": "The Hub card supplies a license but no detailed schema. Dataset purpose is supported by the project's README; exact sample counts and annotation quality remain unverified.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/MurtyShikhar/NNetnav/HEAD/README.md", "target_id": "github:murtyshikhar/nnetnav", "type": "component_of"}], "review_status": "curated", "scope": "core", "summary": "NNetNav's WebArena-oriented browser-interaction data release, with separate training and test files.", "topics": ["data-generation", "web-navigation"], "updated_at": null, "url": "https://huggingface.co/datasets/stanfordnlp/nnetnav-wa"}
{"aliases": ["url:1ff6b78a764bdcc6dd03", "HanXiao1999/UI-Genie-Agent-5k", "https://huggingface.co/datasets/HanXiao1999/UI-Genie-Agent-5k"], "authors": [], "details": {"dataset_profile": {"counts": [], "evidence_urls": ["https://huggingface.co/datasets/HanXiao1999/UI-Genie-Agent-16k/raw/main/README.md", "https://raw.githubusercontent.com/Euphoria16/UI-Genie/HEAD/README.md"], "formats": ["JSONL", "PNG"], "modalities": ["screenshots", "text", "actions"], "notes": "Public JSONL annotations and screenshots are present. MIT is declared in the dataset card. The release name, model-card links and annotation filename use different size labels; this review does not infer a reliable trajectory count from them.", "splits": [], "supervision": ["GUI action trajectories"]}, "disabled": false, "download_url": "https://huggingface.co/datasets/HanXiao1999/UI-Genie-Agent-16k/tree/main", "gated": false, "hub_id": "HanXiao1999/UI-Genie-Agent-16k"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:01:34Z", "role": "primary", "sha256": "a748fc7dbb53d5d751c5ee109e437c377e214b1d837762f8d09995349d86cec9", "supports": ["name", "kind", "relations", "licenses.data"], "url": "https://huggingface.co/datasets/HanXiao1999/UI-Genie-Agent-16k/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:01:33Z", "role": "primary", "sha256": "548df402288350920a26bff2099544bf21835200ee58ff2c657c331fa6200317", "supports": ["identity", "release_status", "details", "licenses.data"], "url": "https://huggingface.co/api/datasets/HanXiao1999/UI-Genie-Agent-16k"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T03:59:54Z", "role": "primary", "sha256": "e08569bd9d8a6ff44537d4fe69243ddfeb8a41ae5c9f87ef84af50a67c64c726", "supports": ["summary", "environments", "topics", "scope"], "url": "https://raw.githubusercontent.com/Euphoria16/UI-Genie/HEAD/README.md"}, {"method": "huggingface_card", "retrieved_at": "2026-10-17T04:01:34Z", "role": "primary", "sha256": "a748fc7dbb53d5d751c5ee109e437c377e214b1d837762f8d09995349d86cec9", "supports": ["licenses.data"], "url": "https://huggingface.co/datasets/HanXiao1999/UI-Genie-Agent-16k/raw/main/README.md"}, {"method": "huggingface_repository_identity", "retrieved_at": "2026-10-17T19:15:09Z", "role": "primary", "sha256": "66006a586e03be3eb1dfc29dc9c951241fd7e4204430e7db258da1037e17ac6e", "supports": ["aliases"], "url": "https://huggingface.co/api/datasets/HanXiao1999/UI-Genie-Agent-5k"}, {"method": "huggingface_repository_identity", "retrieved_at": "2026-10-17T19:15:10Z", "role": "primary", "sha256": "66006a586e03be3eb1dfc29dc9c951241fd7e4204430e7db258da1037e17ac6e", "supports": ["aliases"], "url": "https://huggingface.co/api/datasets/HanXiao1999/UI-Genie-Agent-16k"}], "id": "url:2e1927450da2183a5abe", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-17T04:01:34Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card metadata; third-party assets retain their own terms", "declared": "mit", "retrieved_at": "2026-10-17T04:01:34Z", "url": "https://huggingface.co/datasets/HanXiao1999/UI-Genie-Agent-16k/raw/main/README.md"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/HanXiao1999/UI-Genie-Agent-16k"}], "name": "UI-Genie trajectory data", "notes": "Public JSONL annotations and screenshots are present. MIT is declared in the dataset card. The release name, model-card links and annotation filename use different size labels; this review does not infer a reliable trajectory count from them.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/HanXiao1999/UI-Genie-Agent-16k/raw/main/README.md", "target_id": "arxiv:2505.21496", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Synthetic mobile GUI trajectory supervision and screenshots released with the UI-Genie agent-training pipeline.", "topics": ["data-generation", "mobile-navigation", "imitation-learning"], "updated_at": null, "url": "https://huggingface.co/datasets/HanXiao1999/UI-Genie-Agent-16k"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://huggingface.co/datasets/PrentisAI/GroundGUI_RL/raw/main/README.md", "population": "RL training corpus", "qualifier": "exact", "unit": "prompts", "value": 12600}, {"basis": "reported", "evidence_url": "https://huggingface.co/datasets/PrentisAI/GroundGUI_RL/raw/main/README.md", "population": "RL training corpus", "qualifier": "exact", "unit": "screenshots", "value": 11205}, {"basis": "reported", "evidence_url": "https://huggingface.co/datasets/PrentisAI/GroundGUI_RL/raw/main/README.md", "population": "RL training corpus", "qualifier": "exact", "unit": "desktop applications", "value": 87}], "evidence_urls": ["https://huggingface.co/datasets/PrentisAI/GroundGUI_RL/raw/main/README.md"], "formats": ["Parquet"], "modalities": ["screenshots", "text", "coordinates"], "notes": "Publisher-reported counts cover a selection and reformatting of GroundCUA, not a new disjoint image collection. No independent image-overlap audit was performed.", "splits": ["train"], "supervision": ["instruction-to-click grounding", "target bounding boxes"]}}, "environments": ["desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T01:48:10Z", "role": "primary", "sha256": "34f0025f686d0e48e2baa5244056a4362018aaad77ac54abea7f1f6eb25c4276", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/datasets/PrentisAI/GroundGUI_RL/raw/main/README.md"}, {"method": "huggingface_card", "retrieved_at": "2026-10-17T01:48:10Z", "role": "primary", "sha256": "34f0025f686d0e48e2baa5244056a4362018aaad77ac54abea7f1f6eb25c4276", "supports": ["licenses.data"], "url": "https://huggingface.co/datasets/PrentisAI/GroundGUI_RL/raw/main/README.md"}], "id": "url:4e71dc456c4112eed727", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-17T01:48:10Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card metadata; third-party assets retain their own terms", "declared": "mit", "retrieved_at": "2026-10-17T01:48:10Z", "url": "https://huggingface.co/datasets/PrentisAI/GroundGUI_RL/raw/main/README.md"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/PrentisAI/GroundGUI_RL"}], "name": "GroundGUI RL Dataset", "notes": "The card declares MIT and attributes the source examples to GroundCUA. Its separate claim about UI-Vision overlap is not independently validated by this catalog.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/PrentisAI/GroundGUI_RL/raw/main/README.md", "target_id": "arxiv:2610.19100", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Desktop grounding prompts selected from GroundCUA and formatted for click-based reinforcement learning with target-box rewards.", "topics": ["grounding", "reinforcement-learning", "data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/PrentisAI/GroundGUI_RL"}
{"aliases": [], "authors": [], "details": {"disabled": false, "download_url": "https://huggingface.co/datasets/stanfordnlp/nnetnav-live/tree/main", "gated": false, "hub_id": "stanfordnlp/nnetnav-live", "languages": [], "modalities": ["text"], "size_category_reported": null}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T11:11:53Z", "role": "primary", "sha256": "d8f8571dfc583d3a46d5b3a6a285e77f17057b36a8111d416d711513aa4ecf16", "supports": ["name", "kind"], "url": "https://huggingface.co/api/datasets/stanfordnlp/nnetnav-live"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T10:42:34Z", "role": "primary", "sha256": "3095f7262b7d4ff916c6f58c55e6fc9902c08f77d09bf3951199945501148bb3", "supports": ["name", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/MurtyShikhar/NNetnav/HEAD/README.md"}, {"method": "huggingface_api", "retrieved_at": "2026-10-16T11:11:53Z", "role": "primary", "sha256": "d8f8571dfc583d3a46d5b3a6a285e77f17057b36a8111d416d711513aa4ecf16", "supports": ["details", "licenses.data"], "url": "https://huggingface.co/api/datasets/stanfordnlp/nnetnav-live"}], "id": "url:4f47abde8cd28a1ec1e1", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T11:11:53Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card", "declared": "apache-2.0", "retrieved_at": "2026-10-16T11:11:53Z", "url": "https://huggingface.co/datasets/stanfordnlp/nnetnav-live"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/stanfordnlp/nnetnav-live"}], "name": "NNetNav Live data", "notes": "The Hub card supplies a license but no detailed schema. Dataset purpose is supported by the project's README; exact sample counts and annotation quality remain unverified.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/MurtyShikhar/NNetnav/HEAD/README.md", "target_id": "github:murtyshikhar/nnetnav", "type": "component_of"}], "review_status": "curated", "scope": "core", "summary": "Browser-interaction data released with NNetNav for its live-website setting, with separate training and test files.", "topics": ["data-generation", "web-navigation"], "updated_at": null, "url": "https://huggingface.co/datasets/stanfordnlp/nnetnav-live"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://huggingface.co/datasets/hflqf88888/GUIOdyssey/raw/main/README.md", "population": "recommended hflqf88888/GUIOdyssey release", "qualifier": "exact", "unit": "episodes", "value": 8334}], "evidence_urls": ["https://raw.githubusercontent.com/OpenGVLab/GUI-Odyssey/HEAD/README.md", "https://huggingface.co/datasets/hflqf88888/GUIOdyssey/raw/main/README.md", "https://raw.githubusercontent.com/OpenGVLab/GUI-Odyssey/HEAD/introduction.md"], "formats": ["JSON annotations", "PNG screenshots"], "modalities": ["screenshots", "instructions", "actions", "bounding boxes", "step context"], "notes": "The repository gives conflicting episode totals (8,334 and 8,834). This profile uses 8,334 from the specifically recommended current dataset card. No independent archive recount was performed. Coordinates are normalized to 0–1000.", "splits": ["random", "task", "device", "app"], "supervision": ["cross-app trajectories", "action grounding"]}, "download_url": "https://huggingface.co/datasets/hflqf88888/GUIOdyssey/tree/main", "sample_count": 8334, "sample_unit": "episodes reported by the current release card"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T11:39:52Z", "role": "primary", "sha256": "fd627ee0a6ee0596e0ee8147996b9a7a3436f754fc95ed63cf5f408d778b2831", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/datasets/hflqf88888/GUIOdyssey/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:01Z", "role": "primary", "sha256": "4585b765d7c8b3e9bb6187d36d576edde454d4c0158a6aa8b3fdbedf34c89a15", "supports": ["details.dataset_profile"], "url": "https://raw.githubusercontent.com/OpenGVLab/GUI-Odyssey/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T11:39:52Z", "role": "primary", "sha256": "78c19ee712487af60df1913f11572958564e59a1a2cc663b1b30d971855d7a28", "supports": ["details.dataset_profile"], "url": "https://raw.githubusercontent.com/OpenGVLab/GUI-Odyssey/HEAD/introduction.md"}, {"method": "huggingface_card", "retrieved_at": "2026-10-16T11:39:52Z", "role": "primary", "sha256": "fd627ee0a6ee0596e0ee8147996b9a7a3436f754fc95ed63cf5f408d778b2831", "supports": ["licenses.data"], "url": "https://huggingface.co/datasets/hflqf88888/GUIOdyssey/raw/main/README.md"}], "id": "url:511903ebfac45e6751b2", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T11:39:52Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card metadata; third-party assets retain their own terms", "declared": "cc-by-4.0", "retrieved_at": "2026-10-16T11:39:52Z", "url": "https://huggingface.co/datasets/hflqf88888/GUIOdyssey/raw/main/README.md"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/hflqf88888/GUIOdyssey"}], "name": "GUIOdyssey — current data release", "notes": "The project recommends this release over its earlier OpenGVLab/GUI-Odyssey upload. Episode count follows the current data card; conflicting repository prose is documented in the profile.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/hflqf88888/GUIOdyssey/raw/main/README.md", "target_id": "arxiv:2406.08451", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Cross-app Android interaction episodes with fine-grained step instructions, screen descriptions, action context, and target boxes.", "topics": ["mobile-navigation", "cross-app", "data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/hflqf88888/GUIOdyssey"}
{"aliases": [], "authors": ["Tianlin Shi", "Andrej Karpathy", "Linxi Fan", "Jonathan Hernandez", "Percy Liang"], "details": {"publication": {"authors": ["Tianlin Shi", "Andrej Karpathy", "Linxi Fan", "Jonathan Hernandez", "Percy Liang"], "date_basis": "Publisher citation_publication_date metadata explicitly states 2017/07/17; this is separate from the conference interval.", "date_parts": [2017, 7, 17], "doi": null, "event_dates": "06--11 Aug (conference interval reported in the publisher BibTeX month field)", "evidence_urls": ["https://proceedings.mlr.press/v70/shi17a.html", "https://proceedings.mlr.press/v70/shi17a/shi17a.pdf"], "identity_basis": "No same-work record located in the full active catalog. Similarity screening was not used as automatic identity approval. The raw archive and exclusions are additionally checked during normalization.", "issue": null, "notes": "Publisher citation uses Tianlin Shi and Linxi Fan; the PDF also prints the nicknames Tim and Jim. The ordered publisher author names are retained. DOI not supplied in the retained publisher citation. Publisher metadata supplies the publication date separately from the conference interval.", "pages": "3135-3144", "pdf_url": "https://proceedings.mlr.press/v70/shi17a/shi17a.pdf", "publisher": "PMLR", "series": "Proceedings of Machine Learning Research", "title": "World of Bits: An Open-Domain Platform for Web-Based Agents", "type": "conference-paper", "url": "https://proceedings.mlr.press/v70/shi17a.html", "venue": "Proceedings of the 34th International Conference on Machine Learning", "volume": "70"}}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:53:21Z", "role": "primary", "sha256": "749d2a150e4c2bcde1b219a0ecb86c95546134a057892544e8495b5e45323c44", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.event_dates", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.publisher", "details.publication.series", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue", "details.publication.volume"], "url": "https://proceedings.mlr.press/v70/shi17a.html"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:54:39Z", "role": "primary", "sha256": "087008efb9d056df5fd9fbc724413d6a3f20cb14c7f1597d00bacac53d8342d6", "supports": ["details.publication.notes", "details.publication.pdf_url"], "url": "https://proceedings.mlr.press/v70/shi17a/shi17a.pdf"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:53:21Z", "role": "primary", "sha256": "749d2a150e4c2bcde1b219a0ecb86c95546134a057892544e8495b5e45323c44", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.series", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue", "details.publication.volume", "kind", "name", "scope", "summary", "topics"], "url": "https://proceedings.mlr.press/v70/shi17a.html"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:54:39Z", "role": "primary", "sha256": "087008efb9d056df5fd9fbc724413d6a3f20cb14c7f1597d00bacac53d8342d6", "supports": ["details.publication.notes", "details.publication.pdf_url", "environments"], "url": "https://proceedings.mlr.press/v70/shi17a/shi17a.pdf"}], "id": "url:52a3261eb6a19b402e88", "identifiers": {}, "kind": "paper", "last_verified_at": "2026-10-17T09:53:21Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://proceedings.mlr.press/v70/shi17a.html"}, {"rel": "publication_pdf", "url": "https://proceedings.mlr.press/v70/shi17a/shi17a.pdf"}], "name": "World of Bits: An Open-Domain Platform for Web-Based Agents", "notes": "The annotation was reviewed against retained primary publication sources. Experiments have not been independently reproduced. Publisher citation uses Tianlin Shi and Linxi Fan; the PDF also prints the nicknames Tim and Jim. The ordered publisher author names are retained.", "published_at": "2017-07-17", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces World of Bits, a web-task platform combining human demonstrations, keyboard and mouse actions, and cached HTTP traffic for repeatable task environments. It evaluates agents trained through behavioral cloning and reinforcement learning.", "topics": ["web-navigation", "reinforcement-learning", "evaluation", "data-generation"], "updated_at": null, "url": "https://proceedings.mlr.press/v70/shi17a.html"}
{"aliases": ["url:a778c3411858c38c0f39", "OpenMobile-2/OpenMobile-Data", "https://huggingface.co/datasets/OpenMobile-2/OpenMobile-Data"], "authors": [], "details": {"dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://huggingface.co/datasets/OpenMobile-2/OpenMobile-Data-v2/raw/main/README.md", "population": "SFT collection", "qualifier": "exact", "unit": "trajectories", "value": 11643}, {"basis": "reported", "evidence_url": "https://huggingface.co/datasets/OpenMobile-2/OpenMobile-Data-v2/raw/main/README.md", "population": "SFT collection", "qualifier": "exact", "unit": "decision steps", "value": 118582}, {"basis": "reported", "evidence_url": "https://huggingface.co/datasets/OpenMobile-2/OpenMobile-Data-v2/raw/main/README.md", "population": "RL collection", "qualifier": "exact", "unit": "executable tasks", "value": 2475}], "evidence_urls": ["https://huggingface.co/datasets/OpenMobile-2/OpenMobile-Data-v2/raw/main/README.md"], "formats": ["JSON", "JSONL", "PNG"], "modalities": ["screenshots", "text", "tool calls"], "notes": "These are publisher-reported totals at three distinct grains, not values to sum. Earlier screenshots may be collapsed; each SFT example keeps at most three images. The card and repository disagree on the singular/plural RL filename, so consumers should inspect the actual file manifest.", "splits": ["sft/train", "rl/train"], "supervision": ["next-action supervision", "verifiable task outcomes"]}}, "environments": ["android", "web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T01:48:51Z", "role": "primary", "sha256": "abacf9a10759a9e7a8eeec5a266c96ed83742bd9296a19be95d09c9717e18b7c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/datasets/OpenMobile-2/OpenMobile-Data-v2/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T01:48:10Z", "role": "primary", "sha256": "485279038266d9454df75e27a80a0a0d71c24e90700126d2dcfa0da921b1273d", "supports": ["relations", "environments"], "url": "https://raw.githubusercontent.com/OpenMobile2/OpenMobile2-Code/HEAD/README.md"}, {"method": "huggingface_repository_identity", "retrieved_at": "2026-10-17T19:15:18Z", "role": "primary", "sha256": "14b1211ed709147f4022f311c17303ca70faaffa22be143617eb431d966eec1a", "supports": ["aliases"], "url": "https://huggingface.co/api/datasets/OpenMobile-2/OpenMobile-Data"}, {"method": "huggingface_repository_identity", "retrieved_at": "2026-10-17T19:15:20Z", "role": "primary", "sha256": "14b1211ed709147f4022f311c17303ca70faaffa22be143617eb431d966eec1a", "supports": ["aliases"], "url": "https://huggingface.co/api/datasets/OpenMobile-2/OpenMobile-Data-v2"}], "id": "url:546e6c95825d1318c98d", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-17T01:48:51Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/OpenMobile-2/OpenMobile-Data-v2"}], "name": "OpenMobile-Data-v2", "notes": "The reviewed card contains no explicit license declaration, so data-license status remains unknown. The repository code license is not inherited. The current code README points to the v2 dataset; an older project-page URL was not counted as a separate asset.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/OpenMobile2/OpenMobile2-Code/HEAD/README.md", "target_id": "arxiv:2610.14157", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Mobile-agent supervision from emulator and simulator trajectories, plus executable tasks with starting states and success checks for reinforcement learning.", "topics": ["mobile-navigation", "data-generation", "reinforcement-learning", "tool-integration"], "updated_at": null, "url": "https://huggingface.co/datasets/OpenMobile-2/OpenMobile-Data-v2"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://huggingface.co/datasets/PrentisAI/ScreenRef-Annotations/raw/main/README.md", "population": "ScreenRef image inventory", "qualifier": "exact", "unit": "distinct screenshots", "value": 585981}, {"basis": "reported", "evidence_url": "https://huggingface.co/datasets/PrentisAI/ScreenRef-Annotations/raw/main/README.md", "population": "full recipe", "qualifier": "exact", "unit": "annotation rows", "value": 633369}, {"basis": "reported", "evidence_url": "https://huggingface.co/datasets/PrentisAI/ScreenRef-Annotations/raw/main/README.md", "population": "refusal subset of full recipe", "qualifier": "exact", "unit": "annotation rows", "value": 22475}], "evidence_urls": ["https://huggingface.co/datasets/PrentisAI/ScreenRef-Annotations/raw/main/README.md"], "formats": ["Parquet"], "modalities": ["screenshots", "text", "coordinates"], "notes": "The publisher reports these inventories in the companion annotation card. The refusal subset is included in the full recipe. Screenshot access is gated; no image download or independent corpus recount was performed.", "splits": ["train"], "supervision": ["instruction-to-click grounding", "bounding boxes", "absent-target abstention"]}, "disabled": false, "download_url": "https://huggingface.co/datasets/PrentisAI/ScreenRef/tree/main", "gated": "auto", "hub_id": "PrentisAI/ScreenRef", "languages": ["en"], "license_name": "restricted-research-only", "modalities": ["image", "text"], "size_category_reported": ["100K<n<1M"]}, "environments": ["desktop", "web", "android", "linux"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T01:48:51Z", "role": "primary", "sha256": "634b479c7d089e50735f6d4a25f78a63f4d0c2a8c1955015e7641b7afd61c317", "supports": ["name", "kind", "details.gated", "details.hub_id", "licenses.data"], "url": "https://huggingface.co/api/datasets/PrentisAI/ScreenRef"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T01:48:10Z", "role": "primary", "sha256": "96a87222645c3f823e6f67e50708792e2917ddeb8eea5aacdf2de8c5bbff165f", "supports": ["summary", "environments", "topics", "scope", "notes", "relations", "details.dataset_profile"], "url": "https://huggingface.co/datasets/PrentisAI/ScreenRef-Annotations/raw/main/README.md"}, {"method": "huggingface_api", "retrieved_at": "2026-10-17T01:48:51Z", "role": "primary", "sha256": "634b479c7d089e50735f6d4a25f78a63f4d0c2a8c1955015e7641b7afd61c317", "supports": ["details", "licenses.data"], "url": "https://huggingface.co/api/datasets/PrentisAI/ScreenRef"}], "id": "url:56bda04bd8d3a438deb3", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-17T01:48:51Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card", "declared": "other", "retrieved_at": "2026-10-17T01:48:51Z", "url": "https://huggingface.co/datasets/PrentisAI/ScreenRef"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/PrentisAI/ScreenRef"}], "name": "ScreenRef", "notes": "The public Hub API reports gated=auto and a Restricted Research-Only declaration. The companion card distinguishes annotations from screenshots and limits reuse. Access terms were inspected as public metadata; no access agreement was accepted or gated files downloaded.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/PrentisAI/ScreenRef-Annotations/raw/main/README.md", "target_id": "arxiv:2610.19100", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "GUI screenshots paired with localization instructions and targets, including absent-target examples; image access is gated under publisher terms.", "topics": ["grounding", "data-generation", "screen-understanding"], "updated_at": null, "url": "https://huggingface.co/datasets/PrentisAI/ScreenRef"}
{"aliases": [], "authors": [], "details": {"availability": "Ungated Parquet annotation files for train, validation, test and test_unseen_os; videos external.", "dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://huggingface.co/datasets/runamu/MONDAY/raw/main/README.md", "population": "train split", "qualifier": "exact", "unit": "video annotation records", "value": 19725}, {"basis": "reported", "evidence_url": "https://huggingface.co/datasets/runamu/MONDAY/raw/main/README.md", "population": "validation split", "qualifier": "exact", "unit": "video annotation records", "value": 495}, {"basis": "reported", "evidence_url": "https://huggingface.co/datasets/runamu/MONDAY/raw/main/README.md", "population": "test split", "qualifier": "exact", "unit": "video annotation records", "value": 100}, {"basis": "reported", "evidence_url": "https://huggingface.co/datasets/runamu/MONDAY/raw/main/README.md", "population": "test_unseen_os split", "qualifier": "exact", "unit": "video annotation records", "value": 50}, {"basis": "reported", "evidence_url": "https://huggingface.co/datasets/runamu/MONDAY/raw/main/README.md", "population": "publisher overview of MONDAY", "qualifier": "approximately", "unit": "detected scenes", "value": 333000}, {"basis": "reported", "evidence_url": "https://huggingface.co/datasets/runamu/MONDAY/raw/main/README.md", "population": "publisher overview of MONDAY", "qualifier": "approximately", "unit": "identified actions", "value": 313000}], "evidence_urls": ["https://huggingface.co/datasets/runamu/MONDAY/raw/main/README.md", "https://raw.githubusercontent.com/runamu/monday/HEAD/README.md"], "formats": ["Parquet"], "modalities": ["text", "timestamps", "bounding boxes", "action labels"], "notes": "Split counts are video-annotation rows; scene and action totals are rounded publisher reports and are not added to row counts. Source videos/images are not included in the Hub files.", "splits": ["train", "validation", "test", "test_unseen_os"], "supervision": ["video-derived scene and action annotations"]}, "license_review": {"data": {"evidence_urls": ["https://huggingface.co/datasets/runamu/MONDAY/raw/main/README.md", "https://raw.githubusercontent.com/runamu/monday/HEAD/README.md", "https://raw.githubusercontent.com/runamu/monday/HEAD/LICENSE"], "reason": "No dataset-license declaration in the inspected card; Apache-2.0 applies to companion code and is not transferred to YouTube content or this annotation package."}}, "release_revision": "d7bd929cf629bf0767f154cbdd984eea1eba01e9"}, "environments": ["android", "ios"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:32:07Z", "role": "primary", "sha256": "62a346fd0d328deb5b56a652c8581850b0c8120f5c77e53b33f24b9baf937adb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations", "details.dataset_profile", "licenses.data"], "url": "https://huggingface.co/datasets/runamu/MONDAY/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:32:07Z", "role": "primary", "sha256": "f55c319a70c8ff988049363dc8fc266a9f1b1a09b8ac7096ee573f962e8f9492", "supports": ["identity", "release_status", "details.release_revision"], "url": "https://huggingface.co/api/datasets/runamu/MONDAY"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:30:53Z", "role": "primary", "sha256": "cb3d2329571a80b6edeab98c04f8b63c7f4e976929788d6a8283238b0b7fdadc", "supports": ["notes", "details.dataset_profile", "licenses.data"], "url": "https://raw.githubusercontent.com/runamu/monday/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:34:09Z", "role": "primary", "sha256": "c71d239df91726fc519c6eb72d318ec65820627232b2f796219e87dcf35d0ab4", "supports": ["notes", "licenses.data"], "url": "https://raw.githubusercontent.com/runamu/monday/HEAD/LICENSE"}], "id": "url:570654ff67ecefa2b955", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-17T12:32:07Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/runamu/MONDAY"}], "name": "MONDAY annotations", "notes": "The ungated Hub publishes four Parquet splits containing video identifiers and annotations. Original videos and extracted screenshots must be obtained separately; the code README explains that consumers download source videos from YouTube. A Windows Mobile unseen-OS test split is documented but is not labeled desktop Windows here. The companion code has an Apache-2.0 file, while the data card declares no data license and the repository defers third-party materials to original terms. Training scripts, pretrained checkpoints and the full generation pipeline remain planned updates.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/runamu/MONDAY/raw/main/README.md", "target_id": "arxiv:2505.12632", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Video-derived mobile navigation annotations that pair scene boundaries and screen or element boxes with action labels for training and evaluating GUI agents.", "topics": ["data-generation", "mobile-navigation", "video", "imitation-learning"], "updated_at": null, "url": "https://huggingface.co/datasets/runamu/MONDAY"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://huggingface.co/datasets/sunblaze-ucb/AgentSynth/raw/9ec3f3139f9786e81aee14b7fff1eb487cc7e991/README.md", "population": "collection described in the dataset card; not an independently counted release-row total", "qualifier": "more-than", "unit": "tasks", "value": 6000}], "evidence_urls": ["https://huggingface.co/datasets/sunblaze-ucb/AgentSynth/raw/9ec3f3139f9786e81aee14b7fff1eb487cc7e991/README.md", "https://huggingface.co/api/datasets/sunblaze-ucb/AgentSynth"], "formats": ["JSONL", "ZIP trajectory packages"], "modalities": ["screenshots", "task instructions", "LLM thoughts", "commands"], "notes": "Card maps osworld/train to all_tasks.jsonl and insta/train to task_sequences.jsonl. Six difficulty levels and trajectory screenshots/thoughts/commands are documented. ZIP file labels are not trajectory counts; configurations and composed-task levels are not assumed disjoint.", "splits": ["train (osworld configuration)", "train (insta configuration)"], "supervision": ["synthetic tasks", "agent trajectories"]}, "disabled": false, "download_url": "https://huggingface.co/datasets/sunblaze-ucb/AgentSynth/tree/9ec3f3139f9786e81aee14b7fff1eb487cc7e991", "gated": false, "hub_id": "sunblaze-ucb/AgentSynth", "hub_revision": "9ec3f3139f9786e81aee14b7fff1eb487cc7e991", "license_review": {"data": {"evidence_urls": ["https://huggingface.co/datasets/sunblaze-ucb/AgentSynth/raw/9ec3f3139f9786e81aee14b7fff1eb487cc7e991/README.md"], "reason": "Publisher dataset-card declaration for this release; it does not establish rights to all upstream screenshots, applications or source content."}}, "repository_url": "https://github.com/sunblaze-ucb/AgentSynth"}, "environments": ["desktop", "web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:35:17Z", "role": "primary", "sha256": "c1669d8e3caa27f8cd47c17b243b6a8d42807438d8a6a36074668adc57d33757", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations", "licenses.data", "details.dataset_profile"], "url": "https://huggingface.co/datasets/sunblaze-ucb/AgentSynth/raw/9ec3f3139f9786e81aee14b7fff1eb487cc7e991/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:33:20Z", "role": "primary", "sha256": "0d3797d75dd4089ece8c8d4624f2a726024d25be255255d8770f78de21d75906", "supports": ["identity", "release_status", "details.hub_id", "details.hub_revision", "details.gated", "details.disabled", "details.download_url", "details.dataset_profile"], "url": "https://huggingface.co/api/datasets/sunblaze-ucb/AgentSynth"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:27:12Z", "role": "primary", "sha256": "42e8b8dead7caea3f8b21e93a075e8c895dfc30a25e1470bf97a90940bbf0106", "supports": ["details.repository_url"], "url": "https://raw.githubusercontent.com/sunblaze-ucb/AgentSynth/HEAD/README.md"}], "id": "url:58a209f2b3b970916396", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-18T04:35:17Z", "licenses": {"code": null, "data": {"basis": "Publisher dataset-card declaration for this release; it does not establish rights to all upstream screenshots, applications or source content.", "declared": "Apache-2.0", "retrieved_at": "2026-10-18T04:35:17Z", "url": "https://huggingface.co/datasets/sunblaze-ucb/AgentSynth/raw/9ec3f3139f9786e81aee14b7fff1eb487cc7e991/README.md"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/sunblaze-ucb/AgentSynth"}], "name": "AgentSynth tasks and trajectories", "notes": "Public Hub manifest contains JSONL task files and trajectory ZIP packages. Counts describe the publisher-reported task collection, not independently counted archive rows or numbers inferred from ZIP filenames.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/sunblaze-ucb/AgentSynth/raw/9ec3f3139f9786e81aee14b7fff1eb487cc7e991/README.md", "target_id": "arxiv:2506.14205", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Published AgentSynth task definitions and recorded GUI trajectories, with OSWorld and InSTA training configurations and packaged screenshots, thoughts and commands.", "topics": ["data-generation", "long-horizon", "imitation-learning"], "updated_at": null, "url": "https://huggingface.co/datasets/sunblaze-ucb/AgentSynth"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://huggingface.co/datasets/OpenGVLab/ScaleCUA-Data/raw/main/README.md", "population": "Linux, macOS, Windows, Android, iOS, and Web", "qualifier": "exact", "unit": "platform categories", "value": 6}], "evidence_urls": ["https://huggingface.co/datasets/OpenGVLab/ScaleCUA-Data/raw/main/README.md"], "formats": ["JSONL annotations", "JSON manifest", "split image archives"], "modalities": ["screenshots", "text", "actions"], "notes": "The six categories include Web, so they are not six operating systems. Grounding and trajectory annotations follow different collection processes. No sample total or train/test division is inferred from archive-part counts or Hub size tags.", "splits": [], "supervision": ["GUI grounding", "GUI action trajectories"]}, "disabled": false, "download_url": "https://huggingface.co/datasets/OpenGVLab/ScaleCUA-Data/tree/main", "gated": false, "hub_id": "OpenGVLab/ScaleCUA-Data", "languages": [], "modalities": [], "size_category_reported": null}, "environments": ["cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:38:26Z", "role": "primary", "sha256": "dca4b89176c7653e83545363e4868bd2d5f0bf618df007b68a7e25407262f1ac", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/api/datasets/OpenGVLab/ScaleCUA-Data"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:37Z", "role": "primary", "sha256": "8c545d66cc7199a9258bd91ed48930691d0d99555bc30ee457fae9bf656fd18d", "supports": ["details.dataset_profile", "relations"], "url": "https://huggingface.co/datasets/OpenGVLab/ScaleCUA-Data/raw/main/README.md"}, {"method": "huggingface_api", "retrieved_at": "2026-10-16T06:38:26Z", "role": "primary", "sha256": "dca4b89176c7653e83545363e4868bd2d5f0bf618df007b68a7e25407262f1ac", "supports": ["details", "licenses.data"], "url": "https://huggingface.co/api/datasets/OpenGVLab/ScaleCUA-Data"}], "id": "url:631e74223e198ef40202", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T06:38:26Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/OpenGVLab/ScaleCUA-Data"}], "name": "ScaleCUA Data", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/OpenGVLab/ScaleCUA-Data/raw/main/README.md", "target_id": "arxiv:2509.15221", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Cross-platform training data released with ScaleCUA, with grounding and navigation annotations for Ubuntu, Windows, macOS, Android, iPhone and web screenshots.", "topics": ["data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/OpenGVLab/ScaleCUA-Data"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [], "evidence_urls": ["https://huggingface.co/datasets/cckevinn/GUI-Actor-Data/raw/main/README.md", "https://huggingface.co/api/datasets/cckevinn/GUI-Actor-Data", "https://raw.githubusercontent.com/microsoft/GUI-Actor/HEAD/data/data_config.yaml"], "formats": ["JSON annotations", "ZIP image packages", "multipart image archives"], "modalities": ["GUI screenshots", "bounding boxes", "grounding annotations"], "notes": "Public Hub revision 5a0b35fb includes six bbox JSON files and associated image archives. The training configuration names six input paths with sampling_strategy=all; those input paths are not six disjoint datasets or a verified example count. The card’s source-list wording and mismatched hyperlinks remain unresolved, so its approximately 1M screenshots/10M elements headline is not adopted as a release inventory. No data license is established by the card, and the model repository license is not transferred to these data.", "splits": ["training"], "supervision": ["GUI grounding with bounding-box supervision"]}, "disabled": false, "download_url": "https://huggingface.co/datasets/cckevinn/GUI-Actor-Data/tree/main", "gated": false, "hub_id": "cckevinn/GUI-Actor-Data"}, "environments": ["unspecified"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:42:32Z", "role": "primary", "sha256": "e42bbcf8d466c977f9b2150feb73d8cbda2ff907c279630ad4730a8b1f5d665c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details", "details.dataset_profile"], "url": "https://huggingface.co/datasets/cckevinn/GUI-Actor-Data/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:42:32Z", "role": "primary", "sha256": "d6271cff879b1a2a70490cf3dd95765bef44ecfca531d8ef3545f101673b4423", "supports": ["identity", "release_status", "details", "licenses.data", "details.dataset_profile"], "url": "https://huggingface.co/api/datasets/cckevinn/GUI-Actor-Data"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:00Z", "role": "primary", "sha256": "2dafdebf2fa1e0a5db871addd103e50cb55625f0108c757708029e885920ee4e", "supports": ["summary", "environments"], "url": "https://raw.githubusercontent.com/microsoft/GUI-Actor/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T14:21:12Z", "role": "primary", "sha256": "3521c849b29d7fc88430a8e73a8b2aeb2e613d7fa16ca054dd5bf4ec011b36ca", "supports": ["details.dataset_profile"], "url": "https://raw.githubusercontent.com/microsoft/GUI-Actor/HEAD/data/data_config.yaml"}], "id": "url:66096ad4057a3d3726f5", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-17T02:42:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/cckevinn/GUI-Actor-Data"}], "name": "GUI-Actor data", "notes": "Public JSON annotations and image archives are present. The card has no data-license declaration. Its source list contains mismatched hyperlinks and says six datasets while naming fewer distinct entries; no source-derivation relationships or aggregate count are imported from that list.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/cckevinn/GUI-Actor-Data/raw/main/README.md", "target_id": "arxiv:2506.03143", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "A processed screenshot and bounding-box collection released for training GUI-Actor visual grounding models.", "topics": ["grounding", "data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/cckevinn/GUI-Actor-Data"}
{"aliases": [], "authors": [], "details": {"disabled": false, "download_url": "https://huggingface.co/datasets/xlangai/CUA-Gym/tree/main", "gated": false, "hub_id": "xlangai/CUA-Gym", "languages": ["en"], "modalities": ["tabular", "text"], "size_category_reported": ["10K<n<100K"]}, "environments": ["desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:38:26Z", "role": "primary", "sha256": "43b8ebde3b971e1b8b61d789c786ead59b45a2490bc6ed19e204bafc03a31f44", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://huggingface.co/api/datasets/xlangai/CUA-Gym"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T16:10:25Z", "role": "primary", "sha256": "962cd9162bea83f7adfedb3da50129bcf4203ae65176bdf53f0af6788b9603ef", "supports": ["relations"], "url": "https://huggingface.co/datasets/xlangai/CUA-Gym/raw/main/README.md"}, {"method": "huggingface_api", "retrieved_at": "2026-10-16T06:38:26Z", "role": "primary", "sha256": "43b8ebde3b971e1b8b61d789c786ead59b45a2490bc6ed19e204bafc03a31f44", "supports": ["details", "licenses.data"], "url": "https://huggingface.co/api/datasets/xlangai/CUA-Gym"}], "id": "url:6aa1aa58a7a888395189", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T06:38:26Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card", "declared": "cc-by-4.0", "retrieved_at": "2026-10-16T06:38:26Z", "url": "https://huggingface.co/datasets/xlangai/CUA-Gym"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/xlangai/CUA-Gym"}], "name": "CUA-Gym", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/xlangai/CUA-Gym/raw/main/README.md", "target_id": "arxiv:2605.25624", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Computer-use environment and task resources for agent research.", "topics": ["evaluation", "data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/xlangai/CUA-Gym"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/yiye3/GUICourse/HEAD/README.md", "population": "single-step portion", "qualifier": "approximately", "unit": "action instructions", "value": 67000}, {"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/yiye3/GUICourse/HEAD/README.md", "population": "multi-step portion", "qualifier": "approximately", "unit": "action instructions", "value": 15000}], "evidence_urls": ["https://raw.githubusercontent.com/yiye3/GUICourse/HEAD/README.md"], "formats": ["JSON", "Parquet"], "modalities": ["screenshots", "instructions", "actions"], "notes": "Quantities are rounded publisher descriptions, not downloaded sample counts. Data-license declarations conflict across first-party sources; no permissive license is inferred.", "splits": ["train", "test"], "supervision": ["single-step actions", "multi-step navigation"]}, "disabled": false, "download_url": "https://huggingface.co/datasets/yiye2023/GUIAct/tree/main", "gated": false, "hub_id": "yiye2023/GUIAct", "license_review": {"data": {"evidence_urls": ["https://raw.githubusercontent.com/yiye3/GUICourse/HEAD/README.md", "https://huggingface.co/datasets/yiye2023/GUIAct/raw/main/README.md"], "reason": "Conflicting first-party dataset declarations: the GUICourse project README assigns CC-BY-4.0 to the datasets, while this exact Hub card declares Apache-2.0. The code MIT declaration does not resolve the data conflict."}}}, "environments": ["web", "android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:01Z", "role": "primary", "sha256": "0968273442bfb9a6478fe78f8b9a6247a3bb8e6d028880939b432fd009961e98", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details", "licenses.data"], "url": "https://raw.githubusercontent.com/yiye3/GUICourse/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:01:33Z", "role": "primary", "sha256": "624ccdd6e6c5b29b588a4289d1d94c1a1c0803868bd586175f2c9ebc3bfdd0b6", "supports": ["identity", "release_status", "details"], "url": "https://huggingface.co/api/datasets/yiye2023/GUIAct"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:01:33Z", "role": "primary", "sha256": "98b45ea81164d1e1a1dd82255207053b15cd6c69d922a1c5cf3387ce604d4b74", "supports": ["licenses.data"], "url": "https://huggingface.co/datasets/yiye2023/GUIAct/raw/main/README.md"}], "id": "url:6cd150ec23361b4d2e84", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T06:36:01Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/yiye2023/GUIAct"}], "name": "GUIAct", "notes": "Public JSON annotations and Parquet image packages are present. The project README declares CC-BY-4.0 for datasets, while this Hub card declares Apache-2.0; these conflicting declarations are retained without asserting a single confirmed data license. The project code’s MIT declaration is separate.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/yiye3/GUICourse/HEAD/README.md", "target_id": "arxiv:2406.11317", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Web and Android GUI action-instruction data, with single-step, multi-step and smartphone partitions.", "topics": ["grounding", "data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/yiye2023/GUIAct"}
{"aliases": [], "authors": [], "details": {"disabled": false, "download_url": "https://huggingface.co/datasets/LZXzju/UI-R1-3B-Train/tree/main", "gated": false, "hub_id": "LZXzju/UI-R1-3B-Train"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:03:33Z", "role": "primary", "sha256": "640cb5908abefb770ea6bb8d12f22e9eca622224fa73e9f5c67a3a95f4f7b8e5", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/datasets/LZXzju/UI-R1-3B-Train/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:03:33Z", "role": "primary", "sha256": "482d1304ad53736b3e410694cfa427067ff8b45267374964e4a8ff3833afbbd3", "supports": ["identity", "release_status", "details", "licenses.data"], "url": "https://huggingface.co/api/datasets/LZXzju/UI-R1-3B-Train"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:00Z", "role": "primary", "sha256": "f55ac6a0c50d298d46c5f7083a0879959fd7cd3c1d91496af295ebf2afcc13a3", "supports": ["environments", "summary"], "url": "https://raw.githubusercontent.com/lll6gg/UI-R1/HEAD/README.md"}], "id": "url:7c568d88a472f914d316", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-17T02:03:33Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/LZXzju/UI-R1-3B-Train"}], "name": "UI-R1 3B training data", "notes": "The public manifest includes train.json and a train_imgs directory. The card has no dataset-license declaration, and the model’s MIT metadata is not inherited. The official repository identifies this as v1 training data drawn from AndroidControl and ScreenSpot; later enlarged training mixtures are distinct.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/LZXzju/UI-R1-3B-Train/raw/main/README.md", "target_id": "arxiv:2503.21620", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "The released mobile training subset for the original UI-R1 3B model, with action annotations and individual screenshot files.", "topics": ["data-generation", "grounding", "action-prediction"], "updated_at": null, "url": "https://huggingface.co/datasets/LZXzju/UI-R1-3B-Train"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [], "evidence_urls": ["https://huggingface.co/datasets/HanXiao1999/UI-Genie-RM-517k/raw/main/README.md", "https://raw.githubusercontent.com/Euphoria16/UI-Genie/HEAD/README.md"], "formats": ["JSONL"], "modalities": ["text"], "notes": "The public manifest contains reward_data.jsonl but no separately packaged image archive. The card declares MIT. The 517k identifier is not converted into a verified sample count, and complete image packaging is not inferred from the JSONL filename.", "splits": [], "supervision": ["trajectory rewards"]}, "disabled": false, "download_url": "https://huggingface.co/datasets/HanXiao1999/UI-Genie-RM-517k/tree/main", "gated": false, "hub_id": "HanXiao1999/UI-Genie-RM-517k"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:01:34Z", "role": "primary", "sha256": "dddb70b2b6986525f3980cbe50c3241d17f840a81fda9715b01ac3b25dd35d63", "supports": ["name", "kind", "relations", "licenses.data"], "url": "https://huggingface.co/datasets/HanXiao1999/UI-Genie-RM-517k/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:01:33Z", "role": "primary", "sha256": "ab4bf30b6188959f825f2747ca711b0f190dddfe1658319c881ca9c73be9ecc9", "supports": ["identity", "release_status", "details", "licenses.data"], "url": "https://huggingface.co/api/datasets/HanXiao1999/UI-Genie-RM-517k"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T03:59:54Z", "role": "primary", "sha256": "e08569bd9d8a6ff44537d4fe69243ddfeb8a41ae5c9f87ef84af50a67c64c726", "supports": ["summary", "environments", "topics", "scope"], "url": "https://raw.githubusercontent.com/Euphoria16/UI-Genie/HEAD/README.md"}, {"method": "huggingface_card", "retrieved_at": "2026-10-17T04:01:34Z", "role": "primary", "sha256": "dddb70b2b6986525f3980cbe50c3241d17f840a81fda9715b01ac3b25dd35d63", "supports": ["licenses.data"], "url": "https://huggingface.co/datasets/HanXiao1999/UI-Genie-RM-517k/raw/main/README.md"}], "id": "url:84e5aeeffaf561baa966", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-17T04:01:34Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card metadata; third-party assets retain their own terms", "declared": "mit", "retrieved_at": "2026-10-17T04:01:34Z", "url": "https://huggingface.co/datasets/HanXiao1999/UI-Genie-RM-517k/raw/main/README.md"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/HanXiao1999/UI-Genie-RM-517k"}], "name": "UI-Genie reward data", "notes": "The public manifest contains reward_data.jsonl but no separately packaged image archive. The card declares MIT. The 517k identifier is not converted into a verified sample count, and complete image packaging is not inferred from the JSONL filename.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/HanXiao1999/UI-Genie-RM-517k/raw/main/README.md", "target_id": "arxiv:2505.21496", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Mobile GUI reward annotations released for training UI-Genie’s trajectory-assessment model.", "topics": ["data-generation", "mobile-navigation", "reward-models"], "updated_at": null, "url": "https://huggingface.co/datasets/HanXiao1999/UI-Genie-RM-517k"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://huggingface.co/datasets/showlab/ShowUI-web/raw/main/README.md", "population": "train split", "qualifier": "exact", "unit": "rows", "value": 21988}], "evidence_urls": ["https://huggingface.co/datasets/showlab/ShowUI-web/raw/main/README.md"], "formats": ["Parquet", "JSON", "TAR.GZ"], "modalities": ["screenshots", "text", "bounding boxes", "points"], "notes": "Row count is the card dataset_info value; one row may contain multiple instructions. The card summarizes screenshots as approximately 22K. No data license is declared.", "splits": ["train"], "supervision": ["element grounding"]}, "disabled": false, "download_url": "https://huggingface.co/datasets/showlab/ShowUI-web/tree/main", "gated": false, "hub_id": "showlab/ShowUI-web"}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:42:32Z", "role": "primary", "sha256": "08586aa1ded1f1a1f984fb55697fdb1007d86c7fdf4988a2dc1fcf3fdfd49a47", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/datasets/showlab/ShowUI-web/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:42:32Z", "role": "primary", "sha256": "12412480f1a3957069d9dcdc53abce4ad3e34be7dac7ca6fe1726d04764cadfd", "supports": ["identity", "release_status", "details", "licenses.data"], "url": "https://huggingface.co/api/datasets/showlab/ShowUI-web"}], "id": "url:8933977bb0a57940ed23", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-17T02:42:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/showlab/ShowUI-web"}], "name": "ShowUI-web", "notes": "Public Parquet annotations, metadata JSON and images.tar.gz are present. The card refers to image.tar.gz in its extraction command, while the actual archive is plural. No data-license declaration is present; the ShowUI code or model licenses are not inherited.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/showlab/ShowUI-web/raw/main/README.md", "target_id": "arxiv:2411.17465", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Web-interface grounding examples with instructions, element boxes and points, paired with a separately packaged screenshot archive.", "topics": ["grounding", "data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/showlab/ShowUI-web"}
{"aliases": [], "authors": [], "details": {"availability": "Ungated annotation files and multipart image-archive manifest.", "dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/TongUI-agent/TongUI-agent/HEAD/docs/train.md", "population": "TongUI-collected GUI-Net-1M data, excluding added external mixtures", "qualifier": "approximately", "unit": "training trajectories", "value": 1000000}], "evidence_urls": ["https://huggingface.co/datasets/Bofeee5675/GUI-Net-1M/raw/main/README.md", "https://raw.githubusercontent.com/TongUI-agent/TongUI-agent/HEAD/docs/train.md"], "formats": ["JSON annotations", "multipart image archives"], "modalities": ["text", "screenshots"], "notes": "The source reports an approximate trajectory total; it does not establish per-platform counts or a complete test-split inventory.", "splits": ["training collection"], "supervision": ["GUI trajectory supervision from multimodal tutorials"]}, "license_review": {"data": {"evidence_urls": ["https://huggingface.co/datasets/Bofeee5675/GUI-Net-1M/raw/main/README.md"], "reason": "Publisher declaration for this dataset packaging; original web/tutorial materials retain applicable rights."}}, "release_revision": "60d1e644fcff7d3b62093baec03fe7efeb0d98a1"}, "environments": ["unspecified"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:30:54Z", "role": "primary", "sha256": "f30a18d1a533b8f71f436dae2ee7369f8971422bd76527e069499c5bba911912", "supports": ["name", "kind", "summary", "topics", "scope", "notes", "relations"], "url": "https://raw.githubusercontent.com/TongUI-agent/TongUI-agent/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:32:07Z", "role": "primary", "sha256": "314a77eb0ba43bedffbe6869601e41aea41993eaca59d86ba346652c853802d2", "supports": ["identity", "release_status", "details.release_revision", "details.dataset_profile.formats"], "url": "https://huggingface.co/api/datasets/Bofeee5675/GUI-Net-1M"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:36:55Z", "role": "primary", "sha256": "b51acb70c95d079ccbe1de0a685ad7c8a4b3b5bed11e6162e6f7c2f60d78805e", "supports": ["summary", "details.dataset_profile"], "url": "https://raw.githubusercontent.com/TongUI-agent/TongUI-agent/HEAD/docs/train.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:32:07Z", "role": "primary", "sha256": "c2df5c3a6dd576ae53f0dc0f72ecab07d93ed034b04336abd2072a6f0d812c64", "supports": ["licenses.data", "details.dataset_profile.formats", "notes"], "url": "https://huggingface.co/datasets/Bofeee5675/GUI-Net-1M/raw/main/README.md"}], "id": "url:8c9aaafb918aaf8b4e92", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-17T12:30:54Z", "licenses": {"code": null, "data": {"basis": "Publisher declaration for this dataset packaging; original web/tutorial materials retain applicable rights.", "declared": "apache-2.0", "retrieved_at": "2026-10-17T12:32:07Z", "url": "https://huggingface.co/datasets/Bofeee5675/GUI-Net-1M/raw/main/README.md"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/Bofeee5675/GUI-Net-1M"}], "name": "GUI-Net-1M", "notes": "The ungated manifest lists JSON annotations and multipart image archives. Training documentation describes roughly one million collected training trajectories from Wikihow, Baidu Experience and GUI videos; this is distinct from larger mixtures that add external datasets. Platform-specific distribution and exact row counts were not established from this release metadata. Apache-2.0 is the publisher dataset declaration, without a grant of additional rights to upstream tutorial content. No image archive or example record was downloaded.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/TongUI-agent/TongUI-agent/HEAD/README.md", "target_id": "arxiv:2504.12679", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "A TongUI-collected training corpus from multimodal web tutorials, distributed as annotation files and split screenshot archives for GUI trajectory supervision.", "topics": ["data-generation", "grounding", "imitation-learning"], "updated_at": null, "url": "https://huggingface.co/datasets/Bofeee5675/GUI-Net-1M"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://huggingface.co/datasets/xlangai/AgentNet/raw/main/README.md", "population": "AgentNet collection", "qualifier": "approximately", "unit": "tasks", "value": 22600}], "evidence_urls": ["https://huggingface.co/datasets/xlangai/AgentNet/raw/main/README.md"], "formats": ["JSONL", "image archives"], "modalities": ["screenshots", "instructions", "actions", "generated reasoning"], "notes": "The card distinguishes human task instructions from generated summaries, reasoning, reflections, and quality fields. Generated labels are not independent human quality acceptance records.", "splits": [], "supervision": ["computer-use trajectories", "synthesized reasoning"]}, "disabled": false, "download_url": "https://huggingface.co/datasets/xlangai/AgentNet/tree/main", "gated": false, "hub_id": "xlangai/AgentNet", "languages": ["en"], "modalities": [], "size_category_reported": null}, "environments": ["desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:38:26Z", "role": "primary", "sha256": "f23c588ea9fb27210f9d605e164a32527a24298ca516df98272ff7ca7110eac6", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://huggingface.co/api/datasets/xlangai/AgentNet"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T11:40:46Z", "role": "primary", "sha256": "327a77856df259db0a0c2405478f75a0d0f2bf17c51806abc7c509ad82c3ad57", "supports": ["details.dataset_profile"], "url": "https://huggingface.co/datasets/xlangai/AgentNet/raw/main/README.md"}, {"method": "huggingface_api", "retrieved_at": "2026-10-16T06:38:26Z", "role": "primary", "sha256": "f23c588ea9fb27210f9d605e164a32527a24298ca516df98272ff7ca7110eac6", "supports": ["details", "licenses.data"], "url": "https://huggingface.co/api/datasets/xlangai/AgentNet"}], "id": "url:964d70c40bfc5ca8e1ee", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T06:38:26Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card", "declared": "mit", "retrieved_at": "2026-10-16T06:38:26Z", "url": "https://huggingface.co/datasets/xlangai/AgentNet"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/xlangai/AgentNet"}], "name": "AgentNet", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/api/datasets/xlangai/AgentNet", "target_id": "arxiv:2508.09123", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "22.6K human-annotated desktop computer-use tasks on Windows, macOS and Ubuntu, released with OpenCUA for training and evaluating vision-language-action models.", "topics": ["data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/xlangai/AgentNet"}
{"aliases": [], "authors": [], "details": {"availability": "Public ungated JSON artifact is listed by the Hub manifest.", "dataset_profile": {"counts": [], "evidence_urls": ["https://huggingface.co/api/datasets/BlitherBoom/AutoDroid-V2-Dataset", "https://raw.githubusercontent.com/MobileLLM/AutoDroid-V2/HEAD/step_3_training/README.md"], "formats": ["JSON in ShareGPT conversation format"], "modalities": ["text"], "notes": "The card has license metadata only. The project training README supplies formatting and purpose. No sample count or split inventory is inferred from the training configuration.", "splits": [], "supervision": ["supervised conversations for AutoDroid-V2 fine-tuning"]}, "disabled": false, "download_url": "https://huggingface.co/datasets/BlitherBoom/AutoDroid-V2-Dataset/tree/main", "gated": false, "hub_id": "BlitherBoom/AutoDroid-V2-Dataset", "license_review": {"data": {"evidence_urls": ["https://huggingface.co/datasets/BlitherBoom/AutoDroid-V2-Dataset/raw/main/README.md"], "reason": "Publisher dataset-card license declaration; does not establish permissions for model weights or third-party application content."}}, "release_revision": "f041810dd32a2f2999647b7d4f4b8438482c01b8"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:15:33Z", "role": "primary", "sha256": "1b300edfbf44167da71639a56f8ab668dabd33769575fb313ebef5220fc1bddb", "supports": ["name", "kind", "identity", "release_status", "details.hub_id", "details.gated", "details.disabled", "details.release_revision", "details.availability", "details.download_url", "details.dataset_profile.modalities"], "url": "https://huggingface.co/api/datasets/BlitherBoom/AutoDroid-V2-Dataset"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:15:33Z", "role": "primary", "sha256": "d8d7a46d41a1a37fe4f0a5f637bf55c649310185329127d8a2204632e480be17", "supports": ["licenses.data"], "url": "https://huggingface.co/datasets/BlitherBoom/AutoDroid-V2-Dataset/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:14:58Z", "role": "primary", "sha256": "8ba37983a2c2da53d96a52696c07e4bdbea4e987a1a07b557935d614c3be46d2", "supports": ["summary", "topics", "scope", "notes", "details.dataset_profile", "relations"], "url": "https://raw.githubusercontent.com/MobileLLM/AutoDroid-V2/HEAD/step_3_training/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:14:58Z", "role": "primary", "sha256": "74508100843f8db6db6bb151e77ac1cc604a6b7ebd16d3225266d73434465136", "supports": ["environments", "topics", "scope"], "url": "https://raw.githubusercontent.com/MobileLLM/AutoDroid-V2/HEAD/README_minimal.md"}], "id": "url:9a3506ca6ef034adb424", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-17T09:15:33Z", "licenses": {"code": null, "data": {"basis": "Publisher dataset-card license declaration; does not establish permissions for model weights or third-party application content.", "declared": "mit", "retrieved_at": "2026-10-17T09:15:33Z", "url": "https://huggingface.co/datasets/BlitherBoom/AutoDroid-V2-Dataset/raw/main/README.md"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/BlitherBoom/AutoDroid-V2-Dataset"}], "name": "AutoDroid V2 training data", "notes": "The public Hub manifest lists AutoDroid-V2-Dataset.json; no training records were downloaded. The card contains only an MIT metadata declaration, so the description and intended training use come from the official repository training guide. Sample count, split sizes and third-party-content rights remain unverified; the training command max_samples value is not treated as a dataset count.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/MobileLLM/AutoDroid-V2/HEAD/step_3_training/README.md", "target_id": "github:mobilellm/autodroid-v2", "type": "component_of"}], "review_status": "curated", "scope": "core", "summary": "Conversation-formatted supervision released for fine-tuning AutoDroid V2, with the project documenting a ShareGPT-style JSON configuration.", "topics": ["data-generation", "imitation-learning", "mobile-navigation"], "updated_at": null, "url": "https://huggingface.co/datasets/BlitherBoom/AutoDroid-V2-Dataset"}
{"aliases": [], "authors": [], "details": {"disabled": false, "download_url": "https://huggingface.co/datasets/openbmb/CAGUI/tree/main", "gated": false, "hub_id": "openbmb/CAGUI", "languages": ["zh"], "modalities": [], "size_category_reported": ["1K<n<10K"]}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:38:26Z", "role": "primary", "sha256": "98699e13f7c7f5af812e1cbbd5e71eaceefa8aa73ca37167a772774a40818efd", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://huggingface.co/api/datasets/openbmb/CAGUI"}, {"method": "huggingface_api", "retrieved_at": "2026-10-16T06:38:26Z", "role": "primary", "sha256": "98699e13f7c7f5af812e1cbbd5e71eaceefa8aa73ca37167a772774a40818efd", "supports": ["details", "licenses.data"], "url": "https://huggingface.co/api/datasets/openbmb/CAGUI"}], "id": "url:a06622da7688e3718a7b", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T06:38:26Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card", "declared": "cc-by-nc-4.0", "retrieved_at": "2026-10-16T06:38:26Z", "url": "https://huggingface.co/datasets/openbmb/CAGUI"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/openbmb/CAGUI"}], "name": "CAGUI", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/api/datasets/openbmb/CAGUI", "target_id": "arxiv:2506.01391", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Chinese Android GUI benchmark released with AgentCPM-GUI, with a grounding track for widgets and OCR text and an agent track for multi-step tasks in Chinese apps.", "topics": ["data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/openbmb/CAGUI"}
{"aliases": [], "authors": [], "details": {"disabled": false, "download_url": "https://huggingface.co/datasets/alibabagroup/CMGUI/tree/main", "gated": false, "hub_id": "alibabagroup/CMGUI", "languages": ["zh"], "modalities": [], "size_category_reported": ["100K<n<1M"]}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:38:26Z", "role": "primary", "sha256": "aa1e063e73269acd0cb06adcef67d6230f0fe9498ed361d92234053dcf035685", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://huggingface.co/api/datasets/alibabagroup/CMGUI"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T16:10:26Z", "role": "primary", "sha256": "6f43b9e5f2c783f91cc3271f51db6a9a521c29f21954a2a30ad0cfa186e1c1b7", "supports": ["relations"], "url": "https://huggingface.co/datasets/alibabagroup/CMGUI/raw/main/README.md"}, {"method": "huggingface_api", "retrieved_at": "2026-10-16T06:38:26Z", "role": "primary", "sha256": "aa1e063e73269acd0cb06adcef67d6230f0fe9498ed361d92234053dcf035685", "supports": ["details", "licenses.data"], "url": "https://huggingface.co/api/datasets/alibabagroup/CMGUI"}], "id": "url:ab26f9ef56d651cee587", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T06:38:26Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card", "declared": "cc-by-4.0", "retrieved_at": "2026-10-16T06:38:26Z", "url": "https://huggingface.co/datasets/alibabagroup/CMGUI"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/alibabagroup/CMGUI"}], "name": "CMGUI", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/alibabagroup/CMGUI/raw/main/README.md", "target_id": "arxiv:2603.08533", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "18k trajectories with 98k steps from more than 50 Chinese mobile apps in domains such as e-commerce and social media, for building GUI agents on Chinese apps.", "topics": ["data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/alibabagroup/CMGUI"}
{"aliases": [], "authors": [], "details": {"disabled": false, "download_url": "https://huggingface.co/datasets/mllmTeam/DroidCall/tree/main", "gated": false, "hub_id": "mllmTeam/DroidCall", "license_review": {"data": {"evidence_urls": ["https://huggingface.co/datasets/mllmTeam/DroidCall/raw/main/README.md"], "reason": "The Hub dataset card declares Apache-2.0 in its metadata header (license: apache-2.0). The companion repository's MIT code license is a separate declaration."}}}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:28Z", "role": "primary", "sha256": "08a6c5cfa655e1ecd774a41a9a815e845ad0afd47b289378bc1d6ee66c81dda9", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details", "licenses.data"], "url": "https://huggingface.co/datasets/mllmTeam/DroidCall/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:28Z", "role": "primary", "sha256": "c7804987d56b03314a4c48e3ddf2e5819fc0d9c83cc65ba54a981a2f4a584dd9", "supports": ["identity", "release_status", "details", "licenses.data"], "url": "https://huggingface.co/api/datasets/mllmTeam/DroidCall"}], "id": "url:ae8f5208eea9e7f1f606", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-17T02:02:28Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card metadata; third-party assets retain their own terms", "declared": "apache-2.0", "retrieved_at": "2026-10-17T02:02:28Z", "url": "https://huggingface.co/datasets/mllmTeam/DroidCall/raw/main/README.md"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/mllmTeam/DroidCall"}], "name": "DroidCall data", "notes": "The release provides JSONL data and API annotations. This is Android intent-calling supervision rather than screenshot-and-click trajectories, so it is marked adjacent. The dataset card declares Apache-2.0; the companion repository’s MIT code license is a separate declaration.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/mllmTeam/DroidCall/raw/main/README.md", "target_id": "arxiv:2412.00402", "type": "documented_by"}], "review_status": "curated", "scope": "adjacent", "summary": "Instruction-to-function-call examples for learning Android intent invocation, with training, testing and API-annotation files for mobile task execution.", "topics": ["mobile-navigation", "tool-integration", "data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/mllmTeam/DroidCall"}
{"aliases": [], "authors": ["Biplab Deka", "Zifeng Huang", "Chad Franzen", "Joshua Hibschman", "Daniel Afergan", "Yang Li", "Jeffrey Nichols", "Ranjitha Kumar"], "details": {"publication": {"authors": ["Biplab Deka", "Zifeng Huang", "Chad Franzen", "Joshua Hibschman", "Daniel Afergan", "Yang Li", "Jeffrey Nichols", "Ranjitha Kumar"], "date_basis": "Supplied day precision: 2017-10-20. PDF cover, Published: 20 October 2017 The 18 January 2026 download date and October 22-25 conference dates are not publication dates.", "date_parts": [2017, 10, 20], "doi": "10.1145/3126594.3126651", "event_dates": null, "evidence_urls": ["https://danafergan.com/publications/deka2017rico.pdf", "https://dev.interactionmining.org/archive/rico"], "identity_basis": "No same-work record located in the full active catalog. Similarity screening was not used as automatic identity approval. The raw archive and exclusions are additionally checked during normalization.", "issue": null, "notes": "The paper is a distinct resource from the already cataloged Rico dataset (url:0b134b082a7bd507fc6a). The publisher PDF describes 9.7k apps and 72k screens; the project page describes 9.3k apps and 66k screens. The annotation intentionally supplies no merged size claim. No volume number is supplied for this proceedings article.", "pages": "845-854", "pdf_url": "https://danafergan.com/publications/deka2017rico.pdf", "publisher": null, "series": null, "title": "Rico: A Mobile App Dataset for Building Data-Driven Design Applications", "type": "conference-paper", "url": "https://doi.org/10.1145/3126594.3126651", "venue": "UIST '17: Proceedings of the 30th Annual ACM Symposium on User Interface Software and Technology", "volume": null}}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T10:01:32Z", "role": "primary", "sha256": "a1ec58c526b152c052c5c72f9a62725f1026b5f0e4f45c8fe85fce284295da84", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue"], "url": "https://danafergan.com/publications/deka2017rico.pdf"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:53:23Z", "role": "primary", "sha256": "fb00401fe531ca3aed078dccc57147bcaad33207dacf7cd4cfd4af14f1487ff5", "supports": ["details.publication.notes"], "url": "https://dev.interactionmining.org/archive/rico"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T10:01:32Z", "role": "primary", "sha256": "a1ec58c526b152c052c5c72f9a62725f1026b5f0e4f45c8fe85fce284295da84", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue", "environments", "kind", "name", "scope", "summary", "topics"], "url": "https://danafergan.com/publications/deka2017rico.pdf"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:53:23Z", "role": "primary", "sha256": "fb00401fe531ca3aed078dccc57147bcaad33207dacf7cd4cfd4af14f1487ff5", "supports": ["details.publication.notes"], "url": "https://dev.interactionmining.org/archive/rico"}], "id": "url:b560b5d6abaab98483f6", "identifiers": {"doi": "10.1145/3126594.3126651"}, "kind": "paper", "last_verified_at": "2026-10-17T10:01:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://doi.org/10.1145/3126594.3126651"}, {"rel": "publication_pdf", "url": "https://danafergan.com/publications/deka2017rico.pdf"}], "name": "Rico: A Mobile App Dataset for Building Data-Driven Design Applications", "notes": "The annotation was reviewed against retained primary publication sources. Experiments have not been independently reproduced. The paper is a distinct resource from the already cataloged Rico dataset (url:0b134b082a7bd507fc6a). The publisher PDF describes 9.7k apps and 72k screens; the project page describes 9.3k apps and 66k screens. The annotation intentionally supplies no merged size claim.", "published_at": "2017-10-20", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces Rico, collecting Android interface designs and interaction traces through crowdsourced use followed by automated exploration. The paper demonstrates UI-layout retrieval using learned representations and describes visual, textual, structural, and interaction data.", "topics": ["data-generation", "screen-understanding", "ui-design"], "updated_at": null, "url": "https://doi.org/10.1145/3126594.3126651"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [], "evidence_urls": ["https://huggingface.co/datasets/xlangai/aguvis-stage2/raw/main/README.md", "https://huggingface.co/api/datasets/xlangai/aguvis-stage2", "https://raw.githubusercontent.com/xlang-ai/aguvis/HEAD/README.md", "https://raw.githubusercontent.com/xlang-ai/aguvis/HEAD/data/stage2.yaml"], "formats": ["JSON annotations", "ZIP image packages", "multipart TAR image packages"], "modalities": ["GUI screenshots", "natural-language instructions", "action trajectories"], "notes": "Stage 2 trajectory release, Hub revision bd6f1ed4, is distinct from Stage 1 grounding. The public manifest lists source-specific JSON levels and image ZIP/multipart TAR packages. The training recipe uses all rows for some inputs and random:33% for others; several l2 paths occur twice while the manifest also contains l3 files. Recipe rows/levels are neither independent datasets nor disjoint train/test splits. No example/action/trajectory count is inferred. Apache-2.0 is the card declaration, with separate underlying source terms.", "splits": ["training"], "supervision": ["GUI trajectory supervised fine-tuning", "planning and reasoning with inner monologue"]}, "disabled": false, "download_url": "https://huggingface.co/datasets/xlangai/aguvis-stage2/tree/main", "gated": false, "hub_id": "xlangai/aguvis-stage2"}, "environments": ["cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:03:33Z", "role": "primary", "sha256": "550293f00905615e19e63ac896818f9875dad38fea0ffe196c6302a29719aa72", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details", "details.dataset_profile"], "url": "https://huggingface.co/datasets/xlangai/aguvis-stage2/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:03:33Z", "role": "primary", "sha256": "f030264a7a8c7485dc7cf15ca554d84450cc492d2ac90e5965810a0207ef370b", "supports": ["identity", "release_status", "details", "licenses.data", "details.dataset_profile"], "url": "https://huggingface.co/api/datasets/xlangai/aguvis-stage2"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:00Z", "role": "primary", "sha256": "2dd9f36b818c1b7b6bd772ba8575f6c0ebab6d9e0d307edcc2f196f3e4ba502e", "supports": ["environments", "summary", "details.dataset_profile"], "url": "https://raw.githubusercontent.com/xlang-ai/aguvis/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T14:21:11Z", "role": "primary", "sha256": "8c1b39e9d16067018ce845fd4e996f34e2b47346c342b2ba26b030f4a5338b66", "supports": ["details.dataset_profile"], "url": "https://raw.githubusercontent.com/xlang-ai/aguvis/HEAD/data/stage2.yaml"}, {"method": "huggingface_card", "retrieved_at": "2026-10-17T02:03:33Z", "role": "primary", "sha256": "550293f00905615e19e63ac896818f9875dad38fea0ffe196c6302a29719aa72", "supports": ["licenses.data"], "url": "https://huggingface.co/datasets/xlangai/aguvis-stage2/raw/main/README.md"}], "id": "url:b5e2127e07f692d516b9", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-17T02:03:33Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card metadata; third-party assets retain their own terms", "declared": "apache-2.0", "retrieved_at": "2026-10-17T02:03:33Z", "url": "https://huggingface.co/datasets/xlangai/aguvis-stage2/raw/main/README.md"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/xlangai/aguvis-stage2"}], "name": "Aguvis stage 2 trajectory data", "notes": "The manifest contains trajectory annotations and archives associated with mobile, desktop and web sources. The card declares Apache-2.0; underlying source content may retain separate terms. This stage has a different training purpose and files from the grounding release. No archive or training sample was downloaded.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/xlangai/aguvis-stage2/raw/main/README.md", "target_id": "arxiv:2412.04454", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "The second-stage Aguvis collection of interface trajectories for learning action sequences, planning and reasoning after grounding pretraining.", "topics": ["planning", "data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/xlangai/aguvis-stage2"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [], "evidence_urls": ["https://huggingface.co/datasets/xTRam1/plan-and-act-data/raw/main/README.md", "https://huggingface.co/api/datasets/xTRam1/plan-and-act-data"], "formats": ["JSON", "JSONL"], "modalities": ["text", "plans", "actions"], "notes": "The release contains planner and actor files for two web environments. No sample count is inferred from the package-file manifest. Card loading examples have repository/path inconsistencies.", "splits": ["train (QwQ-32B files)", "validation (QwQ-32B files)"], "supervision": ["high-level plans", "action execution"]}, "disabled": false, "download_url": "https://huggingface.co/datasets/xTRam1/plan-and-act-data/tree/main", "gated": false, "hub_id": "xTRam1/plan-and-act-data"}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:29Z", "role": "primary", "sha256": "2ec8f4a03b8331bdd2ba8d6f6800d6dea9d955cfd9e7c1d2168ecb64deb25448", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/datasets/xTRam1/plan-and-act-data/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:28Z", "role": "primary", "sha256": "a98f98b7d4ceb81b00f217983ff271257f2f1e7d2c1a09836b872b2350d79ae9", "supports": ["identity", "release_status", "details", "licenses.data"], "url": "https://huggingface.co/api/datasets/xTRam1/plan-and-act-data"}, {"method": "huggingface_card", "retrieved_at": "2026-10-17T02:02:29Z", "role": "primary", "sha256": "2ec8f4a03b8331bdd2ba8d6f6800d6dea9d955cfd9e7c1d2168ecb64deb25448", "supports": ["licenses.data"], "url": "https://huggingface.co/datasets/xTRam1/plan-and-act-data/raw/main/README.md"}], "id": "url:ba6453cfdae0eb049ced", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-17T02:02:29Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card metadata; third-party assets retain their own terms", "declared": "mit", "retrieved_at": "2026-10-17T02:02:29Z", "url": "https://huggingface.co/datasets/xTRam1/plan-and-act-data/raw/main/README.md"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/xTRam1/plan-and-act-data"}], "name": "Plan-and-Act training data", "notes": "The public manifest exposes JSON/JSONL data for planner and actor roles, including train and validation variants. The card declares MIT. Its loading examples use a different Hub repository identifier and omit the data/ prefix for one file; use the inspected xTRam1 manifest rather than copying those examples verbatim. Third-party site content retains its own terms.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/xTRam1/plan-and-act-data/raw/main/README.md", "target_id": "arxiv:2503.09572", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Planner and action-executor training examples for WebArena-Lite and WebVoyager, with model-specific processed files and raw trajectory annotations.", "topics": ["planning", "web-navigation", "data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/xTRam1/plan-and-act-data"}
{"aliases": [], "authors": ["Fei Tang", "Zhangxuan Gu", "Zhengxi Lu", "Shangzhan Zhang", "Zhengwen Zeng", "Shuheng Shen", "Changhua Meng", "Yuchen Yan", "Wenqi Zhang", "Yongliang Shen", "Weiming Lu", "Yueting Zhuang"], "details": {"publication": {"authors": ["Fei Tang", "Zhangxuan Gu", "Zhengxi Lu", "Shangzhan Zhang", "Zhengwen Zeng", "Shuheng Shen", "Changhua Meng", "Yuchen Yan", "Wenqi Zhang", "Yongliang Shen", "Weiming Lu", "Yueting Zhuang"], "date_basis": "Explicit publisher metadata: month = {June}, year = {2026}; precision=month.", "date_parts": [2026, 6], "doi": null, "event_dates": null, "evidence_urls": ["https://openaccess.thecvf.com/content/CVPR2026/html/Tang_GUI-SAGE_Enhancing_GUI_Automation_with_Self-Explanatory_Learning_CVPR_2026_paper.html", "https://openaccess.thecvf.com/content/CVPR2026/papers/Tang_GUI-SAGE_Enhancing_GUI_Automation_with_Self-Explanatory_Learning_CVPR_2026_paper.pdf"], "identity_basis": "No title/DOI/canonical URL match in the current paper catalog or raw archive; closest substantive same-team works were reread and have different contributions.", "issue": null, "notes": "CVF open-access accepted version; PDF states that the final proceedings version is on IEEE Xplore.", "pages": "13007-13016", "pdf_url": "https://openaccess.thecvf.com/content/CVPR2026/papers/Tang_GUI-SAGE_Enhancing_GUI_Automation_with_Self-Explanatory_Learning_CVPR_2026_paper.pdf", "publisher": null, "series": null, "title": "GUI-SAGE: Enhancing GUI Automation with Self-Explanatory Learning", "type": "conference-paper", "url": "https://openaccess.thecvf.com/content/CVPR2026/html/Tang_GUI-SAGE_Enhancing_GUI_Automation_with_Self-Explanatory_Learning_CVPR_2026_paper.html", "venue": "Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)", "volume": null}}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:53:48+00:00", "role": "primary", "sha256": "1d5ed4e24cf9763cc43f0a8ce14f0e98c01504c73578e5632cea573dfb7da647", "supports": ["details.publication.date_basis", "details.publication.date_parts", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.type", "details.publication.url", "details.publication.venue"], "url": "https://openaccess.thecvf.com/content/CVPR2026/html/Tang_GUI-SAGE_Enhancing_GUI_Automation_with_Self-Explanatory_Learning_CVPR_2026_paper.html"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:56:04+00:00", "role": "primary", "sha256": "244f40d77a40dfd184e78cf8e1e2dd9e469d3e594a8ce214279c2ce17b8f1012", "supports": ["details.publication.authors", "details.publication.identity_basis", "details.publication.notes", "details.publication.title"], "url": "https://openaccess.thecvf.com/content/CVPR2026/papers/Tang_GUI-SAGE_Enhancing_GUI_Automation_with_Self-Explanatory_Learning_CVPR_2026_paper.pdf"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:53:48+00:00", "role": "primary", "sha256": "1d5ed4e24cf9763cc43f0a8ce14f0e98c01504c73578e5632cea573dfb7da647", "supports": ["details.publication.date_basis", "details.publication.date_parts", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.type", "details.publication.url", "details.publication.venue", "environments", "kind", "scope", "summary", "topics"], "url": "https://openaccess.thecvf.com/content/CVPR2026/html/Tang_GUI-SAGE_Enhancing_GUI_Automation_with_Self-Explanatory_Learning_CVPR_2026_paper.html"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:56:04+00:00", "role": "primary", "sha256": "244f40d77a40dfd184e78cf8e1e2dd9e469d3e594a8ce214279c2ce17b8f1012", "supports": ["details.publication.authors", "details.publication.identity_basis", "details.publication.notes", "details.publication.title", "environments", "name", "scope", "summary", "topics"], "url": "https://openaccess.thecvf.com/content/CVPR2026/papers/Tang_GUI-SAGE_Enhancing_GUI_Automation_with_Self-Explanatory_Learning_CVPR_2026_paper.pdf"}], "id": "url:beb8f1fd6b945ffc9d93", "identifiers": {}, "kind": "paper", "last_verified_at": "2026-10-17T09:53:48+00:00", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://openaccess.thecvf.com/content/CVPR2026/html/Tang_GUI-SAGE_Enhancing_GUI_Automation_with_Self-Explanatory_Learning_CVPR_2026_paper.html"}, {"rel": "publication_pdf", "url": "https://openaccess.thecvf.com/content/CVPR2026/papers/Tang_GUI-SAGE_Enhancing_GUI_Automation_with_Self-Explanatory_Learning_CVPR_2026_paper.pdf"}], "name": "GUI-SAGE: Enhancing GUI Automation with Self-Explanatory Learning", "notes": "The annotation was reviewed against retained primary publication sources. Experiments have not been independently reproduced. New work proposal from the official CVF accepted open-access version. No invented day, DOI, or arXiv identifier; original annotations describe the reviewed sources and not independent reproduction.", "published_at": null, "relations": [], "review_status": "curated", "scope": "core", "summary": "GUI-SAGE generates reasoning conditioned on known actions and weights policy updates using confidence and reward signals. Its experiments use AndroidControl and GUIOdyssey data and separately evaluate interactive AndroidWorld tasks.", "topics": ["reinforcement-learning", "data-generation", "evaluation", "robustness"], "updated_at": null, "url": "https://openaccess.thecvf.com/content/CVPR2026/html/Tang_GUI-SAGE_Enhancing_GUI_Automation_with_Self-Explanatory_Learning_CVPR_2026_paper.html"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://huggingface.co/datasets/PrentisAI/ScreenRef-Annotations/raw/main/README.md", "population": "full recipe", "qualifier": "exact", "unit": "annotation rows", "value": 633369}, {"basis": "reported", "evidence_url": "https://huggingface.co/datasets/PrentisAI/ScreenRef-Annotations/raw/main/README.md", "population": "point recipe", "qualifier": "exact", "unit": "annotation rows", "value": 610894}], "evidence_urls": ["https://huggingface.co/datasets/PrentisAI/ScreenRef-Annotations/raw/main/README.md"], "formats": ["JSONL", "JSON task recipes", "TSV manifests"], "modalities": ["text", "coordinates", "image references"], "notes": "The recipes reuse source images and are not additive datasets. This download contains no screenshots. Counts are publisher-reported; rows named for trajectories are independent decision examples, not full trajectories.", "splits": ["full/train", "point/train"], "supervision": ["instruction-to-click grounding", "bounding boxes", "absent-target abstention"]}, "license_name": "restricted-research-only"}, "environments": ["desktop", "web", "android", "linux"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T01:48:10Z", "role": "primary", "sha256": "96a87222645c3f823e6f67e50708792e2917ddeb8eea5aacdf2de8c5bbff165f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/datasets/PrentisAI/ScreenRef-Annotations/raw/main/README.md"}, {"method": "huggingface_card", "retrieved_at": "2026-10-17T01:48:10Z", "role": "primary", "sha256": "96a87222645c3f823e6f67e50708792e2917ddeb8eea5aacdf2de8c5bbff165f", "supports": ["licenses.data"], "url": "https://huggingface.co/datasets/PrentisAI/ScreenRef-Annotations/raw/main/README.md"}], "id": "url:c1db2ba79e6d6a3db2b9", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-17T01:48:10Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card metadata; third-party assets retain their own terms", "declared": "other", "retrieved_at": "2026-10-17T01:48:10Z", "url": "https://huggingface.co/datasets/PrentisAI/ScreenRef-Annotations/raw/main/README.md"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/PrentisAI/ScreenRef-Annotations"}], "name": "ScreenRef Annotations", "notes": "The card declares restricted academic research use and no redistribution. Annotation coordinates use a normalized 1000 by 1000 space. Some source instructions name gestures other than the single-click target; the publisher recommends treating those as location supervision.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/PrentisAI/ScreenRef-Annotations/raw/main/README.md", "target_id": "arxiv:2610.19100", "type": "documented_by"}, {"evidence_url": "https://huggingface.co/datasets/PrentisAI/ScreenRef-Annotations/raw/main/README.md", "target_id": "url:56bda04bd8d3a438deb3", "type": "component_of"}], "review_status": "curated", "scope": "core", "summary": "Annotation files, task recipes, and image manifests for ScreenRef; screenshots reside in the separately gated image repository.", "topics": ["grounding", "data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/PrentisAI/ScreenRef-Annotations"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [], "evidence_urls": ["https://huggingface.co/api/datasets/osunlp/UGround-V1-Data", "https://raw.githubusercontent.com/OSU-NLP-Group/UGround/HEAD/README.md"], "formats": ["Parquet shards"], "modalities": ["GUI screenshots", "natural-language grounding descriptions"], "notes": "Public API revision 33325877 lists Parquet assets but gated=auto; the pinned raw card returned HTTP 401 and no access was requested. Official README identifies this as V1 training data; the API excerpt also announces a Web-Hybrid bounding-box variant with coordinates normalized to [0,999]. That does not establish every shard schema or a release-wide point/bbox count. No train/test partition, example total, or unrestricted downloadability is claimed. Preserve the recorded CC-BY-NC-SA-4.0 data declaration separately from model/code terms.", "splits": [], "supervision": ["GUI visual-grounding training for the UGround-V1 family"]}, "disabled": false, "download_url": "https://huggingface.co/datasets/osunlp/UGround-V1-Data/tree/main", "gated": "auto", "hub_id": "osunlp/UGround-V1-Data", "languages": [], "modalities": ["tabular", "text"], "size_category_reported": null}, "environments": ["cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:38:26Z", "role": "primary", "sha256": "0342862bd7857dd21e389ee5d351375dfea722cad4725c8e33271efc93e009f6", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details.dataset_profile"], "url": "https://huggingface.co/api/datasets/osunlp/UGround-V1-Data"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:00Z", "role": "primary", "sha256": "72ca3f23114c3a08fefb724573acf10c9f0a1b3d4a50d811c98356264ecf6ffc", "supports": ["details.dataset_profile", "relations"], "url": "https://raw.githubusercontent.com/OSU-NLP-Group/UGround/HEAD/README.md"}, {"method": "huggingface_api", "retrieved_at": "2026-10-16T06:38:26Z", "role": "primary", "sha256": "0342862bd7857dd21e389ee5d351375dfea722cad4725c8e33271efc93e009f6", "supports": ["details", "licenses.data"], "url": "https://huggingface.co/api/datasets/osunlp/UGround-V1-Data"}], "id": "url:c4898ad1bfc049926a63", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T06:38:26Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card", "declared": "cc-by-nc-sa-4.0", "retrieved_at": "2026-10-16T06:38:26Z", "url": "https://huggingface.co/datasets/osunlp/UGround-V1-Data"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/osunlp/UGround-V1-Data"}], "name": "UGround V1 Data", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/OSU-NLP-Group/UGround/HEAD/README.md", "target_id": "github:osu-nlp-group/uground", "type": "component_of"}], "review_status": "curated", "scope": "core", "summary": "Gated web grounding training data for UGround in Web-Hybrid and Web-Direct subsets, with a bounding-box version whose coordinates are normalized to 0-999.", "topics": ["grounding", "data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/osunlp/UGround-V1-Data"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/AMAP-ML/Code2World/HEAD/readme.md", "population": "AndroidCode corpus described by Code2World README; released JSON rows not recounted", "qualifier": "more-than", "unit": "screen–action pairs", "value": 80000}], "evidence_urls": ["https://raw.githubusercontent.com/AMAP-ML/Code2World/HEAD/readme.md", "https://huggingface.co/api/datasets/GD-ML/AndroidCode"], "formats": ["HTML", "JSON", "Python extraction/loading scripts"], "modalities": ["GUI screenshots (described input)", "GUI actions", "HTML"], "notes": "The manifest lists 58,953 HTML files, which are not converted into transition/example counts. The reported >80k pairs and the HTML-file inventory have different units. The card itself has no sample schema or count; archive/image completeness and JSON contents were not audited.", "splits": ["Android-Control train file", "Android-Control test file"], "supervision": ["action-conditioned next-state code prediction", "generated HTML supervision"]}, "disabled": false, "download_url": "https://huggingface.co/datasets/GD-ML/AndroidCode/tree/af316a6be79bbe08c4fc09ea588b5f80f177ed25", "gated": false, "hub_id": "GD-ML/AndroidCode", "hub_revision": "af316a6be79bbe08c4fc09ea588b5f80f177ed25", "license_review": {"data": {"evidence_urls": ["https://huggingface.co/datasets/GD-ML/AndroidCode/raw/af316a6be79bbe08c4fc09ea588b5f80f177ed25/README.md"], "reason": "Publisher dataset-card declaration for this release; it does not establish rights to all upstream screenshots, applications or source content."}}, "repository_url": "https://github.com/AMAP-ML/Code2World"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:27:14Z", "role": "primary", "sha256": "26c14a14166d3b747454d1924ecdf6d96652ac857ae6ad488630dc843036a777", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations", "details.dataset_profile"], "url": "https://raw.githubusercontent.com/AMAP-ML/Code2World/HEAD/readme.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:33:22Z", "role": "primary", "sha256": "5774a3c096264fbba58b7a16587ea65f35628edb9ca39afe8a32c74ec7fd1ce7", "supports": ["identity", "release_status", "details.hub_id", "details.hub_revision", "details.gated", "details.disabled", "details.download_url", "details.dataset_profile"], "url": "https://huggingface.co/api/datasets/GD-ML/AndroidCode"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:35:17Z", "role": "primary", "sha256": "d8d7a46d41a1a37fe4f0a5f637bf55c649310185329127d8a2204632e480be17", "supports": ["licenses.data"], "url": "https://huggingface.co/datasets/GD-ML/AndroidCode/raw/af316a6be79bbe08c4fc09ea588b5f80f177ed25/README.md"}], "id": "url:cf4d2d45a7314cb16acb", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-18T04:27:14Z", "licenses": {"code": null, "data": {"basis": "Publisher dataset-card declaration for this release; it does not establish rights to all upstream screenshots, applications or source content.", "declared": "MIT", "retrieved_at": "2026-10-18T04:35:17Z", "url": "https://huggingface.co/datasets/GD-ML/AndroidCode/raw/af316a6be79bbe08c4fc09ea588b5f80f177ed25/README.md"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/GD-ML/AndroidCode"}], "name": "AndroidCode", "notes": "The Hub card is license-only; semantic description and reported scale come from the official Code2World README. The public manifest contains HTML files, Android-Control train/test JSON files and extraction/loading scripts. It does not list standalone screenshot image files, and embedded/reference image packaging was not inspected. MIT is the data-card declaration, not a grant for every upstream source dataset.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/AMAP-ML/Code2World/HEAD/readme.md", "target_id": "arxiv:2602.09856", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "HTML and annotation release used for Code2World’s action-conditioned GUI-state prediction training, constructed from mobile GUI trajectories.", "topics": ["world-models", "data-generation", "screen-understanding"], "updated_at": null, "url": "https://huggingface.co/datasets/GD-ML/AndroidCode"}
{"aliases": [], "authors": [], "details": {"availability": "Ungated JSONL artifact manifest; no dataset card or license declaration.", "dataset_profile": {"counts": [], "evidence_urls": ["https://raw.githubusercontent.com/Wuzheng02/GUI-CIDER/HEAD/README.md", "https://raw.githubusercontent.com/Wuzheng02/GUI-CIDER/HEAD/data/dataset.md", "https://huggingface.co/api/datasets/wuuuuuz/GUI-CIDER"], "formats": ["JSONL"], "modalities": ["text"], "notes": "Purpose is established by the official project, not by API tags. No sample records were downloaded; count and data-license fields remain unknown.", "splits": [], "supervision": ["synthesized static planning knowledge", "synthesized dynamic causal knowledge"]}, "gated": false, "hub_id": "wuuuuuz/GUI-CIDER", "release_revision": "47ba5a55ed9ad993ad9562904ea6e978c4524d17"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:18:20Z", "role": "primary", "sha256": "9a46741bbde1373ed4b87887cdadde98e05ea8795d34efe39fc25d69b4d50213", "supports": ["name", "kind", "summary", "topics", "scope", "notes", "relations", "details.dataset_profile"], "url": "https://raw.githubusercontent.com/Wuzheng02/GUI-CIDER/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:23:07Z", "role": "primary", "sha256": "aec6da0e872092d737953f8b4db227867024da10e127cd32d15d05ff95552d45", "supports": ["identity", "release_status", "details.hub_id", "details.gated", "details.release_revision"], "url": "https://huggingface.co/api/datasets/wuuuuuz/GUI-CIDER"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:24:47Z", "role": "primary", "sha256": "bffca456e649a08239f6a1a3be10a7d7c21bcb427e783ba2c304e4b8410d385a", "supports": ["environments", "details.dataset_profile"], "url": "https://raw.githubusercontent.com/Wuzheng02/GUI-CIDER/HEAD/data/dataset.md"}], "id": "url:d0300db5ba754b1a64fa", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-17T11:18:20Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/wuuuuuz/GUI-CIDER"}], "name": "GUI-CIDER causal knowledge data", "notes": "The official project explicitly links this corpus, and the ungated Hub API lists GUI-CIDER.jsonl. The repository has no README card at the Hub (404); purpose and Android context are therefore supported by the project README and dataset guide, while the API supports only identity/file availability. No sample count, split inventory, or data license was established. The project README’s MIT badge is not transferred to this data artifact.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/Wuzheng02/GUI-CIDER/HEAD/README.md", "target_id": "github:wuzheng02/gui-cider", "type": "component_of"}, {"evidence_url": "https://raw.githubusercontent.com/Wuzheng02/GUI-CIDER/HEAD/README.md", "target_id": "arxiv:2605.28534", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "A published JSONL corpus of synthesized planning and causal knowledge for GUI-agent mid-training, linked directly from the official GUI-CIDER project.", "topics": ["data-generation", "imitation-learning", "planning"], "updated_at": null, "url": "https://huggingface.co/datasets/wuuuuuz/GUI-CIDER"}
{"aliases": [], "authors": [], "details": {"disabled": false, "download_url": "https://huggingface.co/datasets/Miaosen/Grounding-Anything-GUI/tree/main", "gated": false, "hub_id": "Miaosen/Grounding-Anything-GUI", "languages": ["en"], "modalities": ["image", "text"], "size_category_reported": null}, "environments": ["unspecified"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:42:32Z", "role": "primary", "sha256": "c64f72eae3dd50d28996f99ab744a5093896e8bb29caa6f7b33a2f60a0b530a5", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/api/datasets/Miaosen/Grounding-Anything-GUI"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:42:32Z", "role": "primary", "sha256": "31b964cfd198f99491f0f480024b2322fbdf3a237f7806847e83978109bc5856", "supports": ["licenses.data"], "url": "https://huggingface.co/datasets/Miaosen/Grounding-Anything-GUI/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:00Z", "role": "primary", "sha256": "a0609394a736dc495d077cd57a2945b018123e5b3f9514d7d5bafb9e690aeae2", "supports": ["summary", "release_status", "relations"], "url": "https://raw.githubusercontent.com/microsoft/Phi-Ground/HEAD/README.md"}, {"method": "huggingface_api", "retrieved_at": "2026-10-17T02:42:32Z", "role": "primary", "sha256": "c64f72eae3dd50d28996f99ab744a5093896e8bb29caa6f7b33a2f60a0b530a5", "supports": ["details", "licenses.data"], "url": "https://huggingface.co/api/datasets/Miaosen/Grounding-Anything-GUI"}], "id": "url:d1830de02226155ee16d", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-17T02:42:32Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card", "declared": "apache-2.0", "retrieved_at": "2026-10-17T02:42:32Z", "url": "https://huggingface.co/datasets/Miaosen/Grounding-Anything-GUI"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/Miaosen/Grounding-Anything-GUI"}], "name": "Grounding-Anything-GUI", "notes": "The Hub exposes train.zip and a card containing only license/language metadata. The official repository marks this training-data release complete, but its Phi-Ground-Any paper is still pending. No binding to the earlier Phi-Ground paper, sample count, split structure or operating-system coverage is inferred.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/microsoft/Phi-Ground/HEAD/README.md", "target_id": "github:microsoft/phi-ground", "type": "component_of"}], "review_status": "curated", "scope": "core", "summary": "The public training archive linked by the Phi-Ground project for its Phi-Ground-Any model.", "topics": ["grounding", "data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/Miaosen/Grounding-Anything-GUI"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [], "evidence_urls": ["https://huggingface.co/datasets/OS-Copilot/OS-Genesis-mobile-data/raw/main/README.md", "https://huggingface.co/api/datasets/OS-Copilot/OS-Genesis-mobile-data", "https://raw.githubusercontent.com/OS-Copilot/OS-Genesis/HEAD/README.md"], "formats": ["JSONL", "ZIP"], "modalities": ["screenshots", "text", "actions"], "notes": "AC and AW refer to distinct target training sets; no sample count is inferred from the two JSONL files or ZIP packaging.", "splits": ["AndroidControl training", "AndroidWorld training"], "supervision": ["mobile action trajectories"]}, "disabled": false, "download_url": "https://huggingface.co/datasets/OS-Copilot/OS-Genesis-mobile-data/tree/main", "gated": false, "hub_id": "OS-Copilot/OS-Genesis-mobile-data"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:01:34Z", "role": "primary", "sha256": "de9de63e21e60e882277a6268e8153abb6c14aaf3e9a7f83e7f98ac14bf37170", "supports": ["name", "kind", "relations", "licenses.data"], "url": "https://huggingface.co/datasets/OS-Copilot/OS-Genesis-mobile-data/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:01:33Z", "role": "primary", "sha256": "6a5e0c2baecc2dc8bf4466903798f5ecda07cdac917042bbd4cfbe54d895c11e", "supports": ["identity", "release_status", "details", "licenses.data"], "url": "https://huggingface.co/api/datasets/OS-Copilot/OS-Genesis-mobile-data"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:01Z", "role": "primary", "sha256": "430e495ab250395d666854bd18b0fc29b819cd784cd7730215ef65e461834d0d", "supports": ["summary", "environments", "topics", "scope"], "url": "https://raw.githubusercontent.com/OS-Copilot/OS-Genesis/HEAD/README.md"}, {"method": "huggingface_card", "retrieved_at": "2026-10-17T04:01:34Z", "role": "primary", "sha256": "de9de63e21e60e882277a6268e8153abb6c14aaf3e9a7f83e7f98ac14bf37170", "supports": ["licenses.data"], "url": "https://huggingface.co/datasets/OS-Copilot/OS-Genesis-mobile-data/raw/main/README.md"}], "id": "url:d3ec5122a2b52fd5ae4b", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-17T04:01:34Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card metadata; third-party assets retain their own terms", "declared": "mit", "retrieved_at": "2026-10-17T04:01:34Z", "url": "https://huggingface.co/datasets/OS-Copilot/OS-Genesis-mobile-data/raw/main/README.md"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/OS-Copilot/OS-Genesis-mobile-data"}], "name": "OS-Genesis mobile training data", "notes": "The public manifest contains AC/AW JSONL training annotations and a screenshot archive. MIT is declared for the data in its own card, separately from Apache-2.0 declarations on some model checkpoints. The gated web-data release is not included in this asset.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/OS-Copilot/OS-Genesis-mobile-data/raw/main/README.md", "target_id": "arxiv:2412.19723", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Reverse-task-synthesis mobile trajectories released in separate AndroidControl and AndroidWorld training annotation files.", "topics": ["data-generation", "mobile-navigation", "imitation-learning"], "updated_at": null, "url": "https://huggingface.co/datasets/OS-Copilot/OS-Genesis-mobile-data"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [], "evidence_urls": ["https://huggingface.co/datasets/xlangai/Jedi/raw/main/README.md"], "formats": [], "modalities": ["images", "text", "coordinates"], "notes": "Coordinates were resized for a specified Qwen2.5-VL image processor and must remain aligned with image resizing. AGUVIS++ material requires images from upstream AGUVIS and OS-Atlas releases; this repository is not a self-contained copy of every referenced image collection.", "splits": [], "supervision": ["icon grounding", "component grounding", "layout grounding", "refusal"]}, "disabled": false, "download_url": "https://huggingface.co/datasets/xlangai/Jedi/tree/main", "gated": false, "hub_id": "xlangai/Jedi", "languages": [], "modalities": [], "size_category_reported": null}, "environments": ["desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T09:33:18Z", "role": "primary", "sha256": "63ff72bea0a2cb9218093d7a5a420b39c01861aef66141433d7c56c1e0ad05ad", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/api/datasets/xlangai/Jedi"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:38Z", "role": "primary", "sha256": "1d4d15c0cd27837602d55f0fda1ed91d9e8a98b33b3e3c4ba64decae42399558", "supports": ["details.dataset_profile"], "url": "https://huggingface.co/datasets/xlangai/Jedi/raw/main/README.md"}, {"method": "huggingface_api", "retrieved_at": "2026-10-16T09:33:18Z", "role": "primary", "sha256": "63ff72bea0a2cb9218093d7a5a420b39c01861aef66141433d7c56c1e0ad05ad", "supports": ["details", "licenses.data"], "url": "https://huggingface.co/api/datasets/xlangai/Jedi"}], "id": "url:e062cf1445eaee0f3ade", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T09:33:18Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card", "declared": "apache-2.0", "retrieved_at": "2026-10-16T09:33:18Z", "url": "https://huggingface.co/datasets/xlangai/Jedi"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/xlangai/Jedi"}], "name": "Jedi", "notes": "The dataset card requires coordinate scaling to stay aligned with the image processor. Follow the publisher recipe before using the grounding labels.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/api/datasets/xlangai/Jedi", "target_id": "arxiv:2505.13227", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Synthetic GUI grounding data associated with UI decomposition research and the OSWorld-G benchmark.", "topics": ["grounding", "data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/xlangai/Jedi"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [], "evidence_urls": ["https://huggingface.co/datasets/osunlp/Dreamer-V1-Data/raw/main/README.md"], "formats": [], "modalities": ["images", "prompts", "actions", "responses"], "notes": "The card reports a cleaned size of 3.12M without defining the counting unit, so no sample total is inferred. The data card declares noncommercial share-alike terms, separately from the model-weight license.", "splits": [], "supervision": ["web-world-model prediction"]}, "download_url": "https://huggingface.co/datasets/osunlp/Dreamer-V1-Data/tree/main", "modalities": ["image", "text"]}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T11:09:28Z", "role": "primary", "sha256": "e656246f09340fed08136cf41a8b79880e19e1397ddd176410688ddee4e6365c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details.modalities", "details.download_url", "details.dataset_profile"], "url": "https://huggingface.co/datasets/osunlp/Dreamer-V1-Data/raw/main/README.md"}, {"method": "huggingface_card", "retrieved_at": "2026-10-16T11:09:28Z", "role": "primary", "sha256": "e656246f09340fed08136cf41a8b79880e19e1397ddd176410688ddee4e6365c", "supports": ["licenses.data"], "url": "https://huggingface.co/datasets/osunlp/Dreamer-V1-Data/raw/main/README.md"}], "id": "url:eadc6cba9ea46c38dd1d", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T11:09:28Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card metadata; third-party assets retain their own terms", "declared": "cc-by-nc-sa-4.0", "retrieved_at": "2026-10-16T11:09:28Z", "url": "https://huggingface.co/datasets/osunlp/Dreamer-V1-Data/raw/main/README.md"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/osunlp/Dreamer-V1-Data"}], "name": "Dreamer V1 Data", "notes": "The card reports a cleaned size of 3.12M without explicitly defining its counting unit. The catalog leaves the exact sample count unknown. Its data license differs from the model-weight license.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/osunlp/Dreamer-V1-Data/raw/main/README.md", "target_id": "arxiv:2411.06559", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Web-world-model training data with prompts, images, action descriptions, and responses, released with the WebDreamer project.", "topics": ["world-models", "data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/osunlp/Dreamer-V1-Data"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/yiye3/GUICourse/HEAD/README.md", "population": "single-turn portion", "qualifier": "approximately", "unit": "QA pairs", "value": 44000}, {"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/yiye3/GUICourse/HEAD/README.md", "population": "multi-turn portion", "qualifier": "approximately", "unit": "dialogues", "value": 6000}], "evidence_urls": ["https://raw.githubusercontent.com/yiye3/GUICourse/HEAD/README.md"], "formats": ["JSON", "Parquet"], "modalities": ["screenshots", "text", "bounding boxes"], "notes": "Quantities are rounded publisher descriptions, not downloaded sample counts. Data-license declarations conflict across first-party sources; no permissive license is inferred.", "splits": [], "supervision": ["single-turn question answering", "multi-turn dialogue"]}, "disabled": false, "download_url": "https://huggingface.co/datasets/yiye2023/GUIChat/tree/main", "gated": false, "hub_id": "yiye2023/GUIChat", "license_review": {"data": {"evidence_urls": ["https://raw.githubusercontent.com/yiye3/GUICourse/HEAD/README.md", "https://huggingface.co/datasets/yiye2023/GUIChat/raw/main/README.md"], "reason": "Conflicting first-party dataset declarations: the GUICourse project README assigns CC-BY-4.0 to the datasets, while this exact Hub card declares Apache-2.0. The code MIT declaration does not resolve the data conflict."}}}, "environments": ["unspecified"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:01Z", "role": "primary", "sha256": "0968273442bfb9a6478fe78f8b9a6247a3bb8e6d028880939b432fd009961e98", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details", "licenses.data"], "url": "https://raw.githubusercontent.com/yiye3/GUICourse/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:01:33Z", "role": "primary", "sha256": "3ccb4a9ded4f45aa334f0df5dc9a0f8963c645857e8442c46efe1d9cf4a4360b", "supports": ["identity", "release_status", "details"], "url": "https://huggingface.co/api/datasets/yiye2023/GUIChat"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:01:33Z", "role": "primary", "sha256": "4bcf87ecfbbb8e07a01b21415a970c8b53a5283bf6872b657040d3f45c9241f7", "supports": ["licenses.data"], "url": "https://huggingface.co/datasets/yiye2023/GUIChat/raw/main/README.md"}], "id": "url:edf38d43800be8c955ae", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T06:36:01Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/yiye2023/GUIChat"}], "name": "GUIChat", "notes": "Public JSON annotations and Parquet image packages are present. The project README declares CC-BY-4.0 for datasets, while this Hub card declares Apache-2.0; these conflicting declarations are retained without asserting a single confirmed data license. The project code’s MIT declaration is separate.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/yiye3/GUICourse/HEAD/README.md", "target_id": "arxiv:2406.11317", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Conversational GUI supervision with text-rich images and bounding boxes, organized as single-turn questions and multi-turn dialogues.", "topics": ["grounding", "data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/yiye2023/GUIChat"}
{"aliases": ["url:a6db151f939a6748ff76", "showlab/ShowUI-desktop", "https://huggingface.co/datasets/showlab/ShowUI-desktop"], "authors": [], "details": {"dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://huggingface.co/datasets/showlab/ShowUI-desktop-8K/raw/main/README.md", "population": "train split declared in the dataset card", "qualifier": "exact", "unit": "examples", "value": 7496}], "evidence_urls": ["https://huggingface.co/datasets/showlab/ShowUI-desktop-8K/raw/main/README.md"], "formats": ["Hub table with an image column"], "modalities": ["screenshots", "text", "bounding boxes", "points"], "notes": "The 8K release name is not an exact record count; the card declares 7,496 examples. Images and original annotations come from OmniAct and are augmented with generated descriptions. Examples are not asserted to be unique screenshots.", "splits": ["train"], "supervision": ["instruction-to-element grounding"]}, "disabled": false, "download_url": "https://huggingface.co/datasets/showlab/ShowUI-desktop-8K/tree/main", "gated": false, "hub_id": "showlab/ShowUI-desktop", "languages": [], "modalities": ["image", "text", "timeseries"], "size_category_reported": null}, "environments": ["desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:38:26Z", "role": "primary", "sha256": "84bd2ef05ea837eaeebb248002b2942dd5c72f072beb29a9bd70eda1ea433a31", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/api/datasets/showlab/ShowUI-desktop-8K"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:38Z", "role": "primary", "sha256": "577a78efa03cde05141c23033222014094681cb5d04786dbb4a698df8543b3df", "supports": ["details.dataset_profile", "relations"], "url": "https://huggingface.co/datasets/showlab/ShowUI-desktop-8K/raw/main/README.md"}, {"method": "huggingface_api", "retrieved_at": "2026-10-16T06:38:26Z", "role": "primary", "sha256": "84bd2ef05ea837eaeebb248002b2942dd5c72f072beb29a9bd70eda1ea433a31", "supports": ["details", "licenses.data"], "url": "https://huggingface.co/api/datasets/showlab/ShowUI-desktop-8K"}, {"method": "huggingface_repository_identity", "retrieved_at": "2026-10-17T19:14:57Z", "role": "primary", "sha256": "57eaa720c20e9e3a4428c8292c150e1949f3372c75c6890e424ba3c6600a5e5d", "supports": ["aliases"], "url": "https://huggingface.co/api/datasets/showlab/ShowUI-desktop"}, {"method": "huggingface_repository_identity", "retrieved_at": "2026-10-17T19:14:58Z", "role": "primary", "sha256": "57eaa720c20e9e3a4428c8292c150e1949f3372c75c6890e424ba3c6600a5e5d", "supports": ["aliases"], "url": "https://huggingface.co/api/datasets/showlab/ShowUI-desktop-8K"}], "id": "url:f976dec3d1c5e0bedbd7", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T06:38:26Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/showlab/ShowUI-desktop-8K"}], "name": "ShowUI Desktop", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/showlab/ShowUI-desktop-8K/raw/main/README.md", "target_id": "github:showlab/showui", "type": "component_of"}], "review_status": "curated", "scope": "core", "summary": "7,496 desktop grounding examples built from OmniAct screenshots, with annotations rewritten by GPT-4o to describe element appearance, spatial relations and function.", "topics": ["data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/showlab/ShowUI-desktop-8K"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://huggingface.co/datasets/LangAGI-Lab/WebPRMCollection_checklist_generation/raw/main/README.md", "population": "Hub split named test", "qualifier": "exact", "unit": "rows", "value": 3630}], "evidence_urls": ["https://huggingface.co/datasets/LangAGI-Lab/WebPRMCollection_checklist_generation/raw/main/README.md", "https://huggingface.co/api/collections/LangAGI-Lab/web-shepherd-advancing-prms-for-reinforcing-web-agents-682b4f4ad607fc27c4dc49e8"], "formats": ["Parquet"], "modalities": ["screenshots", "text", "action history"], "notes": "This is the checklist-generation release rather than the separate preference-pair collection. No unique-task count or data permission is inferred.", "splits": ["test"], "supervision": ["checklist generation", "subgoal analysis"]}, "disabled": false, "download_url": "https://huggingface.co/datasets/LangAGI-Lab/WebPRMCollection_checklist_generation/tree/main", "gated": false, "hub_id": "LangAGI-Lab/WebPRMCollection_checklist_generation", "languages": [], "modalities": ["image", "text"], "size_category_reported": null}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:01:33Z", "role": "primary", "sha256": "2e4142675f2cb994c21a7620a5188b10147ab03326a2047946886b5916789ec2", "supports": ["name", "kind", "identity", "release_status", "details", "summary"], "url": "https://huggingface.co/api/datasets/LangAGI-Lab/WebPRMCollection_checklist_generation"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:01:33Z", "role": "primary", "sha256": "3693cb307065a6ced63a4789a7f592b5eba0ad6a5ffe1a3caf0649decd5cb6e8", "supports": ["details"], "url": "https://huggingface.co/datasets/LangAGI-Lab/WebPRMCollection_checklist_generation/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:45:05Z", "role": "primary", "sha256": "647d9e7bdad5fe2c03a8cfcd34d93f3b2bf7c1553b517a4a843bc0113fa47456", "supports": ["identity", "summary"], "url": "https://huggingface.co/api/collections/LangAGI-Lab/web-shepherd-advancing-prms-for-reinforcing-web-agents-682b4f4ad607fc27c4dc49e8"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:44:04Z", "role": "primary", "sha256": "07431058d8b9a71b80e75b7a74a40a8346c1503eec240628459cc4d4e2f0802f", "supports": ["summary", "environments", "topics", "scope"], "url": "https://raw.githubusercontent.com/kyle8581/Web-Shepherd/HEAD/README.md"}, {"method": "huggingface_api", "retrieved_at": "2026-10-17T04:01:33Z", "role": "primary", "sha256": "2e4142675f2cb994c21a7620a5188b10147ab03326a2047946886b5916789ec2", "supports": ["details", "licenses.data"], "url": "https://huggingface.co/api/datasets/LangAGI-Lab/WebPRMCollection_checklist_generation"}], "id": "url:fcc98b149272bbf31b05", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-17T04:01:33Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/LangAGI-Lab/WebPRMCollection_checklist_generation"}], "name": "WebPRM checklist-generation data", "notes": "Five public Parquet shards are present. This collection provides checklist-generation supervision and is distinct from the previously cataloged chosen/rejected preference data. The Hub calls its split test; that label is preserved. No data license is declared.", "published_at": null, "relations": [], "review_status": "curated", "scope": "core", "summary": "Web-task context, starting observations and action history paired with structured checklists and subgoal analyses for checklist generation.", "topics": ["reward-models", "data-generation", "web-navigation"], "updated_at": null, "url": "https://huggingface.co/datasets/LangAGI-Lab/WebPRMCollection_checklist_generation"}