{"aliases": [], "authors": ["Rodrigo Nogueira", "Kyunghyun Cho"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "67f8697c21dabc2c4781eba46bfb60671506b413471efcdb28864c11649b3a2c", "submitted_at": "2016-02-06T14:53:02Z", "updated_at": "2016-05-20T16:26:58Z", "version": 2, "versioned_url": "https://arxiv.org/abs/1602.02261v2"}}, "environments": ["web"], "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": "2026-10-17T09:11:46Z", "role": "primary", "sha256": "5f99bcea7c1585129af46fa5c978002778dedc3d98e3a907e92030f4a5588f00", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/abs/1602.02261v2"}], "id": "arxiv:1602.02261", "identifiers": {"arxiv": "1602.02261"}, "kind": "paper", "last_verified_at": "2026-10-17T09:11:46Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/1602.02261"}], "name": "End-to-End Goal-Driven Web Navigation", "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": "2016-02-06", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces WebNav for converting websites into goal-directed hyperlink navigation tasks and WikiNav for evaluating language-conditioned navigation and question answering on Wikipedia.", "topics": ["web-navigation", "evaluation", "planning"], "updated_at": "2016-05-20", "url": "https://arxiv.org/abs/1602.02261"}
{"aliases": [], "authors": ["Abishek Sridhar", "Robert Lo", "Frank F. Xu", "Hao Zhu", "Shuyan Zhou"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "896d5f5e9248e2e49da254fb383f84e5b6ce4286b6d30a37ed7944fcfa1f4958", "submitted_at": "2023-05-23T17:10:39Z", "updated_at": "2023-10-30T00:55:53Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2305.14257v3"}}, "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:2305.14257", "identifiers": {"arxiv": "2305.14257"}, "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/2305.14257"}], "name": "Hierarchical Prompting Assists Large Language Model on Web Navigation", "notes": "Reviewed against the retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2023-05-23", "relations": [], "review_status": "curated", "scope": "core", "summary": "Separates web navigation prompting into a SUMMARIZER that extracts action-relevant observations and an ACTOR that chooses the next action from the condensed context.", "topics": ["web-navigation", "planning", "efficiency"], "updated_at": "2023-10-30", "url": "https://arxiv.org/abs/2305.14257"}
{"aliases": [], "authors": ["Haotian Sun", "Yuchen Zhuang", "Lingkai Kong", "Bo Dai", "Chao Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "c254a26ebb8e1bcdca70b7581c780aa6b0dd4e5f5ded85e456f7e8fb8b38fcc6", "submitted_at": "2023-05-26T05:52:27Z", "updated_at": "2023-05-26T05:52:27Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2305.16653v1"}}, "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:2305.16653", "identifiers": {"arxiv": "2305.16653"}, "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/2305.16653"}], "name": "AdaPlanner: Adaptive Planning from Feedback with 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. This is a general planning framework; MiniWoB++ supplies the interface-agent evidence, while ALFWorld is a separate environment.", "published_at": "2023-05-26", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "Introduces AdaPlanner, which revises code-structured plans in response to environmental feedback and reuses successful plans as exemplars. Experiments include MiniWoB++ web tasks alongside ALFWorld.", "topics": ["planning", "self-reflection", "evaluation"], "updated_at": "2023-05-26", "url": "https://arxiv.org/abs/2305.16653"}
{"aliases": [], "authors": ["Longtao Zheng", "Rundong Wang", "Xinrun Wang", "Bo An"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "8fdb5fed2b2fa0ecbf5910dd6813d8de7910af286a1e4e00cdebb734a82e78b0", "submitted_at": "2023-06-13T15:49:41Z", "updated_at": "2024-01-19T06:59:26Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2306.07863v3"}}, "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:2306.07863", "identifiers": {"arxiv": "2306.07863"}, "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/2306.07863"}], "name": "Synapse: Trajectory-as-Exemplar Prompting with Memory for Computer 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": "2023-06-13", "relations": [], "review_status": "curated", "scope": "core", "summary": "Synapse filters task-irrelevant page state and places complete abstracted state-action trajectories into the prompt. An embedding-based exemplar memory retrieves relevant demonstrations for new tasks, evaluated on MiniWoB++ and Mind2Web.", "topics": ["memory", "web-navigation", "planning", "efficiency"], "updated_at": "2024-01-19", "url": "https://arxiv.org/abs/2306.07863"}
{"aliases": [], "authors": ["Izzeddin Gur", "Hiroki Furuta", "Austin Huang", "Mustafa Safdari", "Yutaka Matsuo", "Douglas Eck", "Aleksandra Faust"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "65328d525a6458ef8c401fb0ac4a9b6a5142bad462bd15a212d124c83a56209f", "submitted_at": "2023-07-24T14:56:30Z", "updated_at": "2024-02-25T16:17:43Z", "version": 4, "versioned_url": "https://arxiv.org/abs/2307.12856v4"}}, "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:2307.12856", "identifiers": {"arxiv": "2307.12856"}, "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/2307.12856"}], "name": "A Real-World WebAgent with Planning, Long Context Understanding, and Program 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": "2023-07-24", "relations": [], "review_status": "curated", "scope": "core", "summary": "Combines task decomposition, HTML summarization, and generated Python programs to carry out instructions on websites.", "topics": ["web-navigation", "planning"], "updated_at": "2024-02-25", "url": "https://arxiv.org/abs/2307.12856"}
{"aliases": [], "authors": ["Xiao Liu", "Hao Yu", "Hanchen Zhang", "Yifan Xu", "Xuanyu Lei", "Hanyu Lai", "Yu Gu", "Hangliang Ding", "Kaiwen Men", "Kejuan Yang", "Shudan Zhang", "Xiang Deng", "Aohan Zeng", "Zhengxiao Du", "Chenhui Zhang", "Sheng Shen", "Tianjun Zhang", "Yu Su", "Huan Sun", "Minlie Huang", "Yuxiao Dong", "Jie Tang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d8c4ba4db4aeed3a7b7ce7f77d610d7e34b0e8db26a36463e2c0bdffcad8f22a", "submitted_at": "2023-08-07T16:08:11Z", "updated_at": "2025-10-04T03:54:18Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2308.03688v3"}}, "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:2308.03688", "identifiers": {"arxiv": "2308.03688"}, "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/2308.03688"}], "name": "AgentBench: Evaluating LLMs as 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": "2023-08-07", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "Evaluates language agents across eight interactive environments and analyzes failures in reasoning, decisions, and instruction following.", "topics": ["evaluation", "planning"], "updated_at": "2025-10-04", "url": "https://arxiv.org/abs/2308.03688"}
{"aliases": [], "authors": ["Kaixin Ma", "Hongming Zhang", "Hongwei Wang", "Xiaoman Pan", "Wenhao Yu", "Dong Yu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "399c97382cac325a9d60a073a14221dee6ee6254cb0b8aa88a5341a7bb758c97", "submitted_at": "2023-09-15T05:44:08Z", "updated_at": "2024-02-21T17:42:32Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2309.08172v2"}}, "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:2309.08172", "identifiers": {"arxiv": "2309.08172"}, "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/2309.08172"}], "name": "LASER: LLM Agent with State-Space Exploration for Web 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": "2023-09-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "LASER represents web navigation as exploration over a defined set of states, allowing an agent to revisit earlier states after mistakes. The study evaluates this backtracking approach on WebShop and Amazon tasks.", "topics": ["planning", "robustness", "web-navigation"], "updated_at": "2024-02-21", "url": "https://arxiv.org/abs/2309.08172"}
{"aliases": [], "authors": ["Zhuosheng Zhang", "Aston Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6d9cff1ece6115e87b7a9032f5720deb49694fbf570bea1f152018c4fa80d981", "submitted_at": "2023-09-20T16:12:32Z", "updated_at": "2024-06-07T04:52:29Z", "version": 4, "versioned_url": "https://arxiv.org/abs/2309.11436v4"}, "publication": {"authors": ["Zhuosheng Zhang", "Aston Zhang"], "date_basis": "Explicit publisher citation_publication_date: 2024/8. Preserve supplied precision; do not substitute conference event dates.", "date_parts": [2024, 8], "doi": "10.18653/v1/2024.findings-acl.186", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2024.findings-acl.186/"], "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": "3132-3149", "pdf_url": "https://aclanthology.org/2024.findings-acl.186.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "You Only Look at Screens: Multimodal Chain-of-Action Agents", "type": "conference-paper", "url": "https://aclanthology.org/2024.findings-acl.186/", "venue": "Findings of the Association for Computational Linguistics: ACL 2024", "volume": null}}, "environments": ["android"], "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": "2026-10-18T04:09:48Z", "role": "primary", "sha256": "80fb0185db71f714655c22f5e1eb46d890c045b8af09c49d1b5ee2e0b6911478", "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/2024.findings-acl.186/"}, {"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:2309.11436", "identifiers": {"arxiv": "2309.11436"}, "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/2309.11436"}, {"rel": "publication", "url": "https://aclanthology.org/2024.findings-acl.186/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2024.findings-acl.186.pdf"}], "name": "You Only Look at Screens: Multimodal Chain-of-Action 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": "2023-09-20", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces Auto-GUI, a screenshot-based agent using previous actions and future action plans to guide device interaction.", "topics": ["mobile-navigation", "planning"], "updated_at": "2024-06-07", "url": "https://arxiv.org/abs/2309.11436"}
{"aliases": [], "authors": ["Paloma Sodhi", "S. R. K. Branavan", "Yoav Artzi", "Ryan McDonald"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d843f3c65e5961da773ec0265c2cfbf51d7c8214d9232ecad1638cc0dbb9ef64", "submitted_at": "2023-10-05T17:40:09Z", "updated_at": "2024-08-08T18:00:48Z", "version": 4, "versioned_url": "https://arxiv.org/abs/2310.03720v4"}}, "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"], "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:2310.03720", "identifiers": {"arxiv": "2310.03720"}, "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/2310.03720"}], "name": "SteP: Stacked LLM Policies for Web Actions", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2023-10-05", "relations": [], "review_status": "curated", "scope": "core", "summary": "SteP dynamically composes specialized web-action policies through a stack that records the active chain of policy calls. It studies this control structure across WebArena, MiniWoB++, and CRM workflows.", "topics": ["planning", "web-navigation", "workflow"], "updated_at": "2024-08-08", "url": "https://arxiv.org/abs/2310.03720"}
{"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": ["Tao Li", "Gang Li", "Zhiwei Deng", "Bryan Wang", "Yang Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "95092986184a232af62c61fa2aedefb837851dce9e0ff7a1128a67d47edfde40", "submitted_at": "2023-10-12T21:53:37Z", "updated_at": "2023-10-23T17:39:51Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2310.08740v3"}}, "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:2310.08740", "identifiers": {"arxiv": "2310.08740"}, "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/2310.08740"}], "name": "A Zero-Shot Language Agent for Computer Control with Structured Reflection", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2023-10-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "Studies a zero-shot computer-control agent that plans executable actions and organizes its own reflections after mistakes. MiniWoB++ experiments examine learning to complete tasks without supplied expert trajectories.", "topics": ["self-reflection", "planning", "web-navigation"], "updated_at": "2023-10-23", "url": "https://arxiv.org/abs/2310.08740"}
{"aliases": [], "authors": ["Heyi Tao", "Sethuraman T", "Michal Shlapentokh-Rothman", "Derek Hoiem"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "7defced3e3beb77a5a6513dbfdff3f102edd922907b523bd062522e54720fa5e", "submitted_at": "2023-10-24T17:57:03Z", "updated_at": "2023-10-25T03:54:11Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2310.16042v2"}}, "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:2310.16042", "identifiers": {"arxiv": "2310.16042"}, "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/2310.16042"}], "name": "WebWISE: Web Interface Control and Sequential Exploration with Large Language 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": "2023-10-24", "relations": [], "review_status": "curated", "scope": "core", "summary": "WebWISE filters DOM elements into observations and generates short programs for click, scroll, and text-entry steps. It evaluates in-context learning from either one supplied example or an example collected during a successful zero-shot MiniWoB++ trial.", "topics": ["web-navigation", "tool-integration", "planning"], "updated_at": "2023-10-25", "url": "https://arxiv.org/abs/2310.16042"}
{"aliases": [], "authors": ["Sunjae Lee", "Junyoung Choi", "Jungjae Lee", "Munim Hasan Wasi", "Hojun Choi", "Steven Y. Ko", "Sangeun Oh", "Insik Shin"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "b9b6f10b17513f565a13e035802d8bd681b5b0d468ba000df1427a8d59591e2c", "submitted_at": "2023-12-04T06:13:35Z", "updated_at": "2024-10-16T08:15:53Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2312.03003v3"}}, "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", "notes"], "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:2312.03003", "identifiers": {"arxiv": "2312.03003"}, "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/2312.03003"}], "name": "Explore, Select, Derive, and Recall: Augmenting LLM with Human-like Memory for Mobile Task Automation", "notes": "Reviewed against the retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2023-12-04", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces MobileGPT, which organizes mobile-app experience into reusable subtasks using an explore, select, derive, and recall cycle. Its memory supports learning task procedures and adapting them to new contexts.", "topics": ["mobile-navigation", "memory", "planning"], "updated_at": "2024-10-16", "url": "https://arxiv.org/abs/2312.03003"}
{"aliases": [], "authors": ["Yanchu Guan", "Dong Wang", "Zhixuan Chu", "Shiyu Wang", "Feiyue Ni", "Ruihua Song", "Longfei Li", "Jinjie Gu", "Chenyi Zhuang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "598c8c30348d454c638944c932dd55355b5d3ecda6eb6f3910a27fffcab9622a", "submitted_at": "2023-12-04T07:51:58Z", "updated_at": "2023-12-04T07:51:58Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2312.06677v1"}}, "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", "notes"], "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:2312.06677", "identifiers": {"arxiv": "2312.06677"}, "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/2312.06677"}], "name": "Intelligent Virtual Assistants with LLM-based Process Automation", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental findings and artifact availability have not been independently reproduced. The abstract names Alipay but does not establish Android or iOS specifically.", "published_at": "2023-12-04", "relations": [], "review_status": "curated", "scope": "core", "summary": "Presents LLMPA, a mobile process-automation architecture that decomposes requests, identifies interface elements, predicts actions and checks errors. Demonstrations evaluate multistep operations in Alipay.", "topics": ["mobile-navigation", "planning", "automation"], "updated_at": "2023-12-04", "url": "https://arxiv.org/abs/2312.06677"}
{"aliases": [], "authors": ["Yunpeng Song", "Yiheng Bian", "Yongtao Tang", "Guiyu Ma", "Zhongmin Cai"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "df1390eab35db18dedaf135ddbbc9fdfcacbc9a70c51ec2609cf331cc05e06bd", "submitted_at": "2023-12-18T13:35:46Z", "updated_at": "2024-07-29T14:52:38Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2312.11190v2"}}, "environments": ["android"], "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"], "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:2312.11190", "identifiers": {"arxiv": "2312.11190"}, "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/2312.11190"}], "name": "VisionTasker: Mobile Task Automation Using Vision Based UI Understanding and LLM Task 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": "2023-12-18", "relations": [], "review_status": "curated", "scope": "core", "summary": "VisionTasker converts screenshots into textual UI interpretations and presents one interface at a time to an LLM planner for action selection. It evaluates visual representations and execution of 147 tasks on an Android smartphone, including integration with programming by demonstration.", "topics": ["screen-understanding", "planning", "mobile-navigation"], "updated_at": "2024-07-29", "url": "https://arxiv.org/abs/2312.11190"}
{"aliases": [], "authors": ["Difei Gao", "Lei Ji", "Zechen Bai", "Mingyu Ouyang", "Peiran Li", "Dongxing Mao", "Qinchen Wu", "Weichen Zhang", "Peiyi Wang", "Xiangwu Guo", "Hengxu Wang", "Luowei Zhou", "Mike Zheng Shou"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "c771b076044c170728041a4f54662d21493adcc3742d40db08cb2f9bd0614655", "submitted_at": "2023-12-20T15:28:38Z", "updated_at": "2024-01-01T14:26:39Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2312.13108v2"}}, "environments": ["desktop", "windows"], "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:2312.13108", "identifiers": {"arxiv": "2312.13108"}, "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/2312.13108"}], "name": "ASSISTGUI: Task-Oriented Desktop Graphical User Interface 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": "2023-12-20", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces Windows application tasks and an actor-critic agent framework for mouse-and-keyboard automation of desktop workflows.", "topics": ["evaluation", "planning"], "updated_at": "2024-01-01", "url": "https://arxiv.org/abs/2312.13108"}
{"aliases": [], "authors": ["Junyang Wang", "Haiyang Xu", "Jiabo Ye", "Ming Yan", "Weizhou Shen", "Ji Zhang", "Fei Huang", "Jitao Sang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "81ac041a3e5f98bfc23c91f768b32ad7adc3f28d8c9d55a285bb29906c8b5a9c", "submitted_at": "2024-01-29T13:46:37Z", "updated_at": "2024-04-18T06:53:38Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2401.16158v2"}}, "environments": ["android"], "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:2401.16158", "identifiers": {"arxiv": "2401.16158"}, "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/2401.16158"}], "name": "Mobile-Agent: Autonomous Multi-Modal Mobile Device Agent with Visual Perception", "notes": "Summary and topic tags reviewed against the primary abstract. Reported experiments, task counts, and artifact availability are not independently reproduced.", "published_at": "2024-01-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "Uses visual perception tools and task decomposition to operate mobile applications without relying on app interface metadata.", "topics": ["mobile-navigation", "planning"], "updated_at": "2024-04-18", "url": "https://arxiv.org/abs/2401.16158"}
{"aliases": [], "authors": ["Zhiyong Wu", "Chengcheng Han", "Zichen Ding", "Zhenmin Weng", "Zhoumianze Liu", "Shunyu Yao", "Tao Yu", "Lingpeng Kong"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "674b005f670a3593bd215a8ff75eac05c90ee068578fc26afdd998c40f5b7b71", "submitted_at": "2024-02-12T07:29:22Z", "updated_at": "2024-02-15T09:30:48Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2402.07456v2"}}, "environments": ["desktop", "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", "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:2402.07456", "identifiers": {"arxiv": "2402.07456"}, "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/2402.07456"}], "name": "OS-Copilot: Towards Generalist Computer Agents with Self-Improvement", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-02-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "Provides an operating-system agent framework connecting applications, files, terminals, and web tools, with accumulated skills for subsequent tasks.", "topics": ["tool-integration", "planning"], "updated_at": "2024-02-15", "url": "https://arxiv.org/abs/2402.07456"}
{"aliases": [], "authors": ["Chaoyun Zhang", "Liqun Li", "Shilin He", "Xu Zhang", "Bo Qiao", "Si Qin", "Minghua Ma", "Yu Kang", "Qingwei Lin", "Saravan Rajmohan", "Dongmei Zhang", "Qi Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "533acb1c9aa3f1c77adb2c67779da86e3d5de95264ba3ec1f81d803f67da7753", "submitted_at": "2024-02-08T15:40:35Z", "updated_at": "2024-05-23T05:18:58Z", "version": 5, "versioned_url": "https://arxiv.org/abs/2402.07939v5"}, "publication": {"authors": ["Chaoyun Zhang", "Liqun Li", "Shilin He", "Xu Zhang", "Bo Qiao", "Si Qin", "Minghua Ma", "Yu Kang", "Qingwei Lin", "Saravan Rajmohan", "Dongmei Zhang", "Qi Zhang"], "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.26", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2025.naacl-long.26/"], "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": "597-622", "pdf_url": "https://aclanthology.org/2025.naacl-long.26.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "UFO: A UI-Focused Agent for Windows OS Interaction", "type": "conference-paper", "url": "https://aclanthology.org/2025.naacl-long.26/", "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": ["windows", "desktop"], "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": "2026-10-18T04:09:48Z", "role": "primary", "sha256": "bbe15e312bcbc923811ef73e561fa7db6a7244ee3747bb9e3792e166b9327235", "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.26/"}, {"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:2402.07939", "identifiers": {"arxiv": "2402.07939"}, "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/2402.07939"}, {"rel": "publication", "url": "https://aclanthology.org/2025.naacl-long.26/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2025.naacl-long.26.pdf"}], "name": "UFO: A UI-Focused Agent for Windows OS Interaction", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-02-08", "relations": [], "review_status": "curated", "scope": "core", "summary": "UFO uses two agents to inspect Windows application interfaces and their control metadata, then grounds and executes user requests within and across applications. The study evaluates this UI-focused architecture on nine Windows applications.", "topics": ["cross-app", "planning", "grounding", "automation"], "updated_at": "2024-05-23", "url": "https://arxiv.org/abs/2402.07939"}
{"aliases": [], "authors": ["Runliang Niu", "Jindong Li", "Shiqi Wang", "Yali Fu", "Xiyu Hu", "Xueyuan Leng", "He Kong", "Yi Chang", "Qi Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "63424a86302625b15a91ff0252ca16c0a6448160bfd748ba32e386c82883584e", "submitted_at": "2024-02-09T02:33:45Z", "updated_at": "2024-02-09T02:33:45Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2402.07945v1"}}, "environments": ["desktop"], "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:2402.07945", "identifiers": {"arxiv": "2402.07945"}, "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/2402.07945"}], "name": "ScreenAgent: A Vision Language Model-driven Computer Control 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": "2024-02-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "Combines a screenshot-and-action dataset with a planning, acting, and reflecting loop for computer control through mouse and keyboard actions.", "topics": ["automation", "planning"], "updated_at": "2024-02-09", "url": "https://arxiv.org/abs/2402.07945"}
{"aliases": [], "authors": ["Jiwen Zhang", "Jihao Wu", "Yihua Teng", "Minghui Liao", "Nuo Xu", "Xiao Xiao", "Zhongyu Wei", "Duyu Tang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "4de9e7d68aa8450cab65ecb300524e85b1d2b2a91d58fecef370deceaba64b1e", "submitted_at": "2024-03-05T07:09:35Z", "updated_at": "2024-07-13T02:12:30Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2403.02713v2"}, "publication": {"authors": ["Jiwen Zhang (张霁雯)", "Jihao Wu", "Teng Yihua", "Minghui Liao", "Nuo Xu", "Xiao Xiao", "Zhongyu Wei (魏忠钰)", "Duyu Tang"], "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.702", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2024.findings-emnlp.702/"], "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. Publisher spelling, initials, name annotations or author lists differ from the retained preprint.", "pages": "12016-12031", "pdf_url": "https://aclanthology.org/2024.findings-emnlp.702.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "Android in the Zoo: Chain-of-Action-Thought for GUI Agents", "type": "conference-paper", "url": "https://aclanthology.org/2024.findings-emnlp.702/", "venue": "Findings of the Association for Computational Linguistics: EMNLP 2024", "volume": null}}, "environments": ["android"], "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": "2026-10-18T04:09:50Z", "role": "primary", "sha256": "87a46c27e83152e7741c2398f790320f937b8fd770fa7ac7d3a68395427e87d2", "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/2024.findings-emnlp.702/"}, {"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:2403.02713", "identifiers": {"arxiv": "2403.02713"}, "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/2403.02713"}, {"rel": "publication", "url": "https://aclanthology.org/2024.findings-emnlp.702/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2024.findings-emnlp.702.pdf"}], "name": "Android in the Zoo: Chain-of-Action-Thought 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": "2024-03-05", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces Chain-of-Action-Thought and Android-In-The-Zoo annotations that connect prior actions, screen observations, intended actions, and expected outcomes.", "topics": ["mobile-navigation", "planning"], "updated_at": "2024-07-13", "url": "https://arxiv.org/abs/2403.02713"}
{"aliases": [], "authors": ["Weihao Tan", "Wentao Zhang", "Xinrun Xu", "Haochong Xia", "Ziluo Ding", "Boyu Li", "Bohan Zhou", "Junpeng Yue", "Jiechuan Jiang", "Yewen Li", "Ruyi An", "Molei Qin", "Chuqiao Zong", "Longtao Zheng", "Yujie Wu", "Xiaoqiang Chai", "Yifei Bi", "Tianbao Xie", "Pengjie Gu", "Xiyun Li", "Ceyao Zhang", "Long Tian", "Chaojie Wang", "Xinrun Wang", "Börje F. Karlsson", "Bo An", "Shuicheng Yan", "Zongqing Lu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "63c68c596f1c81485e77f98beae696deca9785a5d8725ab4456ee35870e64a99", "submitted_at": "2024-03-05T18:22:29Z", "updated_at": "2024-07-02T17:23:13Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2403.03186v3"}, "publication": {"authors": ["Weihao Tan", "Wentao Zhang", "Xinrun Xu", "Haochong Xia", "Ziluo Ding", "Boyu Li", "Bohan Zhou", "Junpeng Yue", "Jiechuan Jiang", "Yewen Li", "Ruyi An", "Molei Qin", "Chuqiao Zong", "Longtao Zheng", "Yujie Wu", "Xiaoqiang Chai", "Yifei Bi", "Tianbao Xie", "Pengjie Gu", "Xiyun Li", "Ceyao Zhang", "Long Tian", "Chaojie Wang", "Xinrun Wang", "Börje F. Karlsson", "Bo An", "Shuicheng Yan", "Zongqing Lu"], "date_basis": "Explicit publisher metadata: 2025/10/06; precision=day.", "date_parts": [2025, 10, 6], "doi": null, "event_dates": "13--19 Jul 2025", "evidence_urls": ["https://proceedings.mlr.press/v267/tan25h.html"], "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 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": "58658-58725", "pdf_url": "https://raw.githubusercontent.com/mlresearch/v267/main/assets/tan25h/tan25h.pdf", "publisher": "PMLR", "series": "Proceedings of Machine Learning Research", "title": "Cradle: Empowering Foundation Agents towards General Computer Control", "type": "conference-paper", "url": "https://proceedings.mlr.press/v267/tan25h.html", "venue": "Proceedings of the 42nd International Conference on Machine Learning", "volume": "267"}}, "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": "2026-10-17T09:53:47+00:00", "role": "primary", "sha256": "c081bdd38f20d6c3aac756da474b8cb8a6d782ece213f67ebc333bf2b99de2fc", "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.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/v267/tan25h.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:2403.03186", "identifiers": {"arxiv": "2403.03186"}, "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/2403.03186"}, {"rel": "publication", "url": "https://proceedings.mlr.press/v267/tan25h.html"}, {"rel": "publication_pdf", "url": "https://raw.githubusercontent.com/mlresearch/v267/main/assets/tan25h/tan25h.pdf"}], "name": "Cradle: Empowering Foundation Agents Towards General Computer 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": "2024-03-05", "relations": [], "review_status": "curated", "scope": "core", "summary": "Uses screenshot observations, planning modules, and generated low-level keyboard and mouse control for general computer interaction.", "topics": ["planning", "screen-understanding"], "updated_at": "2024-07-02", "url": "https://arxiv.org/abs/2403.03186"}
{"aliases": [], "authors": ["Zirui Song", "Yaohang Li", "Meng Fang", "Yanda Li", "Zhenhao Chen", "Zecheng Shi", "Yuan Huang", "Xiuying Chen", "Ling Chen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "5e531ec2ef9e5d90586b8f3052e16d41da95114b7382047d0a02fd671a0db355", "submitted_at": "2024-04-28T05:33:15Z", "updated_at": "2025-03-23T13:04:57Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2404.18074v3"}}, "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", "notes"], "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:2404.18074", "identifiers": {"arxiv": "2404.18074"}, "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/2404.18074"}], "name": "MMAC-Copilot: Multi-modal Agent Collaboration Operating Copilot", "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": "2024-04-28", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces MMAC-Copilot, which coordinates agents with different expertise to operate PC applications. VIBench evaluates visual interaction with applications that lack usable APIs, spanning office, recreation and game tasks.", "topics": ["automation", "evaluation", "planning"], "updated_at": "2025-03-23", "url": "https://arxiv.org/abs/2404.18074"}
{"aliases": [], "authors": ["Kihoon Son", "Jinhyeon Kwon", "DaEun Choi", "Tae Soo Kim", "Young-Ho Kim", "Sangdoo Yun", "Juho Kim"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a358d417d8bef89a01e799d88564e3ede4678ebda4950e87c96cf1e1954b85e1", "submitted_at": "2024-05-07T17:10:31Z", "updated_at": "2024-05-08T05:44:21Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2405.04497v2"}}, "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", "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=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2405.04497", "identifiers": {"arxiv": "2405.04497"}, "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/2405.04497"}], "name": "Unveiling Disparities in Web Task Handling Between Human and Web Agent", "notes": "Reviewed against the retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2024-05-07", "relations": [], "review_status": "curated", "scope": "core", "summary": "Compares human think-aloud web-task behavior with web-agent planning, action, and reflection. The analysis highlights differences in updating plans, handling ambiguity, and investigating failures.", "topics": ["evaluation", "human-in-the-loop", "planning"], "updated_at": "2024-05-08", "url": "https://arxiv.org/abs/2405.04497"}
{"aliases": [], "authors": ["William E Bishop", "Alice Li", "Christopher Rawles", "Oriana Riva"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "81f62326dddc5c512fb05472eb1f387064c0a5aad77453e9d2e85415bb627fa9", "submitted_at": "2024-05-17T23:27:33Z", "updated_at": "2024-05-17T23:27:33Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2405.11120v1"}}, "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:2405.11120", "identifiers": {"arxiv": "2405.11120"}, "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/2405.11120"}], "name": "Latent State Estimation Helps UI Agents to Reason", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-05-17", "relations": [], "review_status": "curated", "scope": "core", "summary": "Studies explicit textual estimates of latent UI state, including whether commanded actions occurred and how far a task has progressed. The experiments compare agents that reason with these estimates against agents that omit them across public and internal task benchmarks.", "topics": ["state-tracking", "planning", "evaluation"], "updated_at": "2024-05-17", "url": "https://arxiv.org/abs/2405.11120"}
{"aliases": [], "authors": ["Junyang Wang", "Haiyang Xu", "Haitao Jia", "Xi Zhang", "Ming Yan", "Weizhou Shen", "Ji Zhang", "Fei Huang", "Jitao Sang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a1338f3543a04bc2f9734e5fcabbfd5ebda6ac263461c2305f7c39383e938573", "submitted_at": "2024-06-03T05:50:00Z", "updated_at": "2024-06-03T05:50:00Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2406.01014v1"}}, "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:2406.01014", "identifiers": {"arxiv": "2406.01014"}, "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/2406.01014"}], "name": "Mobile-Agent-v2: Mobile Device Operation Assistant with Effective Navigation via Multi-Agent Collaboration", "notes": "Summary and topic tags reviewed against the primary abstract. Reported experiments, task counts, and artifact availability are not independently reproduced.", "published_at": "2024-06-03", "relations": [], "review_status": "curated", "scope": "core", "summary": "Separates mobile planning, decision-making, and reflection, with a memory unit for task progress and relevant content.", "topics": ["planning", "memory", "mobile-navigation"], "updated_at": "2024-06-03", "url": "https://arxiv.org/abs/2406.01014"}
{"aliases": [], "authors": ["Junhee Cho", "Jihoon Kim", "Daseul Bae", "Jinho Choo", "Youngjune Gwon", "Yeong-Dae Kwon"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "5f85b2cf7650f41ba9ee6e4ae83cd9041e077a44adcb685e702d644bd5779963", "submitted_at": "2024-06-11T05:21:20Z", "updated_at": "2024-12-26T11:02:28Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2406.06947v3"}}, "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"], "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:2406.06947", "identifiers": {"arxiv": "2406.06947"}, "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/2406.06947"}], "name": "CAAP: Context-Aware Action Planning Prompting to Solve Computer Tasks with Front-End UI Only", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-06-11", "relations": [], "review_status": "curated", "scope": "core", "summary": "CAAP converts screenshots into textual observations and plans keyboard-and-mouse actions using a prompt that examines task context from multiple perspectives. The agent is evaluated on MiniWoB++ and WebShop using only front-end visual observations.", "topics": ["planning", "screen-understanding", "web-navigation"], "updated_at": "2024-12-26", "url": "https://arxiv.org/abs/2406.06947"}
{"aliases": [], "authors": ["Kevin Qinghong Lin", "Linjie Li", "Difei Gao", "Qinchen WU", "Mingyi Yan", "Zhengyuan Yang", "Lijuan Wang", "Mike Zheng Shou"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "66b5fb1e839490df80e9750fdc049f7fe3d2d4686c4090d78ea01bc542fe55b0", "submitted_at": "2024-06-14T17:59:08Z", "updated_at": "2024-06-14T17:59:08Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2406.10227v1"}, "publication": {"authors": ["Kevin Q. Lin", "Linjie Li", "Difei Gao", "Qinchen Wu", "Mingyi Yan", "Zhengyuan Yang", "Lijuan Wang", "Mike Z. Shou"], "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": "10.52202/079017-2214", "event_dates": null, "evidence_urls": ["https://proceedings.neurips.cc/paper_files/paper/2024/hash/804e757b7d7043c26701c3a313032101-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. Publisher spelling, initials, name annotations or author lists differ from the retained preprint.", "pages": "69329-69360", "pdf_url": "https://proceedings.neurips.cc/paper_files/paper/2024/file/804e757b7d7043c26701c3a313032101-Paper-Datasets_and_Benchmarks_Track.pdf", "publisher": null, "series": null, "title": "VideoGUI: A Benchmark for GUI Automation from Instructional Videos", "type": "conference-paper", "url": "https://proceedings.neurips.cc/paper_files/paper/2024/hash/804e757b7d7043c26701c3a313032101-Abstract-Datasets_and_Benchmarks_Track.html", "venue": "Advances in Neural Information Processing Systems", "volume": "37"}}, "environments": ["desktop", "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": "2026-10-18T04:10:03Z", "role": "primary", "sha256": "5731b34872eb84e83338aa851d7f3922e304162869753bf6100ffdc1183c96cf", "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/804e757b7d7043c26701c3a313032101-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:2406.10227", "identifiers": {"arxiv": "2406.10227"}, "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/2406.10227"}, {"rel": "publication", "url": "https://proceedings.neurips.cc/paper_files/paper/2024/hash/804e757b7d7043c26701c3a313032101-Abstract-Datasets_and_Benchmarks_Track.html"}, {"rel": "publication_pdf", "url": "https://proceedings.neurips.cc/paper_files/paper/2024/file/804e757b7d7043c26701c3a313032101-Paper-Datasets_and_Benchmarks_Track.pdf"}], "name": "VideoGUI: A Benchmark for GUI Automation from Instructional 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": "2024-06-14", "relations": [], "review_status": "curated", "scope": "core", "summary": "Evaluates GUI automation from instructional videos at separate levels of high-level planning, action narration, and primitive execution.", "topics": ["evaluation", "video", "planning"], "updated_at": "2024-06-14", "url": "https://arxiv.org/abs/2406.10227"}
{"aliases": [], "authors": ["Jing Yu Koh", "Stephen McAleer", "Daniel Fried", "Ruslan Salakhutdinov"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "ec7aaf2a34e38f72da652bb2d8350c951229c154e2bc368a927a79287b36c439", "submitted_at": "2024-07-01T17:07:55Z", "updated_at": "2026-02-08T15:06:40Z", "version": 4, "versioned_url": "https://arxiv.org/abs/2407.01476v4"}}, "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": "2026-10-16T06:31:32Z", "role": "primary", "sha256": "b9a83994d8ac36ccdb606a5219511f8bb35ad3948ade4f9f3e28d386221aedfb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "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:2407.01476", "identifiers": {"arxiv": "2407.01476"}, "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/2407.01476"}], "name": "Tree Search for Language Model Agents", "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": "2024-07-01", "relations": [], "review_status": "curated", "scope": "core", "summary": "Applies best-first search to agent actions within interactive web environments, allowing exploration and multistep planning at inference time. WebArena and VisualWebArena experiments examine search behavior and test-time computation.", "topics": ["planning", "web-navigation", "evaluation"], "updated_at": "2026-02-08", "url": "https://arxiv.org/abs/2407.01476"}
{"aliases": [], "authors": ["Jiayi Zhang", "Chuang Zhao", "Yihan Zhao", "Zhaoyang Yu", "Ming He", "Jianping Fan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "aa74755c53e20b3a9ff4b72e03a557f455ecceaefc96a50587aefd78ac3c8176", "submitted_at": "2024-07-04T13:12:19Z", "updated_at": "2024-07-04T13:12:19Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2407.03913v1"}}, "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": "2026-10-16T06:31:32Z", "role": "primary", "sha256": "b9a83994d8ac36ccdb606a5219511f8bb35ad3948ade4f9f3e28d386221aedfb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "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:2407.03913", "identifiers": {"arxiv": "2407.03913"}, "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/2407.03913"}], "name": "MobileExperts: A Dynamic Tool-Enabled Agent Team in Mobile Devices", "notes": "Reviewed against the retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2024-07-04", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces MobileExperts, which assembles mobile-agent teams, lets agents explore and formulate tools, and coordinates them through two levels of planning. A benchmark organizes tasks by hierarchical intelligence levels.", "topics": ["mobile-navigation", "planning", "tool-integration"], "updated_at": "2024-07-04", "url": "https://arxiv.org/abs/2407.03913"}
{"aliases": [], "authors": ["Léo Boisvert", "Megh Thakkar", "Maxime Gasse", "Massimo Caccia", "Thibault Le Sellier De Chezelles", "Quentin Cappart", "Nicolas Chapados", "Alexandre Lacoste", "Alexandre Drouin"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2b56a6208e4fa3523cae3fbae673affcd5fe3d6a9c19df482d27fe71d43ffef5", "submitted_at": "2024-07-07T07:15:49Z", "updated_at": "2025-02-05T21:50:07Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2407.05291v2"}, "publication": {"authors": ["Léo Boisvert", "Megh Thakkar", "Maxime Gasse", "Massimo Caccia", "Thibault L. De Chezelles", "Quentin Cappart", "Nicolas Chapados", "Alexandre Lacoste", "Alexandre Drouin"], "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": "10.52202/079017-0195", "event_dates": null, "evidence_urls": ["https://proceedings.neurips.cc/paper_files/paper/2024/hash/0b82662b6c32e887bb252a74d8cb2d5e-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. Publisher spelling, initials, name annotations or author lists differ from the retained preprint.", "pages": "5996-6051", "pdf_url": "https://proceedings.neurips.cc/paper_files/paper/2024/file/0b82662b6c32e887bb252a74d8cb2d5e-Paper-Datasets_and_Benchmarks_Track.pdf", "publisher": null, "series": null, "title": "WorkArena++: Towards Compositional Planning and Reasoning-based Common Knowledge Work Tasks", "type": "conference-paper", "url": "https://proceedings.neurips.cc/paper_files/paper/2024/hash/0b82662b6c32e887bb252a74d8cb2d5e-Abstract-Datasets_and_Benchmarks_Track.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": "4fe01789f93df840699d604de87c068557fc02aa78d22d53c6b3ae7c9db69cc4", "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/0b82662b6c32e887bb252a74d8cb2d5e-Abstract-Datasets_and_Benchmarks_Track.html"}, {"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:2407.05291", "identifiers": {"arxiv": "2407.05291"}, "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/2407.05291"}, {"rel": "publication", "url": "https://proceedings.neurips.cc/paper_files/paper/2024/hash/0b82662b6c32e887bb252a74d8cb2d5e-Abstract-Datasets_and_Benchmarks_Track.html"}, {"rel": "publication_pdf", "url": "https://proceedings.neurips.cc/paper_files/paper/2024/file/0b82662b6c32e887bb252a74d8cb2d5e-Paper-Datasets_and_Benchmarks_Track.pdf"}], "name": "WorkArena++: Towards Compositional Planning and Reasoning-based Common Knowledge Work Tasks", "notes": "Summary and topic tags reviewed against the primary abstract. Reported experiments, task counts, and artifact availability are not independently reproduced.", "published_at": "2024-07-07", "relations": [], "review_status": "curated", "scope": "core", "summary": "Defines compositional enterprise workflows to evaluate planning and reasoning, with generation of ground-truth interaction traces.", "topics": ["evaluation", "planning", "workflow"], "updated_at": "2025-02-05", "url": "https://arxiv.org/abs/2407.05291"}
{"aliases": [], "authors": ["Tamer Abuelsaad", "Deepak Akkil", "Prasenjit Dey", "Ashish Jagmohan", "Aditya Vempaty", "Ravi Kokku"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "18cfd29fcb66d703d90f747c162db1b3fc6a43f400b4eacbddf081bb83ed7947", "submitted_at": "2024-07-17T21:44:28Z", "updated_at": "2024-07-17T21:44:28Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2407.13032v1"}}, "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": "2026-10-16T06:31:32Z", "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:2407.13032", "identifiers": {"arxiv": "2407.13032"}, "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/2407.13032"}], "name": "Agent-E: From Autonomous Web Navigation to Foundational Design Principles in Agentic Systems", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-07-17", "relations": [], "review_status": "curated", "scope": "core", "summary": "Agent-E combines hierarchical web-task control with DOM distillation and observations of interface changes. WebVoyager experiments inform a broader discussion of primitive skills, observation reduction, and self-improvement in agent architecture.", "topics": ["web-navigation", "planning", "efficiency", "state-tracking"], "updated_at": "2024-07-17", "url": "https://arxiv.org/abs/2407.13032"}
{"aliases": [], "authors": ["Ori Yoran", "Samuel Joseph Amouyal", "Chaitanya Malaviya", "Ben Bogin", "Ofir Press", "Jonathan Berant"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2f028fb7f4d64fd4c29886d49b5d6f0fabddba309b669e019edf13586deb9c90", "submitted_at": "2024-07-22T15:18:45Z", "updated_at": "2024-10-21T15:45:31Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2407.15711v2"}, "publication": {"authors": ["Ori Yoran", "Samuel Joseph Amouyal", "Chaitanya Malaviya", "Ben Bogin", "Ofir Press", "Jonathan Berant"], "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.emnlp-main.505", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2024.emnlp-main.505/"], "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": "8938-8968", "pdf_url": "https://aclanthology.org/2024.emnlp-main.505.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "AssistantBench: Can Web Agents Solve Realistic and Time-Consuming Tasks?", "type": "conference-paper", "url": "https://aclanthology.org/2024.emnlp-main.505/", "venue": "Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing", "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": "2026-10-18T04:09:51Z", "role": "primary", "sha256": "a18f0ea465d83796bcd83b95f48c3c13cbdc0aa778ea0dd454aa61f8082920b4", "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/2024.emnlp-main.505/"}, {"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:2407.15711", "identifiers": {"arxiv": "2407.15711"}, "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/2407.15711"}, {"rel": "publication", "url": "https://aclanthology.org/2024.emnlp-main.505/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2024.emnlp-main.505.pdf"}], "name": "AssistantBench: Can Web Agents Solve Realistic and Time-Consuming 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": "2024-07-22", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces realistic open-web tasks and the SeePlanAct agent, analyzing the limitations of browsing and retrieval in time-consuming assistant work.", "topics": ["evaluation", "web-navigation", "planning"], "updated_at": "2024-10-21", "url": "https://arxiv.org/abs/2407.15711"}
{"aliases": [], "authors": ["Yao Zhang", "Zijian Ma", "Yunpu Ma", "Zhen Han", "Yu Wu", "Volker Tresp"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "9e3a34f16101df7a96ffdabbdf86b060e195494f16499f20bdf54110a0394cae", "submitted_at": "2024-08-28T17:49:29Z", "updated_at": "2024-08-28T17:49:29Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2408.15978v1"}}, "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": "2026-10-16T06:31:32Z", "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:2408.15978", "identifiers": {"arxiv": "2408.15978"}, "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/2408.15978"}], "name": "WebPilot: A Versatile and Autonomous Multi-Agent System for Web Task Execution with Strategic 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": "2024-08-28", "relations": [], "review_status": "curated", "scope": "core", "summary": "WebPilot separates global subtask planning from local Monte Carlo tree search during web execution. It continually refines the high-level plan and adapts local search to uncertain transitions and incomplete observations, evaluated on WebArena and MiniWoB++.", "topics": ["planning", "web-navigation", "long-horizon"], "updated_at": "2024-08-28", "url": "https://arxiv.org/abs/2408.15978"}
{"aliases": [], "authors": ["Segev Shlomov", "Ben wiesel", "Aviad Sela", "Ido Levy", "Liane Galanti", "Roy Abitbol"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "cdb228e652218d7f8981d9d16ca29b465004675c510d1e9c5c7b543bec9dc454", "submitted_at": "2024-09-03T14:17:09Z", "updated_at": "2024-09-03T14:17:09Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2409.01927v1"}}, "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:2409.01927", "identifiers": {"arxiv": "2409.01927"}, "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/2409.01927"}], "name": "From Grounding to Planning: Benchmarking Bottlenecks 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": "2024-09-03", "relations": [], "review_status": "curated", "scope": "core", "summary": "Separates web-agent planning and grounding evaluations on Mind2Web to analyze which components limit task performance.", "topics": ["evaluation", "grounding", "planning"], "updated_at": "2024-09-03", "url": "https://arxiv.org/abs/2409.01927"}
{"aliases": [], "authors": ["Wenyue Hua", "Mengting Wan", "Shashank Vadrevu", "Ryan Nadel", "Yongfeng Zhang", "Chi Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6f4c873d1dd443ffc18fc4b63595fdfc726eb84388eddbfced0dcde67c5e19a7", "submitted_at": "2024-09-30T16:52:51Z", "updated_at": "2024-09-30T16:52:51Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2410.00079v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T07:54:03Z", "role": "primary", "sha256": "13d5969845070cd2877dc51ae5300c2d623e5daef00c8a1552ffa46af3f721a5", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2401.05507%2C2406.07275%2C2409.16427%2C2410.00079%2C2503.07701%2C2503.22738%2C2505.15859%2C2505.17735%2C2505.23596%2C2507.05257%2C2507.06134%2C2507.21953%2C2507.23336%2C2508.06960%2C2508.08997%2C2509.02473%2C2509.23694%2C2510.09116%2C2510.14438%2C2511.18423%2C2512.12818%2C2512.13564%2C2601.10758%2C2601.12449%2C2601.21841%2C2603.01152%2C2603.19993%2C2603.20576%2C2604.04503%2C2604.17562%2C2604.23941%2C2605.16265%2C2606.06448%2C2606.06462%2C2606.29193%2C2607.01647%2C2607.13157&max_results=37"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T07:54:03Z", "role": "primary", "sha256": "13d5969845070cd2877dc51ae5300c2d623e5daef00c8a1552ffa46af3f721a5", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2401.05507%2C2406.07275%2C2409.16427%2C2410.00079%2C2503.07701%2C2503.22738%2C2505.15859%2C2505.17735%2C2505.23596%2C2507.05257%2C2507.06134%2C2507.21953%2C2507.23336%2C2508.06960%2C2508.08997%2C2509.02473%2C2509.23694%2C2510.09116%2C2510.14438%2C2511.18423%2C2512.12818%2C2512.13564%2C2601.10758%2C2601.12449%2C2601.21841%2C2603.01152%2C2603.19993%2C2603.20576%2C2604.04503%2C2604.17562%2C2604.23941%2C2605.16265%2C2606.06448%2C2606.06462%2C2606.29193%2C2607.01647%2C2607.13157&max_results=37"}], "id": "arxiv:2410.00079", "identifiers": {"arxiv": "2410.00079"}, "kind": "paper", "last_verified_at": "2026-10-16T07:54:03Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2410.00079"}], "name": "Interactive Speculative Planning: Enhance Agent Efficiency through Co-design of System and User Interface", "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": "adjacent", "summary": "Studies agent-system and interface co-design that incorporates user interruptions and intermediate feedback to reduce planning latency.", "topics": ["planning", "efficiency", "human-in-the-loop"], "updated_at": "2024-09-30", "url": "https://arxiv.org/abs/2410.00079"}
{"aliases": [], "authors": ["Shaoqing Zhang", "Zhuosheng Zhang", "Kehai Chen", "Xinbei Ma", "Muyun Yang", "Tiejun Zhao", "Min Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "f370d105dc1fc1a377fd1b1fd5d1fc3d0f035694cc7c364e6326c587607c98c6", "submitted_at": "2024-10-01T07:49:24Z", "updated_at": "2024-12-19T14:29:31Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2410.00467v3"}, "publication": {"authors": ["Shaoqing Zhang", "Zhuosheng Zhang", "Kehai Chen (陈科海)", "Xinbei Ma", "Muyun Yang (杨沐昀)", "Tiejun Zhao (赵铁军)", "Min Zhang"], "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.70", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2024.findings-emnlp.70/"], "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": "1304-1320", "pdf_url": "https://aclanthology.org/2024.findings-emnlp.70.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "Dynamic Planning for LLM-based Graphical User Interface Automation", "type": "conference-paper", "url": "https://aclanthology.org/2024.findings-emnlp.70/", "venue": "Findings of the Association for Computational Linguistics: EMNLP 2024", "volume": null}}, "environments": ["unspecified"], "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": "2026-10-18T04:09:53Z", "role": "primary", "sha256": "1d89601c2a7a81756b9e7441c48a312132f385e5370780507d27c9848aa1176d", "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/2024.findings-emnlp.70/"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:35Z", "role": "primary", "sha256": "b3e67984f59a1b05ccd946eb54e2d37b0bb78ab0ca796fbb6d544793dafea3e0", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "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:2410.00467", "identifiers": {"arxiv": "2410.00467"}, "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/2410.00467"}, {"rel": "publication", "url": "https://aclanthology.org/2024.findings-emnlp.70/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2024.findings-emnlp.70.pdf"}], "name": "Dynamic Planning for LLM-based Graphical User Interface 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": "2024-10-01", "relations": [], "review_status": "curated", "scope": "core", "summary": "Dynamic Planning of Thoughts revises a mobile GUI plan using current environmental feedback and execution history. The study compares this planning strategy with long-dialogue prompting and examines its behavior on unfamiliar tasks and different language-model backbones.", "topics": ["planning", "interaction-history", "mobile-navigation", "robustness"], "updated_at": "2024-12-19", "url": "https://arxiv.org/abs/2410.00467"}
{"aliases": [], "authors": ["Nikolai Rozanov", "Marek Rei"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "8a2e16e978f50c8d7161a11a11fba6aeda9617663de1746422151d38c4874cc9", "submitted_at": "2024-09-21T05:54:35Z", "updated_at": "2025-04-08T06:37:51Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2410.02810v3"}}, "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": "2026-10-16T06:31:35Z", "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:2410.02810", "identifiers": {"arxiv": "2410.02810"}, "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/2410.02810"}], "name": "StateAct: Enhancing LLM Base Agents via Self-prompting and State-tracking", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-09-21", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "Maintains explicit task goals and evolving state descriptions to improve base-agent decisions across interactive tasks, including web shopping.", "topics": ["planning", "memory"], "updated_at": "2025-04-08", "url": "https://arxiv.org/abs/2410.02810"}
{"aliases": [], "authors": ["Saaket Agashe", "Jiuzhou Han", "Shuyu Gan", "Jiachen Yang", "Ang Li", "Xin Eric Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "284888d16b049767f76448f23f5d35fda25b109fd67f450105950173ae81c97b", "submitted_at": "2024-10-10T17:43:51Z", "updated_at": "2024-10-10T17:43:51Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2410.08164v1"}, "publication": {"authors": ["Saaket Agashe", "Jiuzhou Han", "Shuyu Gan", "Jiachen Yang", "Ang Li", "Xin Wang"], "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/394c7c30ea87b5c3521b4d9e9d419071-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": "22924-22946", "pdf_url": "https://proceedings.iclr.cc/paper_files/paper/2025/file/394c7c30ea87b5c3521b4d9e9d419071-Paper-Conference.pdf", "publisher": null, "series": null, "title": "Agent S: An Open Agentic Framework that Uses Computers Like a Human", "type": "conference-paper", "url": "https://proceedings.iclr.cc/paper_files/paper/2025/hash/394c7c30ea87b5c3521b4d9e9d419071-Abstract-Conference.html", "venue": "International Conference on Learning Representations", "volume": "2025"}}, "environments": ["desktop"], "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": "2026-10-18T04:10:01Z", "role": "primary", "sha256": "032cfd2f08acc7f85921748e0546bf6e16d038da412c207105ca9f6d433c70b3", "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/394c7c30ea87b5c3521b4d9e9d419071-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", "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:2410.08164", "identifiers": {"arxiv": "2410.08164"}, "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/2410.08164"}, {"rel": "publication", "url": "https://proceedings.iclr.cc/paper_files/paper/2025/hash/394c7c30ea87b5c3521b4d9e9d419071-Abstract-Conference.html"}, {"rel": "publication_pdf", "url": "https://proceedings.iclr.cc/paper_files/paper/2025/file/394c7c30ea87b5c3521b4d9e9d419071-Paper-Conference.pdf"}], "name": "Agent S: An Open Agentic Framework that Uses Computers Like a Human", "notes": "Editorial summary based on the primary paper abstract; experiments and released artifacts have not been independently reproduced.", "published_at": "2024-10-10", "relations": [], "review_status": "curated", "scope": "core", "summary": "Combines hierarchical planning, retrieved experience, and a computer interaction interface to execute multistep GUI tasks.", "topics": ["planning", "memory"], "updated_at": "2024-10-10", "url": "https://arxiv.org/abs/2410.08164"}
{"aliases": [], "authors": ["Hyungjoo Chae", "Namyoung Kim", "Kai Tzu-iunn Ong", "Minju Gwak", "Gwanwoo Song", "Jihoon Kim", "Sunghwan Kim", "Dongha Lee", "Jinyoung Yeo"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "cae86bff4412bbfa7b969ecd60fe7b695f4b7c5ea1be6a67e97919df762d1108", "submitted_at": "2024-10-17T05:37:00Z", "updated_at": "2025-03-29T08:59:09Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2410.13232v2"}}, "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:2410.13232", "identifiers": {"arxiv": "2410.13232"}, "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/2410.13232"}], "name": "Web Agents with World Models: Learning and Leveraging Environment Dynamics in Web 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": "2024-10-17", "relations": [], "review_status": "curated", "scope": "core", "summary": "Uses a learned web world model to predict important state changes and compare candidate actions before executing them.", "topics": ["world-models", "planning"], "updated_at": "2025-03-29", "url": "https://arxiv.org/abs/2410.13232"}
{"aliases": [], "authors": ["Zichen Zhu", "Hao Tang", "Yansi Li", "Dingye Liu", "Hongshen Xu", "Kunyao Lan", "Danyang Zhang", "Yixuan Jiang", "Hao Zhou", "Chenrun Wang", "Situo Zhang", "Liangtai Sun", "Yixiao Wang", "Yuheng Sun", "Lu Chen", "Kai Yu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "c0ef02f9c31bf2f095330cf018e7525aecc7644eb1552fb6b511ffced3889122", "submitted_at": "2024-10-17T16:53:50Z", "updated_at": "2025-05-13T06:25:09Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2410.13757v3"}, "publication": {"authors": ["Zichen Zhu", "Hao Tang", "Yansi Li", "Dingye Liu", "Hongshen Xu", "Kunyao Lan", "Danyang Zhang", "Yixuan Jiang", "Hao Zhou", "Chenrun Wang", "Situo Zhang", "Liangtai Sun", "Yixiao Wang", "Yuheng Sun", "Lu Chen", "Kai Yu"], "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-demo.43", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2025.naacl-demo.43/"], "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": "535-549", "pdf_url": "https://aclanthology.org/2025.naacl-demo.43.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "MobA: Multifaceted Memory-Enhanced Adaptive Planning for Efficient Mobile Task Automation", "type": "conference-paper", "url": "https://aclanthology.org/2025.naacl-demo.43/", "venue": "Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (System Demonstrations)", "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": "2026-10-18T04:09:53Z", "role": "primary", "sha256": "a873b34264be14f64599c0897219ff2796ab74d111df8a2fdc667fd22ad87028", "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-demo.43/"}, {"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:2410.13757", "identifiers": {"arxiv": "2410.13757"}, "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/2410.13757"}, {"rel": "publication", "url": "https://aclanthology.org/2025.naacl-demo.43/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2025.naacl-demo.43.pdf"}], "name": "MobA: Multifaceted Memory-Enhanced Adaptive Planning for Efficient Mobile Task 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": "2024-10-17", "relations": [], "review_status": "curated", "scope": "core", "summary": "Combines adaptive mobile-task planning, reflection-based recovery, and multiple memory components, evaluated with MobBench and AndroidArena.", "topics": ["memory", "planning", "mobile-navigation"], "updated_at": "2025-05-13", "url": "https://arxiv.org/abs/2410.13757"}
{"aliases": [], "authors": ["Ke Yang", "Yao Liu", "Sapana Chaudhary", "Rasool Fakoor", "Pratik Chaudhari", "George Karypis", "Huzefa Rangwala"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "09268d51b3d810ffc0024ae332d840afc4501c8f6a2017128fca525be4809a13", "submitted_at": "2024-10-17T17:50:38Z", "updated_at": "2025-05-24T03:55:22Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2410.13825v2"}}, "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:2410.13825", "identifiers": {"arxiv": "2410.13825"}, "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/2410.13825"}], "name": "AgentOccam: A Simple Yet Strong Baseline for LLM-Based 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": "2024-10-17", "relations": [], "review_status": "curated", "scope": "core", "summary": "Studies how simplifying and aligning a web agent's observations and actions with language-model capabilities affects task completion.", "topics": ["web-navigation", "planning"], "updated_at": "2025-05-24", "url": "https://arxiv.org/abs/2410.13825"}
{"aliases": [], "authors": ["Xiaoqiang Wang", "Bang Liu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "66fa51b108ac79cc43a895ba757ca463d5008bfe6db5cde4bdf3a3bb38c444eb", "submitted_at": "2024-10-24T17:58:08Z", "updated_at": "2024-10-24T17:58:08Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2410.18963v1"}}, "environments": ["desktop", "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"], "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:2410.18963", "identifiers": {"arxiv": "2410.18963"}, "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/2410.18963"}], "name": "OSCAR: Operating System Control via State-Aware Reasoning and Re-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": "2024-10-24", "relations": [], "review_status": "curated", "scope": "core", "summary": "OSCAR translates user instructions into Python control code while observing screen images and acting through mouse and keyboard operations. A state-machine controller handles errors and replans tasks during execution across desktop and mobile applications.", "topics": ["planning", "state-tracking", "tool-integration", "robustness"], "updated_at": "2024-10-24", "url": "https://arxiv.org/abs/2410.18963"}
{"aliases": [], "authors": ["Jaekyeom Kim", "Dong-Ki Kim", "Lajanugen Logeswaran", "Sungryull Sohn", "Honglak Lee"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "bf1fb8aa2e289a72f3c44c99e2d531eaa5eeea08b78ff6f3a21e139d47502b17", "submitted_at": "2024-10-29T21:37:04Z", "updated_at": "2024-10-29T21:37:04Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2410.22552v1"}}, "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:2410.22552", "identifiers": {"arxiv": "2410.22552"}, "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/2410.22552"}], "name": "Auto-Intent: Automated Intent Discovery and Self-Exploration for Large Language Model 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": "2024-10-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "Discovers compact intents from demonstrations and uses intent predictions to guide a language-model web agent's decisions and exploration.", "topics": ["planning", "web-navigation"], "updated_at": "2024-10-29", "url": "https://arxiv.org/abs/2410.22552"}
{"aliases": [], "authors": ["Adam Fourney", "Gagan Bansal", "Hussein Mozannar", "Cheng Tan", "Eduardo Salinas", "Erkang", "Zhu", "Friederike Niedtner", "Grace Proebsting", "Griffin Bassman", "Jack Gerrits", "Jacob Alber", "Peter Chang", "Ricky Loynd", "Robert West", "Victor Dibia", "Ahmed Awadallah", "Ece Kamar", "Rafah Hosn", "Saleema Amershi"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "4a2c499237dbcbe1482824a2710f5185dee05c9ed66a0882d1e422f30afe056a", "submitted_at": "2024-11-07T06:36:19Z", "updated_at": "2024-11-07T06:36:19Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2411.04468v1"}}, "environments": ["web", "desktop"], "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:2411.04468", "identifiers": {"arxiv": "2411.04468"}, "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/2411.04468"}], "name": "Magentic-One: A Generalist Multi-Agent System for Solving Complex 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": "2024-11-07", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "Introduces an orchestrated team of browsing, file, and coding agents with progress tracking and error recovery for general digital tasks.", "topics": ["planning", "tool-integration"], "updated_at": "2024-11-07", "url": "https://arxiv.org/abs/2411.04468"}
{"aliases": [], "authors": ["Yu Gu", "Kai Zhang", "Yuting Ning", "Boyuan Zheng", "Boyu Gou", "Tianci Xue", "Cheng Chang", "Sanjari Srivastava", "Yanan Xie", "Peng Qi", "Huan Sun", "Yu Su"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "9ecaca6448aae5c8849d008d820d700bfa438ed0248f85be585b0b831408b9b6", "submitted_at": "2024-11-10T18:50:51Z", "updated_at": "2025-04-01T05:04:47Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2411.06559v2"}}, "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:2411.06559", "identifiers": {"arxiv": "2411.06559"}, "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/2411.06559"}], "name": "Is Your LLM Secretly a World Model of the Internet? Model-Based Planning 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": "2024-11-10", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces WebDreamer, which simulates candidate web actions with language-model world models and value estimates before acting.", "topics": ["world-models", "planning"], "updated_at": "2025-04-01", "url": "https://arxiv.org/abs/2411.06559"}
{"aliases": [], "authors": ["Siyuan Hu", "Mingyu Ouyang", "Difei Gao", "Mike Zheng Shou"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "255e101e263c4e3a0cb491d4c34b3e75a8f3660cec2141e3a507a30cf5514dbd", "submitted_at": "2024-11-15T16:23:52Z", "updated_at": "2024-11-15T16:23:52Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2411.10323v1"}}, "environments": ["desktop"], "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", "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=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2411.10323", "identifiers": {"arxiv": "2411.10323"}, "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/2411.10323"}], "name": "The Dawn of GUI Agent: A Preliminary Case Study with Claude 3.5 Computer Use", "notes": "Reviewed against the retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2024-11-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "Presents a preliminary study of Claude 3.5 Computer Use across desktop tasks and software, analyzing planning, action, and critique behavior and providing a framework for deploying GUI-automation models.", "topics": ["evaluation", "automation", "planning"], "updated_at": "2024-11-15", "url": "https://arxiv.org/abs/2411.10323"}
{"aliases": [], "authors": ["Minghe Gao", "Wendong Bu", "Bingchen Miao", "Yang Wu", "Yunfei Li", "Juncheng Li", "Siliang Tang", "Qi Wu", "Yueting Zhuang", "Meng Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "daa9b7092616e94a1d0124d36d8b75d0401b9647ec762c963524616a92def9a5", "submitted_at": "2024-11-17T02:44:56Z", "updated_at": "2024-11-17T02:44:56Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2411.10943v1"}}, "environments": ["web", "desktop", "android", "windows"], "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": "2026-10-18T04:29:13.577033+00:00", "role": "primary", "sha256": "92f18020e481d6f42a6d35e96dff820594e9370597f5392d038e9ee908c49f1e", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/html/2411.10943v1"}], "id": "arxiv:2411.10943", "identifiers": {"arxiv": "2411.10943"}, "kind": "survey", "last_verified_at": "2026-10-18T04:29:13.577033+00:00", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2411.10943"}], "name": "Generalist Virtual Agents: A Survey on Autonomous Agents Across Digital Platforms", "notes": "Reviewed the complete retained arXiv abstract and exact-version official HTML sections on browser/application/OS environments, screen observations, input actions and evaluation. The HTML abstract fingerprint and ordered authors match the retained Atom entry. Literature claims and experiments were not independently reproduced.", "published_at": "2024-11-17", "relations": [], "review_status": "curated", "scope": "core", "summary": "Surveys virtual agents for browsers, applications and operating systems, organizing prior work by task type, observation and action spaces, model strategies and evaluation methods. It examines screen-based interaction and keyboard, mouse and touch control, alongside transferability and long-sequence decision-making challenges.", "topics": ["survey", "web-navigation", "mobile-navigation", "evaluation", "planning"], "updated_at": "2024-11-17", "url": "https://arxiv.org/abs/2411.10943"}
{"aliases": [], "authors": ["Yiqin Wang", "Haoji Zhang", "Jingqi Tian", "Yansong Tang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6a4c750017de421d3ca9bf119834dfb96d2fe8a0610f12b30d2fae603e26c66b", "submitted_at": "2024-12-02T08:35:31Z", "updated_at": "2024-12-02T08:35:31Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2412.01268v1"}, "publication": {"authors": ["Yiqin Wang", "Haoji Zhang", "Jingqi Tian", "Yansong Tang"], "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.76", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2025.findings-acl.76/"], "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": "1461-1473", "pdf_url": "https://aclanthology.org/2025.findings-acl.76.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "Ponder & Press: Advancing Visual GUI Agent towards General Computer Control", "type": "conference-paper", "url": "https://aclanthology.org/2025.findings-acl.76/", "venue": "Findings of the Association for Computational Linguistics: ACL 2025", "volume": null}}, "environments": ["cross-platform", "web", "desktop"], "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": "0ec2dd7116411e80f842627d047d06cf60d5921d91350daeb251a9c95a8f7884", "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.76/"}, {"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:2412.01268", "identifiers": {"arxiv": "2412.01268"}, "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/2412.01268"}, {"rel": "publication", "url": "https://aclanthology.org/2025.findings-acl.76/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2025.findings-acl.76.pdf"}], "name": "Ponder & Press: Advancing Visual GUI Agent towards General Computer 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": "2024-12-02", "relations": [], "review_status": "curated", "scope": "core", "summary": "Separates interpretation of user instructions from precise GUI localization using general-purpose and specialized visual-language components.", "topics": ["grounding", "planning"], "updated_at": "2024-12-02", "url": "https://arxiv.org/abs/2412.01268"}
{"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": ["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": ["Yuhang Liu", "Pengxiang Li", "Zishu Wei", "Congkai Xie", "Xueyu Hu", "Xinchen Xu", "Shengyu Zhang", "Xiaotian Han", "Hongxia Yang", "Fei Wu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "18d5b2e34f18c24e5d60d5564c9b08eced39c1b17bf5dac8d997640786688fba", "submitted_at": "2025-01-08T15:45:21Z", "updated_at": "2025-01-08T15:45:21Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2501.04575v1"}}, "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:2501.04575", "identifiers": {"arxiv": "2501.04575"}, "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/2501.04575"}], "name": "InfiGUIAgent: A Multimodal Generalist GUI Agent with Native Reasoning and Reflection", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-01-08", "relations": [], "review_status": "curated", "scope": "core", "summary": "Uses staged supervised training to combine GUI understanding and grounding with hierarchical reasoning and expectation-based reflection.", "topics": ["grounding", "planning"], "updated_at": "2025-01-08", "url": "https://arxiv.org/abs/2501.04575"}
{"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": ["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": ["Qingyuan Wu", "Jianheng Liu", "Jianye Hao", "Jun Wang", "Kun Shao"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "186b7fd8e79db290ea481645d8adab3989a12f787a13ecb11270f4756f5af42a", "submitted_at": "2025-02-11T20:57:46Z", "updated_at": "2025-05-20T19:54:36Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2502.07949v2"}}, "environments": ["web", "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:2502.07949", "identifiers": {"arxiv": "2502.07949"}, "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/2502.07949"}], "name": "Advancing Autonomous VLM Agents via Variational Subgoal-Conditioned 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": "2025-02-11", "relations": [], "review_status": "curated", "scope": "core", "summary": "Uses variational subgoal-conditioned reinforcement learning to improve learning efficiency on sparse-reward mobile and web-control tasks.", "topics": ["reinforcement-learning", "planning"], "updated_at": "2025-05-20", "url": "https://arxiv.org/abs/2502.07949"}
{"aliases": [], "authors": ["Henry Hengyuan Zhao", "Kaiming Yang", "Wendi Yu", "Difei Gao", "Mike Zheng Shou"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "68e6ecfcdff0163debe96f6440fb33529356bf9fd2c8743b5953ae52ae71c9ef", "submitted_at": "2025-02-12T01:06:10Z", "updated_at": "2026-05-23T20:19:43Z", "version": 5, "versioned_url": "https://arxiv.org/abs/2502.08047v5"}}, "environments": ["desktop", "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:2502.08047", "identifiers": {"arxiv": "2502.08047"}, "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/2502.08047"}], "name": "WorldGUI: An Interactive Benchmark for Desktop GUI Automation from Any Starting Point", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-02-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces desktop and web tasks with varied initial states to test planning recovery and adaptability beyond canonical starting configurations.", "topics": ["evaluation", "robustness", "planning"], "updated_at": "2026-05-23", "url": "https://arxiv.org/abs/2502.08047"}
{"aliases": [], "authors": ["Haowei Liu", "Xi Zhang", "Haiyang Xu", "Yuyang Wanyan", "Junyang Wang", "Ming Yan", "Ji Zhang", "Chunfeng Yuan", "Changsheng Xu", "Weiming Hu", "Fei Huang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d75b417f4248397b73934f8b8e45199006e4db6e2ee69b62e0846621411b2dd7", "submitted_at": "2025-02-20T05:41:55Z", "updated_at": "2025-02-21T02:54:09Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2502.14282v2"}}, "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:2502.14282", "identifiers": {"arxiv": "2502.14282"}, "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/2502.14282"}], "name": "PC-Agent: A Hierarchical Multi-Agent Collaboration Framework for Complex Task Automation on PC", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-02-20", "relations": [], "review_status": "curated", "scope": "core", "summary": "Uses active perception and hierarchical agents for instruction decomposition, progress tracking, action selection, and reflection in PC workflows.", "topics": ["planning", "cross-app"], "updated_at": "2025-02-21", "url": "https://arxiv.org/abs/2502.14282"}
{"aliases": [], "authors": ["Zongru Wu", "Pengzhou Cheng", "Zheng Wu", "Tianjie Ju", "Zhuosheng Zhang", "Gongshen Liu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "53781edbb107437b258d0c0e298f80ec14f8f32b51bc0dd11c0417473eb2c606", "submitted_at": "2025-03-01T08:29:59Z", "updated_at": "2025-03-04T12:04:26Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2503.00401v2"}}, "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:2503.00401", "identifiers": {"arxiv": "2503.00401"}, "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/2503.00401"}], "name": "Smoothing Grounding and Reasoning for MLLM-Powered GUI Agents with Query-Oriented Pivot 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": "2025-03-01", "relations": [], "review_status": "curated", "scope": "core", "summary": "Uses inference of possible user queries from screenshots and target coordinates as a training bridge between GUI grounding and action reasoning.", "topics": ["grounding", "planning"], "updated_at": "2025-03-04", "url": "https://arxiv.org/abs/2503.00401"}
{"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": ["Weimin Xiong", "Yifan Song", "Qingxiu Dong", "Bingchan Zhao", "Feifan Song", "Xun Wang", "Sujian Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "3395107d8c7db3562bf75a3c22fee55d3da0564bd2f413491bc0d381fc601323", "submitted_at": "2025-03-04T14:54:45Z", "updated_at": "2025-09-10T16:45:42Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2503.02682v2"}}, "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": "e634a3e576c94f1f7f5409d28a5c4c80c46141f1763c81337bb45f19793815f6", "supports": ["environments", "scope", "notes"], "url": "https://arxiv.org/html/2503.02682"}], "id": "arxiv:2503.02682", "identifiers": {"arxiv": "2503.02682"}, "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/2503.02682"}], "name": "MPO: Boosting LLM Agents with Meta Plan Optimization", "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": "Refines reusable high-level meta plans with execution feedback to guide language-agent planning across interactive environments.", "topics": ["planning", "memory"], "updated_at": "2025-09-10", "url": "https://arxiv.org/abs/2503.02682"}
{"aliases": [], "authors": ["Danqing Zhang", "Balaji Rama", "Jingyi Ni", "Shiying He", "Fu Zhao", "Kunyu Chen", "Arnold Chen", "Junyu Cao"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "df682f883bed5e0e9917074dfa130839fabaa318c55714f26943d944774854fe", "submitted_at": "2025-03-04T19:13:10Z", "updated_at": "2025-05-06T06:42:34Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2503.02950v2"}, "publication": {"authors": ["Danqing Zhang", "Balaji Rama", "Jingyi Ni", "Shiying He", "Fu Zhao", "Kunyu Chen", "Arnold Chen", "Junyu Cao"], "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-demo.36", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2025.naacl-demo.36/"], "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": "449-455", "pdf_url": "https://aclanthology.org/2025.naacl-demo.36.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "LiteWebAgent: The Open-Source Suite for VLM-Based Web-Agent Applications", "type": "conference-paper", "url": "https://aclanthology.org/2025.naacl-demo.36/", "venue": "Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (System Demonstrations)", "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": "9b130228c049f324632f571dd9c9e74c9a293ee240351c7a6fd4fef3359afc12", "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-demo.36/"}, {"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:2503.02950", "identifiers": {"arxiv": "2503.02950"}, "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/2503.02950"}, {"rel": "publication", "url": "https://aclanthology.org/2025.naacl-demo.36/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2025.naacl-demo.36.pdf"}], "name": "LiteWebAgent: The Open-Source Suite for VLM-Based Web-Agent Applications", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-03-04", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces a modular visual web-agent stack with planning and memory components, exposed through a remote-browser application and Chrome extension.", "topics": ["web-navigation", "tool-integration", "planning"], "updated_at": "2025-05-06", "url": "https://arxiv.org/abs/2503.02950"}
{"aliases": [], "authors": ["Yuqi Zhou", "Shuai Wang", "Sunhao Dai", "Qinglin Jia", "Zhaocheng Du", "Zhenhua Dong", "Jun Xu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "380d00c46e5dd221b7c481a68b4fbdc00826e64e4b70a98f06a56f8c12b8f5e4", "submitted_at": "2025-03-05T18:56:16Z", "updated_at": "2025-03-05T18:56:16Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2503.03743v1"}}, "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", "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:2503.03743", "identifiers": {"arxiv": "2503.03743"}, "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/2503.03743"}], "name": "CHOP: Mobile Operating Assistant with Constrained High-frequency Optimized Subtask 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": "2025-03-05", "relations": [], "review_status": "curated", "scope": "core", "summary": "Uses human-planned subtasks to constrain and refine the intermediate plans connecting user goals with executable mobile operations.", "topics": ["mobile-navigation", "planning"], "updated_at": "2025-03-05", "url": "https://arxiv.org/abs/2503.03743"}
{"aliases": [], "authors": ["Fei Tang", "Yongliang Shen", "Hang Zhang", "Siqi Chen", "Guiyang Hou", "Wenqi Zhang", "Wenqiao Zhang", "Kaitao Song", "Weiming Lu", "Yueting Zhuang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "ebc104ee6a868092ae0215f84a3c5c71894bfb321671476a74b794e1645210c7", "submitted_at": "2025-03-09T06:14:17Z", "updated_at": "2025-03-09T06:14:17Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2503.06470v1"}}, "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:2503.06470", "identifiers": {"arxiv": "2503.06470"}, "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/2503.06470"}], "name": "Think Twice, Click Once: Enhancing GUI Grounding via Fast and Slow Systems", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-03-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "Combines rapid target prediction with deliberate interface analysis, adaptively selecting the grounding process according to task complexity.", "topics": ["grounding", "efficiency", "planning"], "updated_at": "2025-03-09", "url": "https://arxiv.org/abs/2503.06470"}
{"aliases": [], "authors": ["Di Zhao", "Longhui Ma", "Siwei Wang", "Miao Wang", "Zhao Lv"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "3ea50d595c5d82b4a5d538b4b20725fa1f5afe5b29b03c022b726c307c8cca6e", "submitted_at": "2025-03-12T11:03:27Z", "updated_at": "2025-03-12T11:03:27Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2503.09263v1"}}, "environments": ["desktop", "windows"], "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:2503.09263", "identifiers": {"arxiv": "2503.09263"}, "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/2503.09263"}], "name": "COLA: A Scalable Multi-Agent Framework For Windows UI Task 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": "2025-03-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "Coordinates specialized Windows UI agents with dynamic task scheduling, shared memory, and human-triggered rollback for recovering from errors.", "topics": ["planning", "memory", "human-in-the-loop"], "updated_at": "2025-03-12", "url": "https://arxiv.org/abs/2503.09263"}
{"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": ["Gaole Dai", "Shiqi Jiang", "Ting Cao", "Yuanchun Li", "Yuqing Yang", "Rui Tan", "Mo Li", "Lili Qiu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6449576cb8ba45569b57844d87f22d01645b3639a222c648d914dc38e3421553", "submitted_at": "2025-03-20T08:25:00Z", "updated_at": "2026-02-21T01:36:43Z", "version": 5, "versioned_url": "https://arxiv.org/abs/2503.15937v5"}}, "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:2503.15937", "identifiers": {"arxiv": "2503.15937"}, "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/2503.15937"}], "name": "Advancing Mobile GUI Agents: A Verifier-Driven Approach to Practical Deployment", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-03-20", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces V-Droid, which verifies candidate actions in a discrete action space rather than directly generating each mobile operation.", "topics": ["mobile-navigation", "efficiency", "planning"], "updated_at": "2026-02-21", "url": "https://arxiv.org/abs/2503.15937"}
{"aliases": [], "authors": ["Li Zhang", "Longxi Gao", "Mengwei Xu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e450216f8f802d8f3d16c5886097cef801d051951af0c57fdf195cc176ac744f", "submitted_at": "2025-03-21T01:52:43Z", "updated_at": "2025-03-21T01:52:43Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2503.16788v1"}}, "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"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%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:2503.16788", "identifiers": {"arxiv": "2503.16788"}, "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/2503.16788"}], "name": "Does Chain-of-Thought Reasoning Help Mobile GUI Agent? An Empirical Study", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-03-21", "relations": [], "review_status": "curated", "scope": "core", "summary": "This study compares matched reasoning and non-reasoning vision-language models on static mobile GUI tasks and interactive AndroidWorld navigation. It finds that chain-of-thought effects vary across tasks and model families.", "topics": ["evaluation", "mobile-navigation", "planning"], "updated_at": "2025-03-21", "url": "https://arxiv.org/abs/2503.16788"}
{"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": ["Ziming Cheng", "Zhiyuan Huang", "Junting Pan", "Zhaohui Hou", "Mingjie Zhan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "1bb09b54379d2fbc5b489f24a27f6311b91bb0821eefe5187aa160ffb00639d9", "submitted_at": "2025-03-31T14:56:24Z", "updated_at": "2025-08-24T08:39:18Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2503.24180v2"}}, "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": "2026-10-16T06:45:04Z", "role": "primary", "sha256": "bab88dfed88ca440c231aa768530374a0f19498777a70921481a904fddd46bb8", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "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:2503.24180", "identifiers": {"arxiv": "2503.24180"}, "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/2503.24180"}], "name": "Navi-plus: Managing Ambiguous GUI Navigation Tasks with Follow-up Questions", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-03-31", "relations": [], "review_status": "curated", "scope": "core", "summary": "Navi-plus evaluates GUI agents that ask follow-up questions to complete ambiguous user instructions. It provides GUI clarification question-answer pairs and Dual-Stream Trajectory Evaluation for this interactive setting.", "topics": ["human-in-the-loop", "evaluation", "planning"], "updated_at": "2025-08-24", "url": "https://arxiv.org/abs/2503.24180"}
{"aliases": [], "authors": ["Saaket Agashe", "Kyle Wong", "Vincent Tu", "Jiachen Yang", "Ang Li", "Xin Eric Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "11958f69503db514a91af8721870a4cbc1fa19dabf94e3f4ce4118d2fad9862a", "submitted_at": "2025-04-01T15:40:27Z", "updated_at": "2025-04-01T15:40:27Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2504.00906v1"}}, "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:2504.00906", "identifiers": {"arxiv": "2504.00906"}, "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/2504.00906"}], "name": "Agent S2: A Compositional Generalist-Specialist Framework for Computer Use Agents", "notes": "Editorial summary based on the primary paper abstract; experiments and released artifacts have not been independently reproduced.", "published_at": "2025-04-01", "relations": [], "review_status": "curated", "scope": "core", "summary": "Combines generalist and specialist models with hierarchical planning and multiple grounding methods for computer-use tasks.", "topics": ["grounding", "planning"], "updated_at": "2025-04-01", "url": "https://arxiv.org/abs/2504.00906"}
{"aliases": [], "authors": ["Boyuan Zheng", "Michael Y. Fatemi", "Xiaolong Jin", "Zora Zhiruo Wang", "Apurva Gandhi", "Yueqi Song", "Yu Gu", "Jayanth Srinivasa", "Gaowen Liu", "Graham Neubig", "Yu Su"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "5f0489433a41b49ab8ab42eea57503685ba60beadabea7822134e96d865a7435", "submitted_at": "2025-04-09T17:51:50Z", "updated_at": "2025-04-09T17:51:50Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2504.07079v1"}}, "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:2504.07079", "identifiers": {"arxiv": "2504.07079"}, "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/2504.07079"}], "name": "SkillWeaver: Web Agents can Self-Improve by Discovering and Honing Skills", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-04-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "SkillWeaver explores websites, practices interactions, and converts the resulting experience into a reusable library of callable skill APIs. Its WebArena and live-website experiments examine skill refinement, composition, and transfer from stronger agents to weaker ones.", "topics": ["web-navigation", "tool-integration", "planning"], "updated_at": "2025-04-09", "url": "https://arxiv.org/abs/2504.07079"}
{"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": ["Zhisong Zhang", "Tianqing Fang", "Kaixin Ma", "Wenhao Yu", "Hongming Zhang", "Haitao Mi", "Dong Yu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "76f84d5375ab7321512a1f4f8724c011cf180bf5a7a0d532ff3218cc9d7a6e75", "submitted_at": "2025-04-16T05:41:20Z", "updated_at": "2026-01-15T02:00:45Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2504.11788v3"}}, "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:2504.11788", "identifiers": {"arxiv": "2504.11788"}, "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/2504.11788"}], "name": "WebRollback: Enhancing Web Agents with Explicit Rollback Mechanisms", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-04-16", "relations": [], "review_status": "curated", "scope": "core", "summary": "WebRollback gives a web agent an explicit action for returning to an earlier state in its navigation trajectory. It studies this recovery mechanism on two live web-navigation benchmarks under both zero-shot and fine-tuned settings.", "topics": ["web-navigation", "planning", "robustness"], "updated_at": "2026-01-15", "url": "https://arxiv.org/abs/2504.11788"}
{"aliases": [], "authors": ["Dezhao Luo", "Bohan Tang", "Kang Li", "Georgios Papoudakis", "Jifei Song", "Shaogang Gong", "Jianye Hao", "Jun Wang", "Kun Shao"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "15e00237ef710c37c0c9f7768d699f57e9d5dcc190e8de7e20e9ec3b21b82750", "submitted_at": "2025-04-15T14:03:10Z", "updated_at": "2025-05-20T14:11:37Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2504.13936v2"}}, "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": "2026-10-16T06:45:04Z", "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:2504.13936", "identifiers": {"arxiv": "2504.13936"}, "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/2504.13936"}], "name": "ViMo: A Generative Visual GUI World Model for App 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": "2025-04-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "ViMo predicts the next mobile-app screenshot after a candidate action by separately modeling graphics and text. A symbolic placeholder representation preserves layout while a second predictor supplies readable text, enabling visual lookahead for GUI-agent planning.", "topics": ["world-models", "planning", "mobile-navigation"], "updated_at": "2025-05-20", "url": "https://arxiv.org/abs/2504.13936"}
{"aliases": [], "authors": ["Yuhang Liu", "Pengxiang Li", "Congkai Xie", "Xavier Hu", "Xiaotian Han", "Shengyu Zhang", "Hongxia Yang", "Fei Wu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "4ce03185cc73c2723f444bd9729c67da9aab7d96b5ef099fb78e47f58790bc18", "submitted_at": "2025-04-19T09:25:55Z", "updated_at": "2025-04-19T09:25:55Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2504.14239v1"}}, "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:2504.14239", "identifiers": {"arxiv": "2504.14239"}, "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/2504.14239"}], "name": "InfiGUI-R1: Advancing Multimodal GUI Agents from Reactive Actors to Deliberative Reasoners", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-04-19", "relations": [], "review_status": "curated", "scope": "core", "summary": "Uses a staged reasoning-focused training framework to move GUI agents from reactive actions toward deliberative planning and recovery.", "topics": ["planning", "reinforcement-learning"], "updated_at": "2025-04-19", "url": "https://arxiv.org/abs/2504.14239"}
{"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": ["Xiaoxi Li", "Jiajie Jin", "Guanting Dong", "Hongjin Qian", "Yongkang Wu", "Ji-Rong Wen", "Yutao Zhu", "Zhicheng Dou"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "351b6d7fbe4e217a5211884832c6ad07d76a0463e6a8c996815999ff3a5f1921", "submitted_at": "2025-04-30T16:25:25Z", "updated_at": "2025-10-13T12:40:15Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2504.21776v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-18T10:54:08Z", "role": "primary", "sha256": "00d275cd5407cec847fa9d43fe73cfc4384c5570b68a5e058c97921799be0bec", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2302.01560%2C2303.16563%2C2305.15486%2C2305.16291%2C2305.17144%2C2306.00003%2C2306.00349%2C2306.17882%2C2310.01557%2C2310.08235%2C2310.08367%2C2311.05997%2C2311.17600%2C2312.02519%2C2312.11865%2C2402.01118%2C2402.10196%2C2402.11208%2C2403.10249%2C2403.12037%2C2404.06647%2C2404.12241%2C2404.13246%2C2407.15325%2C2409.00138%2C2410.14573%2C2410.18927%2C2410.19012%2C2411.09523%2C2411.10349%2C2411.13543%2C2412.01547%2C2501.05366%2C2501.09004%2C2502.07463%2C2502.14538%2C2503.02732%2C2503.08175%2C2503.11387%2C2504.10147%2C2504.11703%2C2504.14064%2C2504.15810%2C2504.16208%2C2504.20635%2C2504.21776%2C2505.12981%2C2505.13076%2C2506.01616%2C2506.13205&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:54:08Z", "role": "primary", "sha256": "00d275cd5407cec847fa9d43fe73cfc4384c5570b68a5e058c97921799be0bec", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2302.01560%2C2303.16563%2C2305.15486%2C2305.16291%2C2305.17144%2C2306.00003%2C2306.00349%2C2306.17882%2C2310.01557%2C2310.08235%2C2310.08367%2C2311.05997%2C2311.17600%2C2312.02519%2C2312.11865%2C2402.01118%2C2402.10196%2C2402.11208%2C2403.10249%2C2403.12037%2C2404.06647%2C2404.12241%2C2404.13246%2C2407.15325%2C2409.00138%2C2410.14573%2C2410.18927%2C2410.19012%2C2411.09523%2C2411.10349%2C2411.13543%2C2412.01547%2C2501.05366%2C2501.09004%2C2502.07463%2C2502.14538%2C2503.02732%2C2503.08175%2C2503.11387%2C2504.10147%2C2504.11703%2C2504.14064%2C2504.15810%2C2504.16208%2C2504.20635%2C2504.21776%2C2505.12981%2C2505.13076%2C2506.01616%2C2506.13205&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T13:00:06.851520+00:00", "role": "primary", "sha256": "d06baa20c7775347cbb4492473418d11cbe56941ab90baa16381353ab6c39e07", "supports": ["summary", "scope", "notes"], "url": "https://arxiv.org/html/2504.21776v2"}], "id": "arxiv:2504.21776", "identifiers": {"arxiv": "2504.21776"}, "kind": "paper", "last_verified_at": "2026-10-18T10:54:08Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2504.21776"}], "name": "WebThinker: Empowering Large Reasoning Models with Deep Research Capability", "notes": "Adjacent web-research method. The retained abstract and full-text Section 3 and Appendix A describe search APIs and Crawl4AI URL retrieval; no visual GUI execution is inferred. Results were not independently reproduced.", "published_at": "2025-04-30", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "Combines iterative web search and link-following with interleaved reasoning and report drafting for deep research. Its reported web explorer retrieves page content through a crawler rather than controlling a visual browser interface.", "topics": ["web-navigation", "planning", "tool-integration"], "updated_at": "2025-10-13", "url": "https://arxiv.org/abs/2504.21776"}
{"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": ["Biao Yi", "Xavier Hu", "Yurun Chen", "Shengyu Zhang", "Hongxia Yang", "Fan Wu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "8f5fdf474c1cb0d7e9f42d617de27afae38a9fd8f5e0b07d1c86e082dc872a35", "submitted_at": "2025-05-08T17:31:20Z", "updated_at": "2025-11-17T09:12:00Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2505.05440v3"}}, "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:2505.05440", "identifiers": {"arxiv": "2505.05440"}, "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/2505.05440"}], "name": "EcoAgent: An Efficient Device-Cloud Collaborative Multi-Agent Framework for Mobile Automation", "notes": "Reviewed against the complete official arXiv abstract. Experimental results and linked artifact availability have not been independently reproduced.", "published_at": "2025-05-08", "relations": [], "review_status": "curated", "scope": "core", "summary": "EcoAgent splits mobile automation between cloud planning and on-device observation and verification. Compact screen descriptions and feedback reduce communication costs, with evaluation on AndroidWorld.", "topics": ["mobile-navigation", "planning", "efficiency"], "updated_at": "2025-11-17", "url": "https://arxiv.org/abs/2505.05440"}
{"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": ["Fanglin Mo", "Junzhe Chen", "Haoxuan Zhu", "Xuming Hu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "de6c501a42c3828bdf795ac59c38b982e738c38bd5d0c59cfb6cb10b51e39292", "submitted_at": "2025-05-20T09:45:55Z", "updated_at": "2025-05-20T09:45:55Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2505.14141v1"}}, "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:2505.14141", "identifiers": {"arxiv": "2505.14141"}, "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/2505.14141"}], "name": "Building a Stable Planner: An Extended Finite State Machine Based Planning Module for Mobile 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": "2025-05-20", "relations": [], "review_status": "curated", "scope": "core", "summary": "SPlanner models mobile-app functions and configurations with extended finite-state machines, traverses them to construct execution paths, and converts those paths into natural-language plans. A VLM follows these plans to perform GUI actions, with evaluation on AndroidWorld.", "topics": ["planning", "mobile-navigation"], "updated_at": "2025-05-20", "url": "https://arxiv.org/abs/2505.14141"}
{"aliases": [], "authors": ["Xiaoran Yin", "Xu Luo", "Hao Wu", "Lianli Gao", "Jingkuan Song"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "93c67222fb0f581e9f9cb3ff59c8d28faec9dc037180060ad5ac962b4469144c", "submitted_at": "2025-05-22T09:08:47Z", "updated_at": "2025-09-05T02:24:00Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2505.16422v3"}}, "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:2505.16422", "identifiers": {"arxiv": "2505.16422"}, "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/2505.16422"}], "name": "Unlocking Smarter Device Control: Foresighted Planning with a World Model-Driven Code Execution Approach", "notes": "Reviewed against the complete official arXiv abstract. Experimental results and linked artifact availability have not been independently reproduced.", "published_at": "2025-05-22", "relations": [], "review_status": "curated", "scope": "core", "summary": "FPWC builds and refines a task-oriented world model before planning mobile-device actions. It executes plans as code and evaluates the approach in simulated environments and on real mobile devices.", "topics": ["world-models", "planning", "mobile-navigation"], "updated_at": "2025-09-05", "url": "https://arxiv.org/abs/2505.16422"}
{"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": ["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": ["Linqiang Guo", "Wei Liu", "Yi Wen Heng", "Tse-Hsun", "Chen", "Yang Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6f0ec81429fb39e56414d3bfe621df6c6adf461e63fc1a1c8886ea2b6850b8ee", "submitted_at": "2025-05-29T16:08:51Z", "updated_at": "2025-11-19T16:48:16Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2505.23596v3"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T07:54:03Z", "role": "primary", "sha256": "13d5969845070cd2877dc51ae5300c2d623e5daef00c8a1552ffa46af3f721a5", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2401.05507%2C2406.07275%2C2409.16427%2C2410.00079%2C2503.07701%2C2503.22738%2C2505.15859%2C2505.17735%2C2505.23596%2C2507.05257%2C2507.06134%2C2507.21953%2C2507.23336%2C2508.06960%2C2508.08997%2C2509.02473%2C2509.23694%2C2510.09116%2C2510.14438%2C2511.18423%2C2512.12818%2C2512.13564%2C2601.10758%2C2601.12449%2C2601.21841%2C2603.01152%2C2603.19993%2C2603.20576%2C2604.04503%2C2604.17562%2C2604.23941%2C2605.16265%2C2606.06448%2C2606.06462%2C2606.29193%2C2607.01647%2C2607.13157&max_results=37"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T07:54:03Z", "role": "primary", "sha256": "13d5969845070cd2877dc51ae5300c2d623e5daef00c8a1552ffa46af3f721a5", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2401.05507%2C2406.07275%2C2409.16427%2C2410.00079%2C2503.07701%2C2503.22738%2C2505.15859%2C2505.17735%2C2505.23596%2C2507.05257%2C2507.06134%2C2507.21953%2C2507.23336%2C2508.06960%2C2508.08997%2C2509.02473%2C2509.23694%2C2510.09116%2C2510.14438%2C2511.18423%2C2512.12818%2C2512.13564%2C2601.10758%2C2601.12449%2C2601.21841%2C2603.01152%2C2603.19993%2C2603.20576%2C2604.04503%2C2604.17562%2C2604.23941%2C2605.16265%2C2606.06448%2C2606.06462%2C2606.29193%2C2607.01647%2C2607.13157&max_results=37"}], "id": "arxiv:2505.23596", "identifiers": {"arxiv": "2505.23596"}, "kind": "paper", "last_verified_at": "2026-10-16T07:54:03Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2505.23596"}], "name": "Agent-SAMA: State-Aware Mobile Assistant", "notes": "Hugging Face discovery used the label MAPLE for this ID; the record follows the current primary arXiv title, Agent-SAMA. The secondary-title discrepancy is not treated as proof of a historical rename. Experiments not independently reproduced.", "published_at": "2025-05-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "Models mobile application execution as a finite-state machine to support planning, verification, and recovery across interface transitions.", "topics": ["planning", "memory", "mobile-navigation"], "updated_at": "2025-11-19", "url": "https://arxiv.org/abs/2505.23596"}
{"aliases": [], "authors": ["Yuyang Wanyan", "Xi Zhang", "Haiyang Xu", "Haowei Liu", "Junyang Wang", "Jiabo Ye", "Yutong Kou", "Ming Yan", "Fei Huang", "Xiaoshan Yang", "Weiming Dong", "Changsheng Xu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "42be944224a9b7d2c8e7042968ab31d373a0bafa304c0e669728cb0f10cc813d", "submitted_at": "2025-06-05T04:12:36Z", "updated_at": "2025-11-17T07:48:59Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2506.04614v2"}}, "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:2506.04614", "identifiers": {"arxiv": "2506.04614"}, "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/2506.04614"}], "name": "Look Before You Leap: A GUI-Critic-R1 Model for Pre-Operative Error Diagnosis in GUI 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": "2025-06-05", "relations": [], "review_status": "curated", "scope": "core", "summary": "Uses a pre-execution critic and reinforcement training to evaluate proposed GUI actions and provide corrective suggestions before execution.", "topics": ["planning", "robustness"], "updated_at": "2025-11-17", "url": "https://arxiv.org/abs/2506.04614"}
{"aliases": [], "authors": ["Shengcheng Yu", "Yuchen Ling", "Chunrong Fang", "Quan Zhou", "Yi Zhao", "Chunyang Chen", "Shaomin Zhu", "Zhenyu Chen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "bad447e8bca06f720a9476492de06978c73a356a38b00671203f8af0b81f1952", "submitted_at": "2025-06-05T14:27:40Z", "updated_at": "2026-05-03T08:25:53Z", "version": 4, "versioned_url": "https://arxiv.org/abs/2506.05079v4"}}, "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:2506.05079", "identifiers": {"arxiv": "2506.05079"}, "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/2506.05079"}], "name": "Scenario-Guided LLM-based Mobile App GUI Testing", "notes": "Reviewed against the complete official arXiv abstract. Experimental results and linked artifact availability have not been independently reproduced.", "published_at": "2025-06-05", "relations": [], "review_status": "curated", "scope": "core", "summary": "ScenGen organizes mobile GUI testing around business scenarios rather than unrestricted exploration. Five collaborating agents observe the interface, choose and execute actions, check scenario completion, and record operations and runtime bugs.", "topics": ["automation", "mobile-navigation", "planning"], "updated_at": "2026-05-03", "url": "https://arxiv.org/abs/2506.05079"}
{"aliases": [], "authors": ["Zihan Deng", "Changyu Du", "Stavros Nousias", "André Borrmann"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2bf4f91de32fdf3bd9b81ba64ace02e4a7d02ff094794f30dfd560389f334cea", "submitted_at": "2025-06-08T16:45:31Z", "updated_at": "2025-06-30T08:31:07Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2506.07217v2"}}, "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:2506.07217", "identifiers": {"arxiv": "2506.07217"}, "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/2506.07217"}], "name": "BIMgent: Towards Autonomous Building Modeling via 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": "2025-06-08", "relations": [], "review_status": "curated", "scope": "core", "summary": "BIMgent plans software-specific workflows and executes GUI actions for building-information modeling. It evaluates multimodal building design and reconstruction tasks in professional authoring software.", "topics": ["industry", "planning", "workflow", "evaluation"], "updated_at": "2025-06-30", "url": "https://arxiv.org/abs/2506.07217"}
{"aliases": [], "authors": ["Junhong Shen", "Hao Bai", "Lunjun Zhang", "Yifei Zhou", "Amrith Setlur", "Shengbang Tong", "Diego Caples", "Nan Jiang", "Tong Zhang", "Ameet Talwalkar", "Aviral Kumar"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e4b6c8b297e88dae689ba2a48724a441698cd5309a3bbe1198f38343e88aceb1", "submitted_at": "2025-06-09T17:50:02Z", "updated_at": "2025-06-10T12:50:18Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2506.07976v2"}}, "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:2506.07976", "identifiers": {"arxiv": "2506.07976"}, "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/2506.07976"}], "name": "Thinking vs. Doing: Agents that Reason by Scaling Test-Time Interaction", "notes": "Reviewed against the complete official arXiv abstract. Experimental results and linked artifact availability have not been independently reproduced.", "published_at": "2025-06-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "Test-Time Interaction scales the interaction horizon so web agents can explore, backtrack and revise plans within a rollout. Its online reinforcement-learning curriculum adjusts rollout lengths and is evaluated on WebVoyager and WebArena.", "topics": ["reinforcement-learning", "web-navigation", "planning", "long-horizon"], "updated_at": "2025-06-10", "url": "https://arxiv.org/abs/2506.07976"}
{"aliases": [], "authors": ["Zefang Liu", "Yinzhu Quan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "43aca3b0b8a1a0d5729a4e90dc1516bbb68e3611de3220876cdb29afc81d847f", "submitted_at": "2025-06-09T18:39:48Z", "updated_at": "2026-05-11T08:25:57Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2506.08136v3"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T07:57:51Z", "role": "primary", "sha256": "b91987360225499418fcd3645d7d5effac91e8202bf7c7e17c9f9e31d25b5be8", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2506.01423%2C2506.03231%2C2506.04301%2C2506.05336%2C2506.07385%2C2506.07398%2C2506.08136%2C2506.08837%2C2506.10953%2C2506.11425%2C2506.11763%2C2506.13538%2C2506.14683%2C2506.15461%2C2506.18959%2C2506.19502%2C2506.20668%2C2506.21252%2C2506.21495%2C2506.21614%2C2507.01599%2C2507.02279%2C2507.02825%2C2507.04036%2C2507.07400%2C2507.09481%2C2507.09884%2C2507.13348%2C2507.14447%2C2507.16704%2C2507.19902%2C2507.21678%2C2507.22925%2C2508.01548%2C2508.02866%2C2508.04002%2C2508.04389%2C2508.05193%2C2508.05294%2C2508.05748%2C2508.07871%2C2508.07999%2C2508.08228%2C2508.09124%2C2508.09981%2C2508.15804%2C2508.16153%2C2508.17398%2C2508.18489%2C2508.19611&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:20:34Z", "role": "primary", "sha256": "2f1d53911fe997bbba01c9d8a2fa7002f18837492448e15056fd7a7a6c78f97d", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/abs/2506.08136v3"}], "id": "arxiv:2506.08136", "identifiers": {"arxiv": "2506.08136"}, "kind": "paper", "last_verified_at": "2026-10-17T11:20:34Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2506.08136"}], "name": "EconWebArena: Benchmarking Autonomous Agents on Economic Tasks in Realistic Web Environments", "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": "2025-06-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces EconWebArena, a curated benchmark of economic-data tasks requiring live website navigation, multimodal interpretation, and precise extraction. It evaluates visual grounding, planning, and interaction design within multistep workflows.", "topics": ["evaluation", "web-navigation", "planning"], "updated_at": "2026-05-11", "url": "https://arxiv.org/abs/2506.08136"}
{"aliases": [], "authors": ["Yuan Guo", "Tingjia Miao", "Zheng Wu", "Pengzhou Cheng", "Ming Zhou", "Zhuosheng Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "8dbc84aa582b94775d490cf84e17eff05be84af7dbcc2de7de61b886fa3c9178", "submitted_at": "2025-06-10T16:45:29Z", "updated_at": "2025-06-10T16:45:29Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2506.08972v1"}}, "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"], "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:2506.08972", "identifiers": {"arxiv": "2506.08972"}, "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/2506.08972"}], "name": "Atomic-to-Compositional Generalization for Mobile Agents with A New Benchmark and Scheduling System", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-06-10", "relations": [], "review_status": "curated", "scope": "core", "summary": "UI-NEXUS evaluates compositional mobile tasks in local utility apps and online Chinese and English service apps. AGENT-NEXUS decomposes those long tasks into self-contained atomic subtasks to study generalization beyond isolated screens or short action chains.", "topics": ["evaluation", "long-horizon", "planning", "mobile-navigation"], "updated_at": "2025-06-10", "url": "https://arxiv.org/abs/2506.08972"}
{"aliases": [], "authors": ["Yuquan Xie", "Zaijing Li", "Rui Shao", "Gongwei Chen", "Kaiwen Zhou", "Yinchuan Li", "Dongmei Jiang", "Liqiang Nie"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "9b353606ce29a5bc60d263e86827a825af46883e25da5bd290b48affa9dd3838", "submitted_at": "2025-06-12T06:21:19Z", "updated_at": "2025-06-12T06:21:19Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2506.10387v1"}}, "environments": ["android", "web", "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:2506.10387", "identifiers": {"arxiv": "2506.10387"}, "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/2506.10387"}], "name": "Mirage-1: Augmenting and Updating GUI Agent with Hierarchical Multimodal Skills", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-06-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "Mirage-1 organizes interaction trajectories into execution, core, and meta-skills, then uses them to guide Monte Carlo tree search during online GUI tasks. It introduces AndroidLH for long-horizon evaluation and tests the approach on Android and web-navigation benchmarks.", "topics": ["planning", "memory", "long-horizon", "mobile-navigation", "web-navigation"], "updated_at": "2025-06-12", "url": "https://arxiv.org/abs/2506.10387"}
{"aliases": [], "authors": ["Yining Hong", "Rui Sun", "Bingxuan Li", "Xingcheng Yao", "Maxine Wu", "Alexander Chien", "Da Yin", "Ying Nian Wu", "Zhecan James Wang", "Kai-Wei Chang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "3e8f714b234a1aad60d0306df23c90410aebfd257fb5906644d0382479e8f3d0", "submitted_at": "2025-06-18T17:58:17Z", "updated_at": "2025-07-29T22:40:49Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2506.15677v3"}}, "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", "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:2506.15677", "identifiers": {"arxiv": "2506.15677"}, "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/2506.15677"}], "name": "Embodied Web Agents: Bridging Physical-Digital Realms for Integrated Agent Intelligence", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental findings and artifact availability have not been independently reproduced. The physical component is a simulation; the supported catalog connection is its functional web interaction.", "published_at": "2025-06-18", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces environments that couple 3D physical-world simulations with functional web interfaces. The accompanying benchmark evaluates tasks requiring agents to coordinate online information and physical actions across cooking, navigation and shopping scenarios.", "topics": ["evaluation", "web-navigation", "planning"], "updated_at": "2025-07-29", "url": "https://arxiv.org/abs/2506.15677"}
{"aliases": [], "authors": ["Jinjie Wei", "Jiyao Liu", "Lihao Liu", "Ming Hu", "Junzhi Ning", "Mingcheng Li", "Weijie Yin", "Junjun He", "Xiao Liang", "Chao Feng", "Dingkang Yang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "f12d2ae58243b5f89a6a510e8f3ef4073a605c4bafdf63c37fd38db57ea84602", "submitted_at": "2025-06-22T06:30:52Z", "updated_at": "2026-07-21T05:30:48Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2506.17913v2"}}, "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:2506.17913", "identifiers": {"arxiv": "2506.17913"}, "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/2506.17913"}], "name": "Learning, Reasoning, Refinement: A Framework for Kahneman's Dual-System Intelligence in 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": "2025-06-22", "relations": [], "review_status": "curated", "scope": "core", "summary": "CogniGUI combines hierarchical visual parsing of interface elements with a GRPO-based grounding agent that compares alternative interaction paths. Its ScreenSeek benchmark evaluates navigation across applications, changing interface states, and consistency between interfaces.", "topics": ["grounding", "planning", "reinforcement-learning", "evaluation"], "updated_at": "2026-07-21", "url": "https://arxiv.org/abs/2506.17913"}
{"aliases": [], "authors": ["Imene Kerboua", "Sahar Omidi Shayegan", "Megh Thakkar", "Xing Han Lù", "Massimo Caccia", "Véronique Eglin", "Alexandre Aussem", "Jérémy Espinas", "Alexandre Lacoste"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "b8972688fb95908c024a85e3ffc258be61280ee8d22e6ab97dfb0c4997b83445", "submitted_at": "2025-06-30T19:24:45Z", "updated_at": "2025-06-30T19:24:45Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2507.00210v1"}}, "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:2507.00210", "identifiers": {"arxiv": "2507.00210"}, "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/2507.00210"}], "name": "LineRetriever: Planning-Aware Observation Reduction 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": "2025-06-30", "relations": [], "review_status": "curated", "scope": "core", "summary": "LineRetriever uses a language model to select observation lines that matter for upcoming web-navigation actions. It reduces DOM or accessibility-tree context by considering the planning horizon instead of relying solely on semantic similarity.", "topics": ["web-navigation", "efficiency", "planning"], "updated_at": "2025-06-30", "url": "https://arxiv.org/abs/2507.00210"}
{"aliases": [], "authors": ["Yan Yang", "Dongxu Li", "Yutong Dai", "Yuhao Yang", "Ziyang Luo", "Zirui Zhao", "Zhiyuan Hu", "Junzhe Huang", "Amrita Saha", "Zeyuan Chen", "Ran Xu", "Liyuan Pan", "Silvio Savarese", "Caiming Xiong", "Junnan Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "3c1f659a30e1a0bd04219b9383927ce72e5f5b0d20783a89b850036d087155a4", "submitted_at": "2025-07-08T08:52:18Z", "updated_at": "2025-10-03T23:50:19Z", "version": 5, "versioned_url": "https://arxiv.org/abs/2507.05791v5"}}, "environments": ["unspecified"], "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:2507.05791", "identifiers": {"arxiv": "2507.05791"}, "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/2507.05791"}], "name": "GTA1: GUI Test-time Scaling 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": "2025-07-08", "relations": [], "review_status": "curated", "scope": "core", "summary": "GTA1 samples multiple action proposals at each GUI step and uses a judge to select among them. A separate grounding model is trained with reinforcement learning to map selected proposals to visual targets, linking planning-time search with click localization.", "topics": ["planning", "grounding", "reinforcement-learning"], "updated_at": "2025-10-03", "url": "https://arxiv.org/abs/2507.05791"}
{"aliases": [], "authors": ["Yuheng Zhao", "Xueli Shu", "Liwen Fan", "Lin Gao", "Yu Zhang", "Siming Chen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e3c271f330c5bf27e2831efd96fb1d6f843bd79942791686a866df1b4812de2f", "submitted_at": "2025-07-24T08:02:35Z", "updated_at": "2025-07-24T08:02:35Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2507.18165v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-18T10:01:16.030926+00:00", "role": "primary", "sha256": "4dcd1ba66ba21c0f2cb2aeb00c8bcab7c0cee4da37ff2fa6c16b45066aaa9db0", "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+OR+ti%3A%22mobile+agent%22+OR+ti%3A%22UI+agent%22+OR+ti%3A%22desktop+agent%22+OR+ti%3A%22screen+grounding%22%29+AND+submittedDate%3A%5B200001010000+TO+202610180000%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:01:16.030926+00:00", "role": "primary", "sha256": "4dcd1ba66ba21c0f2cb2aeb00c8bcab7c0cee4da37ff2fa6c16b45066aaa9db0", "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+OR+ti%3A%22mobile+agent%22+OR+ti%3A%22UI+agent%22+OR+ti%3A%22desktop+agent%22+OR+ti%3A%22screen+grounding%22%29+AND+submittedDate%3A%5B200001010000+TO+202610180000%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:05:30.529800+00:00", "role": "primary", "sha256": "7ca9569113479791a0482f4c196002a8459f1826eb6a3c9fab8a701b33360593", "supports": ["summary", "scope", "environments", "notes"], "url": "https://arxiv.org/html/2507.18165v1"}], "id": "arxiv:2507.18165", "identifiers": {"arxiv": "2507.18165"}, "kind": "paper", "last_verified_at": "2026-10-18T10:01:16.030926+00:00", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2507.18165"}], "name": "ProactiveVA: Proactive Visual Analytics with LLM-Based UI Agent", "notes": "Adjacent work on embedded assistance in two instrumented visual-analytics systems. Its in-lab user-study evidence is context-specific, and the paper discusses clarification and intervention-control trade-offs. General GUI-agent performance is not inferred. Basis: complete retained abstract and source-reviewed sections of the exact-version paper (5.3 User study; 5.3.2 Observed issues; 6.1 Design trade-off; 6.2 Limitations); not an entire-paper reading or reproduction claim.", "published_at": "2025-07-24", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "Studies proactive assistance in visual-analytics interfaces, using interaction logs to detect support needs and provide context-aware suggestions and guidance.", "topics": ["human-in-the-loop", "planning"], "updated_at": "2025-07-24", "url": "https://arxiv.org/abs/2507.18165"}
{"aliases": [], "authors": ["Xuetian Chen", "Yinghao Chen", "Xinfeng Yuan", "Zhuo Peng", "Lu Chen", "Yuekeng Li", "Zhoujia Zhang", "Yingqian Huang", "Leyan Huang", "Jiaqing Liang", "Tianbao Xie", "Zhiyong Wu", "Qiushi Sun", "Biqing Qi", "Bowen Zhou"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "fcd5a34bf6187f48da8e08842678e1e21cd558b2bcc7aab2611ceee7bc0d0179", "submitted_at": "2025-07-25T10:14:53Z", "updated_at": "2025-07-25T10:14:53Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2507.19132v1"}}, "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:2507.19132", "identifiers": {"arxiv": "2507.19132"}, "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/2507.19132"}], "name": "OS-MAP: How Far Can Computer-Using Agents Go in Breadth and Depth?", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-07-25", "relations": [], "review_status": "curated", "scope": "core", "summary": "OS-MAP organizes 416 computer-use tasks across 15 applications by a five-level automation taxonomy and a hierarchy of user demands. The benchmark separates required autonomy from generalization scope to diagnose where agents struggle as task demands increase.", "topics": ["evaluation", "automation", "planning"], "updated_at": "2025-07-25", "url": "https://arxiv.org/abs/2507.19132"}
{"aliases": [], "authors": ["Xuehui Wang", "Zhenyu Wu", "JingJing Xie", "Zichen Ding", "Bowen Yang", "Zehao Li", "Zhaoyang Liu", "Qingyun Li", "Xuan Dong", "Zhe Chen", "Weiyun Wang", "Xiangyu Zhao", "Jixuan Chen", "Haodong Duan", "Tianbao Xie", "Chenyu Yang", "Shiqian Su", "Yue Yu", "Yuan Huang", "Yiqian Liu", "Xiao Zhang", "Yanting Zhang", "Xiangyu Yue", "Weijie Su", "Xizhou Zhu", "Wei Shen", "Jifeng Dai", "Wenhai Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "8b91f7bacbac917e374aa6f656dcfbcb2e220ab3227fa2cff295c63a1c005f9f", "submitted_at": "2025-07-25T17:59:26Z", "updated_at": "2025-07-25T17:59:26Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2507.19478v1"}}, "environments": ["web", "desktop", "windows", "macos", "linux", "ios", "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:2507.19478", "identifiers": {"arxiv": "2507.19478"}, "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/2507.19478"}], "name": "MMBench-GUI: Hierarchical Multi-Platform Evaluation Framework 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": "2025-07-25", "relations": [], "review_status": "curated", "scope": "core", "summary": "MMBench-GUI evaluates interface understanding, element grounding, task automation, and collaboration across six named desktop, mobile, and web platforms. It introduces the Efficiency-Quality Area metric for examining execution efficiency alongside task quality in online GUI automation.", "topics": ["evaluation", "grounding", "efficiency", "planning"], "updated_at": "2025-07-25", "url": "https://arxiv.org/abs/2507.19478"}
{"aliases": [], "authors": ["Yi Kong", "Dianxi Shi", "Guoli Yang", "Zhang ke-di", "Chenlin Huang", "Xiaopeng Li", "Songchang Jin"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "197dbb0e2734f30349260444fca80a8ac830a5d4000a677d6417f46509737475", "submitted_at": "2025-07-29T16:05:32Z", "updated_at": "2025-07-29T16:05:32Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2507.21953v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T07:54:03Z", "role": "primary", "sha256": "13d5969845070cd2877dc51ae5300c2d623e5daef00c8a1552ffa46af3f721a5", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2401.05507%2C2406.07275%2C2409.16427%2C2410.00079%2C2503.07701%2C2503.22738%2C2505.15859%2C2505.17735%2C2505.23596%2C2507.05257%2C2507.06134%2C2507.21953%2C2507.23336%2C2508.06960%2C2508.08997%2C2509.02473%2C2509.23694%2C2510.09116%2C2510.14438%2C2511.18423%2C2512.12818%2C2512.13564%2C2601.10758%2C2601.12449%2C2601.21841%2C2603.01152%2C2603.19993%2C2603.20576%2C2604.04503%2C2604.17562%2C2604.23941%2C2605.16265%2C2606.06448%2C2606.06462%2C2606.29193%2C2607.01647%2C2607.13157&max_results=37"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T07:54:03Z", "role": "primary", "sha256": "13d5969845070cd2877dc51ae5300c2d623e5daef00c8a1552ffa46af3f721a5", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2401.05507%2C2406.07275%2C2409.16427%2C2410.00079%2C2503.07701%2C2503.22738%2C2505.15859%2C2505.17735%2C2505.23596%2C2507.05257%2C2507.06134%2C2507.21953%2C2507.23336%2C2508.06960%2C2508.08997%2C2509.02473%2C2509.23694%2C2510.09116%2C2510.14438%2C2511.18423%2C2512.12818%2C2512.13564%2C2601.10758%2C2601.12449%2C2601.21841%2C2603.01152%2C2603.19993%2C2603.20576%2C2604.04503%2C2604.17562%2C2604.23941%2C2605.16265%2C2606.06448%2C2606.06462%2C2606.29193%2C2607.01647%2C2607.13157&max_results=37"}], "id": "arxiv:2507.21953", "identifiers": {"arxiv": "2507.21953"}, "kind": "paper", "last_verified_at": "2026-10-16T07:54:03Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2507.21953"}], "name": "MapAgent: Trajectory-Constructed Memory-Augmented Planning for Mobile Task Automation", "notes": "Editorial summary based on the primary paper abstract; experiments and released artifacts have not been independently reproduced.", "published_at": "2025-07-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "Builds reusable page memories from past mobile trajectories and retrieves them to support task planning and execution.", "topics": ["memory", "planning", "mobile-navigation"], "updated_at": "2025-07-29", "url": "https://arxiv.org/abs/2507.21953"}
{"aliases": [], "authors": ["Zihan Zheng", "Tianle Cui", "Taoran Wang", "Fengtao Wang", "Jiahui Pan", "Lewei He", "Qianglong Chen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "57d89909534af90cbc10b97a992238bde3ae976ac6cc97778b4345635cef552c", "submitted_at": "2025-08-02T11:53:41Z", "updated_at": "2026-04-20T02:44:32Z", "version": 4, "versioned_url": "https://arxiv.org/abs/2508.01330v4"}}, "environments": ["unspecified"], "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:2508.01330", "identifiers": {"arxiv": "2508.01330"}, "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/2508.01330"}], "name": "NaturalGAIA: A Verifiable Benchmark and Hierarchical Framework for Long-Horizon GUI 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": "2025-08-02", "relations": [], "review_status": "curated", "scope": "core", "summary": "NaturalGAIA builds verifiable long-horizon GUI tasks from human interaction intents while separating causal task structure from its verbal description. Its LightManus-Jarvis framework divides dynamic planning and context management from execution based on visual and structural interface information.", "topics": ["evaluation", "long-horizon", "planning"], "updated_at": "2026-04-20", "url": "https://arxiv.org/abs/2508.01330"}
{"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": ["Linxin Song", "Yutong Dai", "Viraj Prabhu", "Jieyu Zhang", "Taiwei Shi", "Li Li", "Junnan Li", "Silvio Savarese", "Zeyuan Chen", "Jieyu Zhao", "Ran Xu", "Caiming Xiong"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "b6e4272a69c1a93f22bd35032547ffb6021a79c97d179a1075e14aee69dea83a", "submitted_at": "2025-08-05T21:33:36Z", "updated_at": "2026-02-20T10:14:08Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2508.03923v3"}}, "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:2508.03923", "identifiers": {"arxiv": "2508.03923"}, "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/2508.03923"}], "name": "CoAct-1: Computer-using Multi-Agent System with Coding Actions", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-08-05", "relations": [], "review_status": "curated", "scope": "core", "summary": "CoAct-1 delegates subtasks between a GUI operator and a programmer that executes Python or Bash. Its OSWorld evaluation studies a hybrid action space in which scripts handle suitable file and data operations while visual interaction handles other steps.", "topics": ["tool-integration", "automation", "efficiency", "planning"], "updated_at": "2026-02-20", "url": "https://arxiv.org/abs/2508.03923"}
{"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": ["Jiarun Liu", "Chunhong Zhang", "Zheng Hu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2d574d4157b775639c08ed12813892584b011dd3c0271ca2586f9d6a61d00c14", "submitted_at": "2025-08-07T07:05:22Z", "updated_at": "2025-08-07T07:05:22Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2508.05081v1"}}, "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", "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:2508.05081", "identifiers": {"arxiv": "2508.05081"}, "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/2508.05081"}], "name": "Cognitive Duality for Adaptive Web Agents", "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": "2025-08-07", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces CogniWeb, which switches between fast reactive processing and slower deliberative reasoning according to task complexity. WebArena experiments examine this combination of learned behaviors and online planning.", "topics": ["planning", "efficiency", "web-navigation"], "updated_at": "2025-08-07", "url": "https://arxiv.org/abs/2508.05081"}
{"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": ["Tanvir Bhathal", "Asanshay Gupta"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "67da98d98fec647de9b482fabd8c7861063c0fdb91f1ec303fe8df759ee0f69d", "submitted_at": "2025-08-23T11:02:59Z", "updated_at": "2025-08-23T11:02:59Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2508.16987v1"}}, "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:2508.16987", "identifiers": {"arxiv": "2508.16987"}, "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/2508.16987"}], "name": "WebSight: A Vision-First Architecture for Robust 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": "2025-08-23", "relations": [], "review_status": "curated", "scope": "core", "summary": "WebSight uses a vision-language model trained on web-interface examples within a modular agent with planning, action, verification, and episodic memory components. It evaluates element clicking on Showdown Clicks and full browsing tasks on WebVoyager using visual observations.", "topics": ["web-navigation", "grounding", "planning", "memory"], "updated_at": "2025-08-23", "url": "https://arxiv.org/abs/2508.16987"}
{"aliases": [], "authors": ["Xin Wang", "Zhiyao Cui", "Hao Li", "Ya Zeng", "Chenxu Wang", "Ruiqi Song", "Yihang Chen", "Kun Shao", "Qiaosheng Zhang", "Jinzhuo Liu", "Siyue Ren", "Shuyue Hu", "Zhen Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "57d5fc98a2eb2232f237d71b34060df0acef89cfeaa4bc247580a4dda3223e0b", "submitted_at": "2025-08-25T13:57:02Z", "updated_at": "2025-08-25T13:57:02Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2508.18040v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-18T10:01:10.946679+00:00", "role": "primary", "sha256": "594b5261dd2039777d36c46cdb8525d973ad57a154b902ff1a171bcbed2c64cc", "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+OR+ti%3A%22mobile+agent%22+OR+ti%3A%22UI+agent%22+OR+ti%3A%22desktop+agent%22+OR+ti%3A%22screen+grounding%22%29+AND+submittedDate%3A%5B200001010000+TO+202610180000%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:01:10.946679+00:00", "role": "primary", "sha256": "594b5261dd2039777d36c46cdb8525d973ad57a154b902ff1a171bcbed2c64cc", "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+OR+ti%3A%22mobile+agent%22+OR+ti%3A%22UI+agent%22+OR+ti%3A%22desktop+agent%22+OR+ti%3A%22screen+grounding%22%29+AND+submittedDate%3A%5B200001010000+TO+202610180000%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:18:36.598801+00:00", "role": "primary", "sha256": "3c3474b771af7bdd500e6aaa91b83d39ec0d455e59bb83df906e7a94700ad2b7", "supports": ["summary", "scope", "environments", "notes"], "url": "https://arxiv.org/pdf/2508.18040v1"}], "id": "arxiv:2508.18040", "identifiers": {"arxiv": "2508.18040"}, "kind": "paper", "last_verified_at": "2026-10-18T10:01:10.946679+00:00", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2508.18040"}], "name": "PerPilot: Personalizing VLM-based Mobile Agents via Memory and Exploration", "notes": "Personalization depends on the validity of stored or inferred user information and retrieval access. This is instruction completion with interaction, not demonstrated permanent model-weight personalization. No specific operating system or reproduced result is asserted. Basis: complete retained abstract and source-reviewed sections of the exact-version paper (The Framework of PerPilot; Personalization Perception; Personalization Completion; PerInstruct Data Construction); not an entire-paper reading or reproduction claim.", "published_at": "2025-08-25", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces PerPilot and the PerInstruct dataset for completing user-specific mobile instructions through memory retrieval, app exploration, and clarification when missing context remains unresolved.", "topics": ["memory", "planning", "human-in-the-loop", "mobile-navigation"], "updated_at": "2025-08-25", "url": "https://arxiv.org/abs/2508.18040"}
{"aliases": [], "authors": ["Yuyang Zhao", "Wentao Shi", "Fuli Feng", "Xiangnan He"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "02f4a76ae220caa289b322b622be6356086c5b04bc0f1598d0cd580c9ee25b6d", "submitted_at": "2025-08-26T05:23:24Z", "updated_at": "2025-08-27T04:25:35Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2508.18689v2"}}, "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:2508.18689", "identifiers": {"arxiv": "2508.18689"}, "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/2508.18689"}], "name": "AppAgent-Pro: A Proactive GUI Agent System for Multidomain Information Integration and User Assistance", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-08-26", "relations": [], "review_status": "curated", "scope": "core", "summary": "Proposes a proactive GUI assistant that gathers information across domains in response to user requests and anticipated information needs.", "topics": ["planning", "workflow"], "updated_at": "2025-08-27", "url": "https://arxiv.org/abs/2508.18689"}
{"aliases": [], "authors": ["Zeyi Sun", "Yuhang Cao", "Jianze Liang", "Qiushi Sun", "Ziyu Liu", "Zhixiong Zhang", "Yuhang Zang", "Xiaoyi Dong", "Kai Chen", "Dahua Lin", "Jiaqi Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "ac401f8d5c56050e72839f3bffba4f2c78bbdf34a00cab05fe504f03efd96a19", "submitted_at": "2025-08-27T17:59:50Z", "updated_at": "2025-08-27T17:59:50Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2508.20096v1"}}, "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:2508.20096", "identifiers": {"arxiv": "2508.20096"}, "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/2508.20096"}], "name": "CODA: Coordinating the Cerebrum and Cerebellum for a Dual-Brain Computer Use Agent with Decoupled 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": "2025-08-27", "relations": [], "review_status": "curated", "scope": "core", "summary": "CODA combines a generalist planner with a specialist GUI executor and trains application-specific planners through decoupled GRPO. Successful specialist trajectories then support supervised training of a consolidated planner, evaluated across four ScienceBoard applications.", "topics": ["reinforcement-learning", "planning", "science", "imitation-learning"], "updated_at": "2025-08-27", "url": "https://arxiv.org/abs/2508.20096"}
{"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": ["Ziyi Guan", "Jason Chun Lok Li", "Zhijian Hou", "Pingping Zhang", "Donglai Xu", "Yuzhi Zhao", "Mengyang Wu", "Jinpeng Chen", "Thanh-Toan Nguyen", "Pengfei Xian", "Wenao Ma", "Shengchao Qin", "Graziano Chesi", "Ngai Wong"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "c22f63bb5de22dd922ba37ea45df9404def017fee492e8ade03c83ab8b4f9b58", "submitted_at": "2025-08-30T05:32:32Z", "updated_at": "2025-08-30T05:32:32Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2509.00366v1"}}, "environments": ["android", "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:2509.00366", "identifiers": {"arxiv": "2509.00366"}, "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/2509.00366"}], "name": "KG-RAG: Enhancing GUI Agent Decision-Making via Knowledge Graph-Driven Retrieval-Augmented 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": "2025-08-30", "relations": [], "review_status": "curated", "scope": "core", "summary": "KG-RAG converts UI transition graphs into a searchable vector store and uses intent-guided retrieval to construct actionable navigation paths. It introduces KG-Android-Bench and KG-Harmony-Bench and evaluates transfer from mobile apps to web and desktop tasks.", "topics": ["memory", "planning", "mobile-navigation", "evaluation"], "updated_at": "2025-08-30", "url": "https://arxiv.org/abs/2509.00366"}
{"aliases": [], "authors": ["Jaewoo Ahn", "Junseo Kim", "Heeseung Yun", "Jaehyeon Son", "Dongmin Park", "Jaewoong Cho", "Gunhee Kim"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "3bf86652aaeb252972f3010be38b75b78dfaa9a8063c480c80d0cea47397c2e7", "submitted_at": "2025-09-01T01:33:16Z", "updated_at": "2025-10-15T10:33:27Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2509.01052v2"}}, "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:2509.01052", "identifiers": {"arxiv": "2509.01052"}, "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/2509.01052"}], "name": "FlashAdventure: A Benchmark for GUI Agents Solving Full Story Arcs in Diverse Adventure Games", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-09-01", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "FlashAdventure evaluates GUI agents on completing story arcs in 34 adventure games. It pairs an automated gameplay judge with COAST, a framework that keeps long-term clue memory to plan and complete sequential objectives.", "topics": ["evaluation", "long-horizon", "memory", "planning"], "updated_at": "2025-10-15", "url": "https://arxiv.org/abs/2509.01052"}
{"aliases": [], "authors": ["Jingru Fan", "Yufan Dang", "Jingyao Wu", "Huatao Li", "Runde Yang", "Xiyuan Yang", "Yuheng Wang", "Chen Qian"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "49a887b12b3395821aa100a66e61aac808309ef947ae752475df3d2db9a4fcd8", "submitted_at": "2025-09-02T15:48:21Z", "updated_at": "2025-10-17T00:57:58Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2509.02444v2"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-18T10:01:10.946679+00:00", "role": "primary", "sha256": "594b5261dd2039777d36c46cdb8525d973ad57a154b902ff1a171bcbed2c64cc", "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+OR+ti%3A%22mobile+agent%22+OR+ti%3A%22UI+agent%22+OR+ti%3A%22desktop+agent%22+OR+ti%3A%22screen+grounding%22%29+AND+submittedDate%3A%5B200001010000+TO+202610180000%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:01:10.946679+00:00", "role": "primary", "sha256": "594b5261dd2039777d36c46cdb8525d973ad57a154b902ff1a171bcbed2c64cc", "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+OR+ti%3A%22mobile+agent%22+OR+ti%3A%22UI+agent%22+OR+ti%3A%22desktop+agent%22+OR+ti%3A%22screen+grounding%22%29+AND+submittedDate%3A%5B200001010000+TO+202610180000%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:14:33.619660+00:00", "role": "primary", "sha256": "6c2591dad8893f87cffdea211584a3733dc2137dafa9416c1885d1b0126fd540", "supports": ["summary", "scope", "environments", "notes"], "url": "https://arxiv.org/pdf/2509.02444v2"}], "id": "arxiv:2509.02444", "identifiers": {"arxiv": "2509.02444"}, "kind": "paper", "last_verified_at": "2026-10-18T10:01:10.946679+00:00", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2509.02444"}], "name": "AppCopilot: Toward General, Accurate, Long-Horizon, and Efficient Mobile Agent", "notes": "Trace reuse assumes stable layouts and fixed tasks. The described function-call routes are user-selected, and coordinate calibration depends on valid detected widgets. No particular operating-system coverage, latency guarantee or reproduced result is asserted. Basis: complete retained abstract and source-reviewed sections of the exact-version paper (4.2 OCR/OR Region Localization; 4.2.2 Coordinate Calibration; 4.3 Multi-Agent Voting; 5.2 Task Decomposition; 6.2 Experience Reuse; 6.3 Function Calling); not an entire-paper reading or reproduction claim.", "published_at": "2025-09-02", "relations": [], "review_status": "curated", "scope": "core", "summary": "Describes a mobile-agent stack combining region-based coordinate correction, hierarchical task planning, multi-agent voting, and reuse of successful interaction traces.", "topics": ["planning", "grounding", "workflow", "mobile-navigation", "long-horizon"], "updated_at": "2025-10-17", "url": "https://arxiv.org/abs/2509.02444"}
{"aliases": [], "authors": ["Liangtao Lin", "Zhaomeng Zhu", "Tianwei Zhang", "Yonggang Wen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2c1a4019df77278943829e16b155771abb1c8ca30726044cf5969227e0437f5f", "submitted_at": "2025-09-17T05:14:11Z", "updated_at": "2025-09-17T05:14:11Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2509.13704v1"}}, "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:2509.13704", "identifiers": {"arxiv": "2509.13704"}, "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/2509.13704"}], "name": "InfraMind: A Novel Exploration-based GUI Agentic Framework for Mission-critical Industrial 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": "2025-09-17", "relations": [], "review_status": "curated", "scope": "core", "summary": "InfraMind explores industrial management GUIs using virtual-machine snapshots and builds memory for planning and hierarchical state recognition. Its framework also studies knowledge distillation and safety checks on open-source and commercial data-center management platforms.", "topics": ["industry", "planning", "memory", "safety-security"], "updated_at": "2025-09-17", "url": "https://arxiv.org/abs/2509.13704"}
{"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": ["Gaole Dai", "Shiqi Jiang", "Ting Cao", "Yuqing Yang", "Yuanchun Li", "Rui Tan", "Mo Li", "Lili Qiu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "c96f310fb0f28888c59dce71d08697b603e7a7562fdf99b51e3504b93e9e81b3", "submitted_at": "2025-09-26T03:29:36Z", "updated_at": "2026-04-15T09:19:46Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2509.21823v2"}}, "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:2509.21823", "identifiers": {"arxiv": "2509.21823"}, "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/2509.21823"}], "name": "ProRe: A Proactive Reward System for GUI Agents via Reasoner-Actor Collaboration", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-09-26", "relations": [], "review_status": "curated", "scope": "core", "summary": "ProRe evaluates GUI task outcomes by having a reasoner assign targeted environment-probing tasks to specialist actors. The additional observations support reward judgments, with separate evaluation of judge accuracy and downstream agent task success.", "topics": ["reward-models", "evaluation", "planning"], "updated_at": "2026-04-15", "url": "https://arxiv.org/abs/2509.21823"}
{"aliases": [], "authors": ["Zhecheng Li", "Guoxian Song", "Yiwei Wang", "Zhen Xiong", "Junsong Yuan", "Yujun Cai"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "9b42e784de8dd007c60d234b7331bd98b5a04c58c13696bf436c81c7eccee37f", "submitted_at": "2025-09-29T00:06:31Z", "updated_at": "2025-09-29T00:06:31Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2509.24133v1"}}, "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:2509.24133", "identifiers": {"arxiv": "2509.24133"}, "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/2509.24133"}], "name": "Generalist Scanner Meets Specialist Locator: A Synergistic Coarse-to-Fine Framework for Robust GUI Grounding", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-09-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "GMS pairs a general vision-language scanner that proposes interface regions with a specialist locator that predicts coordinates within them. Hierarchical coarse-to-fine search coordinates the two components for GUI grounding on ScreenSpot-Pro.", "topics": ["grounding", "planning", "evaluation"], "updated_at": "2025-09-29", "url": "https://arxiv.org/abs/2509.24133"}
{"aliases": [], "authors": ["Ran Xu", "Kaixin Ma", "Wenhao Yu", "Hongming Zhang", "Joyce C. Ho", "Carl Yang", "Dong Yu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e1c6fd3e84212bac4c4c8341ffb3478c97becce08f92dac89e5de77ad42cd07d", "submitted_at": "2025-09-29T02:04:20Z", "updated_at": "2025-09-29T02:04:20Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2509.24183v1"}}, "environments": ["web", "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", "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": "2026-10-17T09:20:58+00:00", "role": "primary", "sha256": "fe9f70994c119a85cb8075461174e9699c423d79094e611f1ac95a51666600c1", "supports": ["summary", "environments", "scope"], "url": "https://arxiv.org/html/2509.24183v1"}], "id": "arxiv:2509.24183", "identifiers": {"arxiv": "2509.24183"}, "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/2509.24183"}], "name": "Retrieval-augmented GUI Agents with Generative Guidelines", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-09-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "RAG-GUI trains a small guidance adapter to filter and summarize retrieved web tutorials while keeping the acting agent fixed. Supervised and rejection-sampling stages prepare the adapter for inference-time assistance, tested on offline web/mobile tasks and interactive Android tasks.", "topics": ["memory", "planning", "imitation-learning", "web-navigation", "mobile-navigation"], "updated_at": "2025-09-29", "url": "https://arxiv.org/abs/2509.24183"}
{"aliases": [], "authors": ["Zikang Liu", "Junyi Li", "Wayne Xin Zhao", "Dawei Gao", "Yaliang Li", "Ji-rong Wen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "277315ecc1a341c0a511bce84b87190fc0aeeb2306666ea8d4746aa8e9b39984", "submitted_at": "2025-10-01T01:48:39Z", "updated_at": "2025-10-04T15:23:55Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2510.00413v2"}}, "environments": ["web", "cross-platform"], "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:2510.00413", "identifiers": {"arxiv": "2510.00413"}, "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/2510.00413"}], "name": "PAL-UI: Planning with Active Look-back for Vision-Based GUI Agents", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-10-01", "relations": [], "review_status": "curated", "scope": "core", "summary": "PAL-UI combines observation/action summaries with selective retrieval of historical screenshots during planning. It trains mobile navigation models from step-level supervision and evaluates transfer to web navigation without additional training.", "topics": ["memory", "planning", "interaction-history", "mobile-navigation"], "updated_at": "2025-10-04", "url": "https://arxiv.org/abs/2510.00413"}
{"aliases": [], "authors": ["Bin Lei", "Nuo Xu", "Ali Payani", "Mingyi Hong", "Chunhua Liao", "Yu Cao", "Caiwen Ding"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "798fc5ca4fe483a1235f0979b0c641e9c1732660780c13340c5221319a1b28fc", "submitted_at": "2025-10-05T05:15:45Z", "updated_at": "2025-10-05T05:15:45Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2510.04039v1"}}, "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"], "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:2510.04039", "identifiers": {"arxiv": "2510.04039"}, "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/2510.04039"}], "name": "\\textsc{GUI-Spotlight}: Adaptive Iterative Focus Refinement for Enhanced GUI Visual Grounding", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-10-05", "relations": [], "review_status": "curated", "scope": "core", "summary": "GUI-Spotlight trains a grounding model to reason over images and call tools that iteratively narrow its visual focus. The paper evaluates this refinement method on ScreenSpot-Pro after training on 18,500 samples.", "topics": ["grounding", "planning", "tool-integration"], "updated_at": "2025-10-05", "url": "https://arxiv.org/abs/2510.04039"}
{"aliases": [], "authors": ["Yuan Wang", "Mingyu Li", "Haibo Chen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "1c22547bf82eb2fb87e8bf6a3846d25c17cac296763726ed9a25faf88bdca33c", "submitted_at": "2025-10-06T09:14:58Z", "updated_at": "2026-03-25T11:29:08Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2510.04607v2"}}, "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", "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:2510.04607", "identifiers": {"arxiv": "2510.04607"}, "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/2510.04607"}], "name": "From Imperative to Declarative: Towards LLM-friendly OS Interfaces for Boosted Computer-Use Agents", "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": "2025-10-06", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces a Declarative Model Interface that maps existing GUIs to access, state and observation primitives while handling low-level interactions. Evaluation uses Word, PowerPoint and Excel on Windows with a GUI-agent baseline.", "topics": ["automation", "tool-integration", "planning"], "updated_at": "2026-03-25", "url": "https://arxiv.org/abs/2510.04607"}
{"aliases": [], "authors": ["Shang Ma", "Xusheng Xiao", "Yanfang Ye"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "4bf0a561460c080b14da6fd13f6d92781e01e52b08dc691fe019e70b88e44c86", "submitted_at": "2025-10-07T15:36:04Z", "updated_at": "2026-09-27T17:25:57Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2510.06042v3"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-18T10:01:10.946679+00:00", "role": "primary", "sha256": "594b5261dd2039777d36c46cdb8525d973ad57a154b902ff1a171bcbed2c64cc", "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+OR+ti%3A%22mobile+agent%22+OR+ti%3A%22UI+agent%22+OR+ti%3A%22desktop+agent%22+OR+ti%3A%22screen+grounding%22%29+AND+submittedDate%3A%5B200001010000+TO+202610180000%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:01:10.946679+00:00", "role": "primary", "sha256": "594b5261dd2039777d36c46cdb8525d973ad57a154b902ff1a171bcbed2c64cc", "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+OR+ti%3A%22mobile+agent%22+OR+ti%3A%22UI+agent%22+OR+ti%3A%22desktop+agent%22+OR+ti%3A%22screen+grounding%22%29+AND+submittedDate%3A%5B200001010000+TO+202610180000%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:14:29.861666+00:00", "role": "primary", "sha256": "5e2cb2f053983140904ec15c097d060ff18315c6c3c956bbcacbec26921709c5", "supports": ["summary", "scope", "environments", "notes"], "url": "https://arxiv.org/pdf/2510.06042v3"}], "id": "arxiv:2510.06042", "identifiers": {"arxiv": "2510.06042"}, "kind": "paper", "last_verified_at": "2026-10-18T10:01:10.946679+00:00", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2510.06042"}], "name": "Agent+P: Guiding UI Agents via Symbolic Planning", "notes": "Planning guarantees are relative to the modeled graph and selected target. They do not establish complete graphs, correct grounding or faithful goal translation. The described design depends on static source analysis; no experiments were reproduced. Basis: complete retained abstract and source-reviewed sections of the exact-version paper (3.3 UI Automation to Classical Planning; 4.1 UTG Builder; 4.2 Node Selector; 4.3 Plan Generator; 4.4 UI Explorer; 5.1 Datasets); not an entire-paper reading or reproduction claim.", "published_at": "2025-10-07", "relations": [], "review_status": "curated", "scope": "core", "summary": "Guides UI agents with symbolic planning over a UI transition graph initialized from app-code analysis and updated through online exploration, with evaluation on AndroidWorld.", "topics": ["planning", "long-horizon", "mobile-navigation"], "updated_at": "2026-09-27", "url": "https://arxiv.org/abs/2510.06042"}
{"aliases": [], "authors": ["Kai Mei", "Jiang Guo", "Shuaichen Chang", "Mingwen Dong", "Dongkyu Lee", "Xing Niu", "Jiarong Jiang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "7c37dae372655e8ef556163744e47ed42db3533175ca7d1db7560033774ed2a0", "submitted_at": "2025-10-13T19:52:04Z", "updated_at": "2026-03-09T15:25:16Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2510.11892v2"}}, "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:2510.11892", "identifiers": {"arxiv": "2510.11892"}, "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/2510.11892"}], "name": "R-WoM: Retrieval-augmented World Model 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": "2025-10-13", "relations": [], "review_status": "curated", "scope": "core", "summary": "R-WoM evaluates how well world models predict nearby GUI states, align complete procedures, and recognize milestone transitions. It retrieves external tutorials to guide simulated future interactions, with experiments on subsets of OSWorld and WebArena.", "topics": ["world-models", "memory", "planning", "evaluation"], "updated_at": "2026-03-09", "url": "https://arxiv.org/abs/2510.11892"}
{"aliases": [], "authors": ["Zhe Wu", "Hongjin Lu", "Junliang Xing", "Changhao Zhang", "Yuxuan Li", "Yin Zhu", "Yuhao Yang", "Yuheng Jing", "Kai Li", "Kun Shao", "Jianye Hao", "Jun Wang", "Yuanchun Shi"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "95fd3e1286ed49d569b05644d77f822ea9747cfb4769cb7cb93f39c9271cf57e", "submitted_at": "2025-10-16T07:38:21Z", "updated_at": "2026-02-06T02:04:25Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2510.14388v2"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T07:57:59Z", "role": "primary", "sha256": "240e3d6528900f4a3e87f12f5e55cb7654a4d25d4478e291e9bbf725e7013118", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2510.10587%2C2510.10909%2C2510.11144%2C2510.12798%2C2510.14388%2C2510.14513%2C2510.14969%2C2510.16156%2C2510.17017%2C2510.17800%2C2510.19207%2C2510.19247%2C2510.19386%2C2510.19864%2C2510.20333%2C2510.21652%2C2510.22443%2C2510.22936%2C2510.23730%2C2510.23883%2C2510.24702%2C2510.25726%2C2510.26328%2C2510.27135%2C2510.27410%2C2511.00153%2C2511.02650%2C2511.02687%2C2511.03690%2C2511.04220%2C2511.05797%2C2511.06251%2C2511.08319%2C2511.08487%2C2511.10110%2C2511.13725%2C2511.14210%2C2511.14945%2C2511.15705%2C2511.16842%2C2511.17649%2C2511.17775%2C2511.19496%2C2511.19516%2C2511.20636%2C2511.21990%2C2511.22074%2C2512.04302%2C2512.04529%2C2512.06726&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:20:56Z", "role": "primary", "sha256": "fecd42f8ca75572410c354c892734cb8eefbd3fa622064f6bc6db0f535cf5607", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/abs/2510.14388v2"}], "id": "arxiv:2510.14388", "identifiers": {"arxiv": "2510.14388"}, "kind": "paper", "last_verified_at": "2026-10-17T11:20:56Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2510.14388"}], "name": "Hi-Agent: Hierarchical Vision-Language Agents for Mobile Device Control", "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. This Hi-Agent has a distinct author team and method from the earlier HiAgent working-memory paper; the shared project-name prefix is not treated as an identity match.", "published_at": "2025-10-16", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces Hi-Agent, jointly training a high-level reasoning model and a low-level mobile action model. Execution feedback supplies a foresight advantage for optimizing subgoal decisions, with evaluations on Android control benchmarks.", "topics": ["planning", "reinforcement-learning", "mobile-navigation"], "updated_at": "2026-02-06", "url": "https://arxiv.org/abs/2510.14388"}
{"aliases": [], "authors": ["Mathieu Andreux", "Märt Bakler", "Yanael Barbier", "Hamza Benchekroun", "Emilien Biré", "Antoine Bonnet", "Riaz Bordie", "Nathan Bout", "Matthias Brunel", "Aleix Cambray", "Pierre-Louis Cedoz", "Antoine Chassang", "Gautier Cloix", "Ethan Connelly", "Alexandra Constantinou", "Ramzi De Coster", "Hubert de la Jonquiere", "Aurélien Delfosse", "Maxime Delpit", "Alexis Deprez", "Augustin Derupti", "Mathieu Diaz", "Shannon D'Souza", "Julie Dujardin", "Abai Edmund", "Michael Eickenberg", "Armand Fatalot", "Wissem Felissi", "Isaac Herring", "Xavier Koegler", "Erwan Le Jumeau de Kergaradec", "Aurélien Lac", "Maxime Langevin", "Corentin Lauverjat", "Antonio Loison", "Avshalom Manevich", "Axel Moyal", "Axel Nguyen Kerbel", "Marinela Parovic", "Julien Revelle", "Guillaume Richard", "Mats Richter", "Ronan Riochet", "María Santos", "Romain Savidan", "Laurent Sifre", "Maxime Theillard", "Marc Thibault", "Ivan Valentini", "Tony Wu", "Laura Yie", "Kai Yuan", "Jevgenij Zubovskij"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "de25494af1bc2ceaa760a21da51803aaf89d9d2d1961887cfd0023fb87f28f63", "submitted_at": "2025-10-22T18:21:52Z", "updated_at": "2025-10-24T11:52:29Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2510.19949v2"}}, "environments": ["web", "desktop", "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:2510.19949", "identifiers": {"arxiv": "2510.19949"}, "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/2510.19949"}], "name": "Surfer 2: The Next Generation of Cross-Platform 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": "2025-10-22", "relations": [], "review_status": "curated", "scope": "core", "summary": "Surfer2 combines hierarchical context management with separate planning and execution, self-verification, and recovery for visual computer use. It evaluates the framework across web, desktop, and Android tasks without task-specific fine-tuning.", "topics": ["planning", "memory", "long-horizon", "robustness"], "updated_at": "2025-10-24", "url": "https://arxiv.org/abs/2510.19949"}
{"aliases": [], "authors": ["Hongrui Jia", "Jitong Liao", "Xi Zhang", "Haiyang Xu", "Tianbao Xie", "Chaoya Jiang", "Ming Yan", "Si Liu", "Wei Ye", "Fei Huang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d115356f723746ce95c3e40fed933786a0747b3b560c8f73a6b7af05ca22d24f", "submitted_at": "2025-10-28T15:56:36Z", "updated_at": "2025-11-11T04:10:32Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2510.24563v2"}}, "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:2510.24563", "identifiers": {"arxiv": "2510.24563"}, "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/2510.24563"}], "name": "OSWorld-MCP: Benchmarking MCP Tool Invocation 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": "2025-10-28", "relations": [], "review_status": "curated", "scope": "core", "summary": "OSWorld-MCP evaluates how agents choose between GUI operations and MCP tool calls in computer tasks. It combines generated and curated tools, with 158 manually validated tools across seven applications, to examine tool use alongside task completion.", "topics": ["evaluation", "tool-integration", "planning"], "updated_at": "2025-11-11", "url": "https://arxiv.org/abs/2510.24563"}
{"aliases": [], "authors": ["Tao Liu", "Chongyu Wang", "Rongjie Li", "Yingchen Yu", "Xuming He", "Bai Song"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "283b1a1de2a4e1d4ec8175f0c15af66c10097a7f5ad9fd2793e93b888526ff7e", "submitted_at": "2025-10-31T06:10:57Z", "updated_at": "2025-10-31T06:10:57Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2510.27210v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-18T10:54:12Z", "role": "primary", "sha256": "5f5bae19420f79231516d4bcf604ea18d4f930fc7790cde5a3e7fcb8d41a1126", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2508.19461%2C2508.19493%2C2509.21782%2C2509.25140%2C2510.03485%2C2510.07809%2C2510.14276%2C2510.27140%2C2510.27210%2C2511.13248%2C2512.08417%2C2601.07263%2C2602.13516%2C2603.00476%2C2603.11862%2C2603.18762%2C2603.23791%2C2604.00986%2C2604.16515%2C2604.16548%2C2604.24348%2C2605.05509%2C2605.08876%2C2605.20833%2C2605.24785%2C2607.00333%2C2607.18659%2C2607.25255%2C2608.00202%2C2608.08939%2C2608.24913%2C2608.29596&max_results=32"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:54:12Z", "role": "primary", "sha256": "5f5bae19420f79231516d4bcf604ea18d4f930fc7790cde5a3e7fcb8d41a1126", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2508.19461%2C2508.19493%2C2509.21782%2C2509.25140%2C2510.03485%2C2510.07809%2C2510.14276%2C2510.27140%2C2510.27210%2C2511.13248%2C2512.08417%2C2601.07263%2C2602.13516%2C2603.00476%2C2603.11862%2C2603.18762%2C2603.23791%2C2604.00986%2C2604.16515%2C2604.16548%2C2604.24348%2C2605.05509%2C2605.08876%2C2605.20833%2C2605.24785%2C2607.00333%2C2607.18659%2C2607.25255%2C2608.00202%2C2608.08939%2C2608.24913%2C2608.29596&max_results=32"}], "id": "arxiv:2510.27210", "identifiers": {"arxiv": "2510.27210"}, "kind": "paper", "last_verified_at": "2026-10-18T10:54:12Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2510.27210"}], "name": "GUI-Rise: Structured Reasoning and History Summarization for GUI 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": "2025-10-31", "relations": [], "review_status": "curated", "scope": "core", "summary": "Combines structured progress-and-decision reasoning with history summarization to improve cross-domain generalization in GUI navigation agents.", "topics": ["planning", "interaction-history", "reinforcement-learning"], "updated_at": "2025-10-31", "url": "https://arxiv.org/abs/2510.27210"}
{"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": ["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": ["Zehao Deng", "Tianjie Ju", "Zheng Wu", "Zhuosheng Zhang", "Gongshen Liu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "9acb395dfaad18b84ea4b52d97be194a42eb42d6ac922c55f2179e38cc258a31", "submitted_at": "2025-11-27T09:01:38Z", "updated_at": "2026-03-04T12:15:06Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2511.22235v2"}}, "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:2511.22235", "identifiers": {"arxiv": "2511.22235"}, "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/2511.22235"}], "name": "Training High-Level Schedulers with Execution-Feedback Reinforcement Learning for Long-Horizon GUI Automation", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-11-27", "relations": [], "review_status": "curated", "scope": "core", "summary": "This work trains high-level schedulers with staged reinforcement learning driven by execution feedback. Its CES framework separates a Coordinator for planning and decomposition, a StateTracker for compressed context, and a low-level Executor.", "topics": ["planning", "state-tracking", "reinforcement-learning", "long-horizon"], "updated_at": "2026-03-04", "url": "https://arxiv.org/abs/2511.22235"}
{"aliases": [], "authors": ["Zishu Wei", "Qixiang Ma", "Xavier Hu", "Yuhang Liu", "Hui Zang", "Yudong Zhao", "Tao Wang", "Shengyu Zhang", "Fei Wu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "49e76d19d0477bcc56f19e8170fed7e9cc80f9c4bafb5f03af3a2d172cb113bc", "submitted_at": "2025-12-10T07:40:23Z", "updated_at": "2025-12-10T07:40:23Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2512.09396v1"}}, "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:2512.09396", "identifiers": {"arxiv": "2512.09396"}, "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/2512.09396"}], "name": "GAIR: GUI Automation via Information-Joint Reasoning and Group Reflection", "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": "2025-12-10", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces GAIR, which combines information from specialized GUI models through a general-purpose coordinating model. When the available information is insufficient, group reflection sends targeted requests back to the specialists before selecting an operation.", "topics": ["planning", "self-reflection", "automation"], "updated_at": "2025-12-10", "url": "https://arxiv.org/abs/2512.09396"}
{"aliases": [], "authors": ["Shufan Li", "Konstantinos Kallidromitis", "Akash Gokul", "Yusuke Kato", "Kazuki Kozuka", "Aditya Grover"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e6228822cf64c6aeb1836bee52a6515123bd0d44d77cceb23e785bb4ff9aef47", "submitted_at": "2025-12-16T02:16:42Z", "updated_at": "2026-10-07T23:06:49Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2512.14014v2"}}, "environments": ["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", "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:2512.14014", "identifiers": {"arxiv": "2512.14014"}, "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/2512.14014"}], "name": "MobileWorldBench: Towards Semantic World Modeling For 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": "2025-12-16", "relations": [], "review_status": "curated", "scope": "core", "summary": "Evaluates semantic world models for mobile interfaces and studies natural-language state transitions as a tool for agent planning.", "topics": ["world-models", "planning", "evaluation"], "updated_at": "2026-10-07", "url": "https://arxiv.org/abs/2512.14014"}
{"aliases": [], "authors": ["Baixuan Li", "Jialong Wu", "Wenbiao Yin", "Kuan Li", "Zhongwang Zhang", "Huifeng Yin", "Zhengwei Tao", "Liwen Zhang", "Pengjun Xie", "Jingren Zhou", "Yong Jiang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "fde71dd935acb953bc5d01bcd48f0d6c3351db6a66126761baf134c114e7893c", "submitted_at": "2025-12-29T17:59:14Z", "updated_at": "2025-12-29T17:59:14Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2512.23647v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T12:31:09Z", "role": "primary", "sha256": "f8cee8b281f54ecb3112bb26b59d166617d33f08f098f44818968a9c9b777ce6", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2512.06866%2C2512.07287%2C2512.08769%2C2512.12594%2C2512.12730%2C2512.13159%2C2512.14012%2C2512.16925%2C2512.16962%2C2512.16975%2C2512.21373%2C2512.23647%2C2512.23738%2C2601.01743%2C2601.02046%2C2601.02163%2C2601.02454%2C2601.02702%2C2601.03468%2C2601.03928%2C2601.04795%2C2601.05107%2C2601.05432%2C2601.06411%2C2601.06487%2C2601.06490%2C2601.06565%2C2601.06663%2C2601.07143%2C2601.07375%2C2601.08012%2C2601.08079%2C2601.08225%2C2601.08237%2C2601.08731%2C2601.09113%2C2601.09749%2C2601.09883%2C2601.10378%2C2601.10440%2C2601.11653%2C2601.11895%2C2601.11903%2C2601.11974%2C2601.12042%2C2601.12307%2C2601.12560%2C2601.12641%2C2601.13186%2C2601.13383&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:31:09Z", "role": "primary", "sha256": "f8cee8b281f54ecb3112bb26b59d166617d33f08f098f44818968a9c9b777ce6", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2512.06866%2C2512.07287%2C2512.08769%2C2512.12594%2C2512.12730%2C2512.13159%2C2512.14012%2C2512.16925%2C2512.16962%2C2512.16975%2C2512.21373%2C2512.23647%2C2512.23738%2C2601.01743%2C2601.02046%2C2601.02163%2C2601.02454%2C2601.02702%2C2601.03468%2C2601.03928%2C2601.04795%2C2601.05107%2C2601.05432%2C2601.06411%2C2601.06487%2C2601.06490%2C2601.06565%2C2601.06663%2C2601.07143%2C2601.07375%2C2601.08012%2C2601.08079%2C2601.08225%2C2601.08237%2C2601.08731%2C2601.09113%2C2601.09749%2C2601.09883%2C2601.10378%2C2601.10440%2C2601.11653%2C2601.11895%2C2601.11903%2C2601.11974%2C2601.12042%2C2601.12307%2C2601.12560%2C2601.12641%2C2601.13186%2C2601.13383&max_results=50"}], "id": "arxiv:2512.23647", "identifiers": {"arxiv": "2512.23647"}, "kind": "paper", "last_verified_at": "2026-10-17T12:31:09Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2512.23647"}], "name": "Nested Browser-Use Learning for Agentic Information Seeking", "notes": "Reviewed against the complete retained official abstract. Experimental findings and release availability have not been independently reproduced or verified.", "published_at": "2025-12-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "NestBrowse organizes browser actions in a nested framework that separates interaction control from page exploration. It evaluates this design for deep information seeking beyond snippet retrieval and direct URL fetching.", "topics": ["web-navigation", "planning", "efficiency", "tool-integration"], "updated_at": "2025-12-29", "url": "https://arxiv.org/abs/2512.23647"}
{"aliases": [], "authors": ["Yilin Cao", "Yufeng Zhong", "Zhixiong Zeng", "Siran Dai", "Liming Zheng", "Jing Huang", "Haibo Qiu", "Peng Shi", "Wenji Mao"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "c40e3db3f538f089c19d7c8233ea01174d0591c3211a7acc412ee4df28e902a8", "submitted_at": "2026-01-07T15:51:44Z", "updated_at": "2026-08-31T13:27:52Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2601.04035v2"}}, "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:2601.04035", "identifiers": {"arxiv": "2601.04035"}, "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/2601.04035"}], "name": "MobileDreamer: Generative Sketch World Model for GUI Agent", "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": "MobileDreamer predicts post-action GUI states as task-relevant textual sketches with an order-invariant learning objective. Agents use these predictions for lookahead action selection, evaluated on AndroidWorld.", "topics": ["world-models", "planning", "mobile-navigation"], "updated_at": "2026-08-31", "url": "https://arxiv.org/abs/2601.04035"}
{"aliases": [], "authors": ["Bowen Yang", "Kaiming Jin", "Zhenyu Wu", "Zhaoyang Liu", "Qiushi Sun", "Zehao Li", "JingJing Xie", "Zhoumianze Liu", "Fangzhi Xu", "Kanzhi Cheng", "Qingyun Li", "Yian Wang", "Yu Qiao", "Zun Wang", "Zichen Ding"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "cfdaabaf3c6df80eb718e51dc666c54fb2ecf38ec520f6a71ccf5e5ce54248c8", "submitted_at": "2026-01-12T17:55:51Z", "updated_at": "2026-01-12T17:55:51Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2601.07779v1"}}, "environments": ["desktop", "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:2601.07779", "identifiers": {"arxiv": "2601.07779"}, "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/2601.07779"}], "name": "OS-Symphony: A Holistic Framework for Robust and Generalist Computer-Using Agent", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-01-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "OS-Symphony coordinates milestone-based reflection and memory with tools that build visually aligned tutorials through browser interaction. The framework evaluates trajectory correction and adaptation on online computer-use benchmarks including OSWorld.", "topics": ["memory", "self-reflection", "planning", "tool-integration"], "updated_at": "2026-01-12", "url": "https://arxiv.org/abs/2601.07779"}
{"aliases": [], "authors": ["Wenyuan Zhang", "Xinghua Zhang", "Haiyang Yu", "Shuaiyi Nie", "Bingli Wu", "Juwei Yue", "Tingwen Liu", "Yongbin Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "73df8c5b417c3387530b7730be3aed8b69374c06fdd3573f56147b19359c6912", "submitted_at": "2026-01-13T14:48:34Z", "updated_at": "2026-04-15T14:59:17Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2601.08605v2"}}, "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:2601.08605", "identifiers": {"arxiv": "2601.08605"}, "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/2601.08605"}], "name": "ExpSeek: Self-Triggered Experience Seeking for 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": "2026-01-13", "relations": [], "review_status": "curated", "scope": "core", "summary": "ExpSeek uses step-level entropy signals to decide when a web agent should request experience-based guidance. It supplies guidance tailored to the current step and evaluates the intervention across web-agent benchmarks.", "topics": ["memory", "planning", "web-navigation", "evaluation"], "updated_at": "2026-04-15", "url": "https://arxiv.org/abs/2601.08605"}
{"aliases": [], "authors": ["Hanna Foerster", "Tom Blanchard", "Kristina Nikolić", "Ilia Shumailov", "Cheng Zhang", "Robert Mullins", "Nicolas Papernot", "Florian Tramèr", "Yiren Zhao"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e32917bc92b65701eea06c09a2c681c7504418f8ce888af2b72012a1da50c9f6", "submitted_at": "2026-01-14T23:06:35Z", "updated_at": "2026-06-04T15:38:02Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2601.09923v3"}}, "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:2601.09923", "identifiers": {"arxiv": "2601.09923"}, "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/2601.09923"}], "name": "CaMeLs Can Use Computers Too: System-level Security for Computer Use Agents", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-01-14", "relations": [], "review_status": "curated", "scope": "core", "summary": "NOVA separates trusted task planning from screen-derived observations by producing a branching plan before execution and resolving runtime values through perception. Its OSWorld study examines security and utility while identifying remaining risks from perception-driven branch steering.", "topics": ["safety-security", "planning", "robustness"], "updated_at": "2026-06-04", "url": "https://arxiv.org/abs/2601.09923"}
{"aliases": [], "authors": ["Mingxian Yu", "Siqi Luo", "Xu Chen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "4ce0909d1c9081f099ae0dc0b0182e0442fd71fd58c62b13d486d154afc2187e", "submitted_at": "2026-01-24T11:35:37Z", "updated_at": "2026-01-24T11:35:37Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2601.17418v1"}}, "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:2601.17418", "identifiers": {"arxiv": "2601.17418"}, "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/2601.17418"}], "name": "GraphPilot: GUI Task Automation with One-Step LLM Reasoning Powered by Knowledge Graph", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-01-24", "relations": [], "review_status": "curated", "scope": "core", "summary": "GraphPilot builds app-specific knowledge graphs from explored pages, elements and transitions, then uses them to plan action sequences. It requests additional interface observations when uncertain and checks plans against graph transition rules before correction and execution.", "topics": ["mobile-navigation", "planning", "memory", "efficiency"], "updated_at": "2026-01-24", "url": "https://arxiv.org/abs/2601.17418"}
{"aliases": [], "authors": ["Sijia Li", "Xiaoyu Tan", "Shahir Ali", "Niels Schmidt", "Gengchen Ma", "Xihe Qiu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "223b9a915e5ea9849644318e1e5426e64ad8655ab46d0b007c30e1967130dc50", "submitted_at": "2026-01-27T07:46:05Z", "updated_at": "2026-01-27T07:46:05Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2601.19306v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-18T10:01:04.629294+00:00", "role": "primary", "sha256": "53521b0a80976320b10efcf7840629e4a2085119710324dd48db2cfd9c5d9cf6", "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+OR+ti%3A%22mobile+agent%22+OR+ti%3A%22UI+agent%22+OR+ti%3A%22desktop+agent%22+OR+ti%3A%22screen+grounding%22%29+AND+submittedDate%3A%5B200001010000+TO+202610180000%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:01:04.629294+00:00", "role": "primary", "sha256": "53521b0a80976320b10efcf7840629e4a2085119710324dd48db2cfd9c5d9cf6", "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+OR+ti%3A%22mobile+agent%22+OR+ti%3A%22UI+agent%22+OR+ti%3A%22desktop+agent%22+OR+ti%3A%22screen+grounding%22%29+AND+submittedDate%3A%5B200001010000+TO+202610180000%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:14:20.444793+00:00", "role": "primary", "sha256": "ef3bbc8adcc147239ea910e0c59904d7a26ee1610a7a4d1613be2cb845098536", "supports": ["summary", "scope", "environments", "notes"], "url": "https://arxiv.org/pdf/2601.19306v1"}], "id": "arxiv:2601.19306", "identifiers": {"arxiv": "2601.19306"}, "kind": "paper", "last_verified_at": "2026-10-18T10:01:04.629294+00:00", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2601.19306"}], "name": "Curiosity Driven Knowledge Retrieval for Mobile Agents", "notes": "The curiosity signal is a model-based proxy, not a calibrated probability of failure. Accurate external app information is a dependency, and AndroidWorld is the reported evaluation environment. No performance numbers were independently reproduced. Basis: complete retained abstract and source-reviewed sections of the exact-version paper (3.1 Problem Formulation; 3.2 Curiosity-Driven Information Gain Estimation; 3.3 Retrieving and Embedding External Knowledge); not an entire-paper reading or reproduction claim.", "published_at": "2026-01-27", "relations": [], "review_status": "curated", "scope": "core", "summary": "Uses execution uncertainty to trigger retrieval from application documentation, code, and past trajectories, organizing the results into structured AppCards for Android agent planning.", "topics": ["memory", "planning", "mobile-navigation"], "updated_at": "2026-01-27", "url": "https://arxiv.org/abs/2601.19306"}
{"aliases": [], "authors": ["Tianyi Chen", "Yinheng Li", "Michael Solodko", "Sen Wang", "Nan Jiang", "Tingyuan Cui", "Junheng Hao", "Jongwoo Ko", "Sara Abdali", "Leon Xu", "Suzhen Zheng", "Hao Fan", "Pashmina Cameron", "Justin Wagle", "Kazuhito Koishida"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "5a3efa3cd0add437e327ea49896a3ee794daa178664dfb4096aba66fed27554b", "submitted_at": "2026-01-28T23:38:25Z", "updated_at": "2026-02-02T23:11:55Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2601.21123v2"}}, "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:2601.21123", "identifiers": {"arxiv": "2601.21123"}, "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/2601.21123"}], "name": "CUA-Skill: Develop Skills for Computer Using Agent", "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": "CUA-Skill encodes Windows application knowledge in parameterized skills and execution/composition graphs. Its agent retrieves and instantiates these skills and uses memory for failure recovery, with end-to-end evaluation on WindowsAgentArena.", "topics": ["tool-integration", "planning", "memory", "automation"], "updated_at": "2026-02-02", "url": "https://arxiv.org/abs/2601.21123"}
{"aliases": [], "authors": ["Ziyu Lu", "Tengjin Weng", "Yiying Yang", "Yuhang Zhao", "Xinxin Huang", "Wenhao Jiang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "fa36a9fbd0b189717b2a5311f06b6127d54f2eca09961cedd2a0b9d2a78b1b8b", "submitted_at": "2026-01-29T07:22:50Z", "updated_at": "2026-01-29T07:22:50Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2601.21352v1"}}, "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:2601.21352", "identifiers": {"arxiv": "2601.21352"}, "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/2601.21352"}], "name": "BEAP-Agent: Backtrackable Execution and Adaptive Planning 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": "2026-01-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "BEAP-Agent treats GUI execution as depth-first search with coordinated planning, execution and task-state tracking. It supports backtracking across multiple levels of earlier states and evaluates recovery on OSWorld.", "topics": ["planning", "state-tracking", "robustness", "long-horizon"], "updated_at": "2026-01-29", "url": "https://arxiv.org/abs/2601.21352"}
{"aliases": [], "authors": ["Yurun Chen", "Zeyi Liao", "Ping Yin", "Taotao Xie", "Keting Yin", "Shengyu Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d1a6f98c0ab0a221c169796d9680e4b2a7bf7a0775c0493254f47d5e0a9b3c4e", "submitted_at": "2026-02-02T07:04:06Z", "updated_at": "2026-02-02T07:04:06Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2602.01725v1"}}, "environments": ["desktop", "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", "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": "2026-10-17T11:22:38+00:00", "role": "primary", "sha256": "6e1266c5a47cd667d681d3143a0331b74b6f64416b7e01f44a93a001c38ef74b", "supports": ["summary", "environments", "scope"], "url": "https://arxiv.org/html/2602.01725v1"}], "id": "arxiv:2602.01725", "identifiers": {"arxiv": "2602.01725"}, "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/2602.01725"}], "name": "SafePred: A Predictive Guardrail for Computer-Using Agents via World Models", "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": "2026-02-02", "relations": [], "review_status": "curated", "scope": "core", "summary": "SafePred uses world-model risk forecasts to guide action filtering and task replanning. Its desktop and web safety evaluations distinguish policy compliance, task success and success under the specified policy.", "topics": ["safety-security", "world-models", "planning", "evaluation"], "updated_at": "2026-02-02", "url": "https://arxiv.org/abs/2602.01725"}
{"aliases": [], "authors": ["Aiden Yiliu Li", "Xinyue Hao", "Shilong Liu", "Mengdi Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a5b730fa81936d6a1b2b0a8b74a522a22de39b977929165c44a5050e490c301f", "submitted_at": "2026-02-02T18:50:07Z", "updated_at": "2026-02-02T18:50:07Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2602.02468v1"}}, "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:2602.02468", "identifiers": {"arxiv": "2602.02468"}, "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/2602.02468"}], "name": "Avenir-Web: Human-Experience-Imitating Multimodal Web Agents with Mixture of Grounding Experts", "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": "Avenir-Web combines multiple grounding specialists with procedural experience, a task checklist and adaptive memory. It evaluates this planning and execution framework on live Online-Mind2Web tasks.", "topics": ["grounding", "planning", "memory", "web-navigation"], "updated_at": "2026-02-02", "url": "https://arxiv.org/abs/2602.02468"}
{"aliases": [], "authors": ["Sizhe Tang", "Rongqian Chen", "Tian Lan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "0a01499ce704bb7916a3157ff2d3fc755a1115d9b27c7e529c153bce360a5404", "submitted_at": "2026-02-03T02:07:12Z", "updated_at": "2026-02-03T02:07:12Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2602.02995v1"}}, "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:2602.02995", "identifiers": {"arxiv": "2602.02995"}, "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/2602.02995"}], "name": "Agent Alpha: Tree Search Unifying Generation, Exploration and Evaluation for Computer-Use Agents", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-02-03", "relations": [], "review_status": "curated", "scope": "core", "summary": "Agent Alpha uses step-level Monte Carlo tree search to reuse successful prefixes and recover from early GUI mistakes. Comparison-based evaluation and diversity-constrained expansion guide the search, with experiments on OSWorld.", "topics": ["planning", "efficiency", "evaluation"], "updated_at": "2026-02-03", "url": "https://arxiv.org/abs/2602.02995"}
{"aliases": [], "authors": ["Syed Nazmus Sakib", "Nafiul Haque", "Tapodhir Karmakar Taton", "Shahrear Bin Amin", "Shifat E. Arman"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "76fbeae086fb9ef21f820b02777a2ceb66ee37f8f1c978c7426589ea4e888e49", "submitted_at": "2026-02-05T06:24:23Z", "updated_at": "2026-09-08T14:40:01Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2602.05354v3"}}, "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:2602.05354", "identifiers": {"arxiv": "2602.05354"}, "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/2602.05354"}], "name": "Do Web Agents Investigate Before They Decide?", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-02-05", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces MIRAGE, a web decision benchmark where agents must actively uncover evidence that can contradict the visible context. Separate investigation, reasoning, decision, and hallucination measures diagnose failures beyond reaching the right page.", "topics": ["evaluation", "web-navigation", "planning"], "updated_at": "2026-09-08", "url": "https://arxiv.org/abs/2602.05354"}
{"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": ["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": ["Xiwen Teoh", "Yun Lin", "Duc-Minh Nguyen", "Ruofei Ren", "Wenjie Zhang", "Jin Song Dong"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2076fc8b0b7b37bdf43fe2e0b0cc5fa4c3a8cb7b44a41fdf80f59a8c05c7e1b3", "submitted_at": "2026-02-12T08:51:07Z", "updated_at": "2026-04-23T15:00:58Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2602.11724v3"}}, "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:2602.11724", "identifiers": {"arxiv": "2602.11724"}, "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/2602.11724"}], "name": "WebTestPilot: Agentic End-to-End Web Testing against Natural Language Specification by Inferring Oracles with Symbolized 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": "2026-02-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "WebTestPilot turns natural-language requirements into web test steps with inferred preconditions and postconditions over symbolized GUI elements. A benchmark of bug-injected web applications tests whether these context-aware oracles distinguish application defects from agent mistakes.", "topics": ["automation", "evaluation", "planning"], "updated_at": "2026-04-23", "url": "https://arxiv.org/abs/2602.11724"}
{"aliases": [], "authors": ["Nicholas Lee", "Lutfi Eren Erdogan", "Chris Joseph John", "Surya Krishnapillai", "Michael W. Mahoney", "Kurt Keutzer", "Amir Gholami"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "c64cd65fc7fec601658baf7836bdbe9c0e494bde2b0a4b66ec9d3090ee106b1f", "submitted_at": "2026-02-12T18:58:30Z", "updated_at": "2026-08-14T21:25:10Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2602.12276v2"}}, "environments": ["web"], "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:2602.12276", "identifiers": {"arxiv": "2602.12276"}, "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/2602.12276"}], "name": "Agentic Test-Time Scaling for WebAgents", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-02-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "CATTS allocates additional inference computation when an agent’s action votes show uncertainty. The paper compares voting and arbiter strategies and evaluates this selective scaling on WebArena-Lite, Online-Mind2Web and GoBrowse.", "topics": ["efficiency", "planning", "evaluation", "web-navigation"], "updated_at": "2026-08-14", "url": "https://arxiv.org/abs/2602.12276"}
{"aliases": [], "authors": ["Renjun Xu", "Yang Yan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "f0edc70265a8d46a7c22a9c6384de8aa19f79dbb6987f8c23f4a260647a6a3f0", "submitted_at": "2026-02-12T21:33:25Z", "updated_at": "2026-06-02T16:14:54Z", "version": 4, "versioned_url": "https://arxiv.org/abs/2602.12430v4"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T12:31:12Z", "role": "primary", "sha256": "c8c95278aa5dcc6a8f2eda3f3b44d61ec1f35adcb6a4c6c9ab328691a1a4f102", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2601.13462%2C2601.14232%2C2601.15331%2C2601.17548%2C2601.18418%2C2601.18491%2C2601.18827%2C2601.20171%2C2601.20613%2C2601.20742%2C2601.21233%2C2601.21253%2C2601.22037%2C2601.22154%2C2601.22511%2C2601.22607%2C2602.01550%2C2602.03117%2C2602.03361%2C2602.05665%2C2602.05746%2C2602.05877%2C2602.06052%2C2602.07398%2C2602.07458%2C2602.07918%2C2602.09007%2C2602.09463%2C2602.09514%2C2602.10122%2C2602.10525%2C2602.11224%2C2602.12268%2C2602.12430%2C2602.13318%2C2602.13543%2C2602.13692%2C2602.14038%2C2602.14337%2C2602.14364%2C2602.16313%2C2602.16317%2C2602.16666%2C2602.16901%2C2602.16930%2C2602.17547%2C2602.17646%2C2602.18846%2C2602.19320%2C2602.19594&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:31:12Z", "role": "primary", "sha256": "c8c95278aa5dcc6a8f2eda3f3b44d61ec1f35adcb6a4c6c9ab328691a1a4f102", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2601.13462%2C2601.14232%2C2601.15331%2C2601.17548%2C2601.18418%2C2601.18491%2C2601.18827%2C2601.20171%2C2601.20613%2C2601.20742%2C2601.21233%2C2601.21253%2C2601.22037%2C2601.22154%2C2601.22511%2C2601.22607%2C2602.01550%2C2602.03117%2C2602.03361%2C2602.05665%2C2602.05746%2C2602.05877%2C2602.06052%2C2602.07398%2C2602.07458%2C2602.07918%2C2602.09007%2C2602.09463%2C2602.09514%2C2602.10122%2C2602.10525%2C2602.11224%2C2602.12268%2C2602.12430%2C2602.13318%2C2602.13543%2C2602.13692%2C2602.14038%2C2602.14337%2C2602.14364%2C2602.16313%2C2602.16317%2C2602.16666%2C2602.16901%2C2602.16930%2C2602.17547%2C2602.17646%2C2602.18846%2C2602.19320%2C2602.19594&max_results=50"}], "id": "arxiv:2602.12430", "identifiers": {"arxiv": "2602.12430"}, "kind": "survey", "last_verified_at": "2026-10-17T12:31:12Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2602.12430"}], "name": "Agent Skills for Large Language Models: Architecture, Acquisition, Security, and the Path Forward", "notes": "Reviewed against the complete retained official abstract. Experimental findings and release availability have not been independently reproduced or verified. Adjacent survey of the skill layer: CUA deployment and GUI grounding are explicit parts of the review, but its overall scope extends well beyond interface agents.", "published_at": "2026-02-12", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "This survey organizes agent skills by architecture, acquisition, deployment and security, including their role in computer-use systems and GUI grounding. It also proposes a permission and lifecycle governance framework for skill deployment.", "topics": ["survey", "tool-integration", "safety-security", "planning"], "updated_at": "2026-06-02", "url": "https://arxiv.org/abs/2602.12430"}
{"aliases": [], "authors": ["Yuyu Guo", "Wenjie Yang", "Siyuan Yang", "Ziyang Liu", "Cheng Chen", "Yuan Wei", "Yun Hu", "Yang Huang", "Guoliang Hao", "Dongsheng Yuan", "Jianming Wang", "Xin Chen", "Hang Yu", "Lei Lei", "Peng Di"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "04298ed8a4bc0f2811f52eeee6da0a13faa265bb3c5cbc5ea6be573a77c21f27", "submitted_at": "2026-02-14T02:33:55Z", "updated_at": "2026-04-30T02:05:10Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2602.13559v2"}}, "environments": ["web"], "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", "notes"], "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:2602.13559", "identifiers": {"arxiv": "2602.13559"}, "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/2602.13559"}], "name": "OpAgent: Operator Agent for Web Navigation", "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": "2026-02-14", "relations": [], "review_status": "curated", "scope": "core", "summary": "OpAgent combines multi-task GUI fine-tuning with online web reinforcement learning using outcome judgments and rule-based progress rewards. A modular planner, grounder, reflector and summarizer support navigation and recovery, evaluated separately from the underlying trained policy.", "topics": ["web-navigation", "reinforcement-learning", "planning", "self-reflection"], "updated_at": "2026-04-30", "url": "https://arxiv.org/abs/2602.13559"}
{"aliases": [], "authors": ["Zexue He", "Yu Wang", "Churan Zhi", "Yuanzhe Hu", "Tzu-Ping Chen", "Lang Yin", "Ze Chen", "Tong Arthur Wu", "Siru Ouyang", "Zihan Wang", "Jiaxin Pei", "Julian McAuley", "Yejin Choi", "Alex Pentland"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2317a94d54b4458a1b7c5e92754aeaefcbf270aabfa0bd8ea333c673cd4596b3", "submitted_at": "2026-02-18T09:49:14Z", "updated_at": "2026-09-17T22:55:03Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2602.16313v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T12:31:12Z", "role": "primary", "sha256": "c8c95278aa5dcc6a8f2eda3f3b44d61ec1f35adcb6a4c6c9ab328691a1a4f102", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2601.13462%2C2601.14232%2C2601.15331%2C2601.17548%2C2601.18418%2C2601.18491%2C2601.18827%2C2601.20171%2C2601.20613%2C2601.20742%2C2601.21233%2C2601.21253%2C2601.22037%2C2601.22154%2C2601.22511%2C2601.22607%2C2602.01550%2C2602.03117%2C2602.03361%2C2602.05665%2C2602.05746%2C2602.05877%2C2602.06052%2C2602.07398%2C2602.07458%2C2602.07918%2C2602.09007%2C2602.09463%2C2602.09514%2C2602.10122%2C2602.10525%2C2602.11224%2C2602.12268%2C2602.12430%2C2602.13318%2C2602.13543%2C2602.13692%2C2602.14038%2C2602.14337%2C2602.14364%2C2602.16313%2C2602.16317%2C2602.16666%2C2602.16901%2C2602.16930%2C2602.17547%2C2602.17646%2C2602.18846%2C2602.19320%2C2602.19594&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:31:12Z", "role": "primary", "sha256": "c8c95278aa5dcc6a8f2eda3f3b44d61ec1f35adcb6a4c6c9ab328691a1a4f102", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2601.13462%2C2601.14232%2C2601.15331%2C2601.17548%2C2601.18418%2C2601.18491%2C2601.18827%2C2601.20171%2C2601.20613%2C2601.20742%2C2601.21233%2C2601.21253%2C2601.22037%2C2601.22154%2C2601.22511%2C2601.22607%2C2602.01550%2C2602.03117%2C2602.03361%2C2602.05665%2C2602.05746%2C2602.05877%2C2602.06052%2C2602.07398%2C2602.07458%2C2602.07918%2C2602.09007%2C2602.09463%2C2602.09514%2C2602.10122%2C2602.10525%2C2602.11224%2C2602.12268%2C2602.12430%2C2602.13318%2C2602.13543%2C2602.13692%2C2602.14038%2C2602.14337%2C2602.14364%2C2602.16313%2C2602.16317%2C2602.16666%2C2602.16901%2C2602.16930%2C2602.17547%2C2602.17646%2C2602.18846%2C2602.19320%2C2602.19594&max_results=50"}], "id": "arxiv:2602.16313", "identifiers": {"arxiv": "2602.16313"}, "kind": "paper", "last_verified_at": "2026-10-17T12:31:12Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2602.16313"}], "name": "MemoryArena: Benchmarking Agent Memory in Interdependent Multi-Session Agentic Tasks", "notes": "Reviewed against the complete retained official abstract. Experimental findings and release availability have not been independently reproduced or verified.", "published_at": "2026-02-18", "relations": [], "review_status": "curated", "scope": "core", "summary": "MemoryArena links tasks across sessions so that agents must distill earlier interaction experience and use it in later decisions. Its evaluation includes web navigation alongside planning, search and formal reasoning, testing memory through subsequent actions rather than recall alone.", "topics": ["memory", "interaction-history", "planning", "evaluation", "long-horizon"], "updated_at": "2026-09-17", "url": "https://arxiv.org/abs/2602.16313"}
{"aliases": [], "authors": ["Yiming Guan", "Rui Yu", "John Zhang", "Lu Wang", "Chaoyun Zhang", "Liqun Li", "Bo Qiao", "Si Qin", "He Huang", "Fangkai Yang", "Pu Zhao", "Lukas Wutschitz", "Samuel Kessler", "Huseyin A Inan", "Robert Sim", "Saravan Rajmohan", "Qingwei Lin", "Dongmei Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "3d2237dd43f4c876ba69b68c1871ca3efd3aa501e93c86655eb2de6d9f35d2e8", "submitted_at": "2026-02-19T13:48:29Z", "updated_at": "2026-02-19T13:48:29Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2602.17365v1"}}, "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", "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:2602.17365", "identifiers": {"arxiv": "2602.17365"}, "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/2602.17365"}], "name": "Computer-Using World Model", "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": "2026-02-19", "relations": [], "review_status": "curated", "scope": "core", "summary": "Presents a desktop world model that predicts textual UI changes before synthesizing the next screenshot. It learns from Microsoft Office transitions and is evaluated by letting a frozen agent compare simulated actions before execution.", "topics": ["world-models", "planning", "action-prediction"], "updated_at": "2026-02-19", "url": "https://arxiv.org/abs/2602.17365"}
{"aliases": [], "authors": ["Hongbin Zhong", "Fazle Faisal", "Luis França", "Tanakorn Leesatapornwongsa", "Adriana Szekeres", "Kexin Rong", "Suman Nath"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e179c833d43a88098d576e31e726eb3e4289f728659697364d0b448282df3fc1", "submitted_at": "2026-02-24T03:03:18Z", "updated_at": "2026-09-28T17:40:20Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2602.20502v2"}}, "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:2602.20502", "identifiers": {"arxiv": "2602.20502"}, "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/2602.20502"}], "name": "ActionEngine: From Reactive to Programmatic Web Agents via State Machine Memory", "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": "2026-02-24", "relations": [], "review_status": "curated", "scope": "core", "summary": "ActionEngine explores web applications to build an updateable state-machine memory of GUI states, operations and transitions. It generates complete task programs from that memory, using reactive repair to handle changed interfaces or missing knowledge.", "topics": ["memory", "planning", "efficiency", "web-navigation"], "updated_at": "2026-09-28", "url": "https://arxiv.org/abs/2602.20502"}
{"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": ["Syed Yusuf Ahmed", "Shiwei Feng", "Chanwoo Bae", "Calix Barrus Xiangyu Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "8ad18d7fe815f5529fef663130b7ce8841763589e7f4319c0c7556ec2df1da03", "submitted_at": "2026-03-10T22:56:03Z", "updated_at": "2026-03-10T22:56:03Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2603.10268v1"}}, "environments": ["web"], "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:2603.10268", "identifiers": {"arxiv": "2603.10268"}, "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/2603.10268"}], "name": "SpecOps: A Fully Automated AI Agent Testing Framework in Real-World GUI 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": "2026-03-10", "relations": [], "review_status": "curated", "scope": "core", "summary": "SpecOps divides testing of GUI-based agents into test generation, environment setup, execution and validation handled by specialist agents. Its evaluation spans five real-world agents, including web applications and browser extensions, and measures execution reliability and bug discovery.", "topics": ["automation", "evaluation", "planning"], "updated_at": "2026-03-10", "url": "https://arxiv.org/abs/2603.10268"}
{"aliases": [], "authors": ["Taiyi Wang", "Sian Gooding", "Florian Hartmann", "Oriana Riva", "Edward Grefenstette"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "5538933a0e3d11df6ff4d767849979703d83412599fbbb0f8b432c4d9850633b", "submitted_at": "2026-03-20T06:32:16Z", "updated_at": "2026-03-20T06:32:16Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2603.19685v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T12:31:16Z", "role": "primary", "sha256": "9289bc047b59e93f75703b72f5e92d9277e115238427519c343df550220e351b", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2602.19870%2C2602.22769%2C2602.23166%2C2602.23320%2C2603.00198%2C2603.00842%2C2603.00991%2C2603.01050%2C2603.01145%2C2603.02236%2C2603.02795%2C2603.03278%2C2603.04549%2C2603.06713%2C2603.07106%2C2603.07670%2C2603.07853%2C2603.08706%2C2603.09716%2C2603.10002%2C2603.10664%2C2603.11139%2C2603.11320%2C2603.12483%2C2603.13026%2C2603.13247%2C2603.15714%2C2603.16065%2C2603.16289%2C2603.16839%2C2603.17216%2C2603.17378%2C2603.19313%2C2603.19685%2C2603.21489%2C2603.22527%2C2603.23638%2C2603.24157%2C2603.24639%2C2603.27646%2C2603.28069%2C2603.28407%2C2603.29418%2C2603.29847%2C2604.00528%2C2604.00886%2C2604.01194%2C2604.01438%2C2604.01560%2C2604.02985&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:31:16Z", "role": "primary", "sha256": "9289bc047b59e93f75703b72f5e92d9277e115238427519c343df550220e351b", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2602.19870%2C2602.22769%2C2602.23166%2C2602.23320%2C2603.00198%2C2603.00842%2C2603.00991%2C2603.01050%2C2603.01145%2C2603.02236%2C2603.02795%2C2603.03278%2C2603.04549%2C2603.06713%2C2603.07106%2C2603.07670%2C2603.07853%2C2603.08706%2C2603.09716%2C2603.10002%2C2603.10664%2C2603.11139%2C2603.11320%2C2603.12483%2C2603.13026%2C2603.13247%2C2603.15714%2C2603.16065%2C2603.16289%2C2603.16839%2C2603.17216%2C2603.17378%2C2603.19313%2C2603.19685%2C2603.21489%2C2603.22527%2C2603.23638%2C2603.24157%2C2603.24639%2C2603.27646%2C2603.28069%2C2603.28407%2C2603.29418%2C2603.29847%2C2604.00528%2C2604.00886%2C2604.01194%2C2604.01438%2C2604.01560%2C2604.02985&max_results=50"}], "id": "arxiv:2603.19685", "identifiers": {"arxiv": "2603.19685"}, "kind": "paper", "last_verified_at": "2026-10-17T12:31:16Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2603.19685"}], "name": "A Subgoal-driven Framework for Improving Long-Horizon LLM Agents", "notes": "Reviewed against the complete retained official abstract. Experimental findings and release availability have not been independently reproduced or verified.", "published_at": "2026-03-20", "relations": [], "review_status": "curated", "scope": "core", "summary": "The framework separates inference-time subgoal planning from MiRA, a reinforcement-learning method that supplies milestone rewards. WebArena-Lite experiments evaluate these mechanisms for long-horizon web navigation.", "topics": ["planning", "reinforcement-learning", "reward-models", "long-horizon", "evaluation"], "updated_at": "2026-03-20", "url": "https://arxiv.org/abs/2603.19685"}
{"aliases": [], "authors": ["Zijian Lu", "Yiping Zuo", "Yupeng Nie", "Xin He", "Weibei Fan", "Lianyong Qi", "Shi Jin"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "8e8182e64c7a57a2dd6eb381762bc9a60b08e13ba5a337d0e9125ab6a5518b6c", "submitted_at": "2026-03-20T09:25:50Z", "updated_at": "2026-04-15T07:34:34Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2603.20340v3"}}, "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:2603.20340", "identifiers": {"arxiv": "2603.20340"}, "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/2603.20340"}], "name": "ContractSkill: Repairable Contract-Based Skills for Multimodal 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. The abstract limits its portability evidence to transfer within a benchmark family, not generalization across arbitrary benchmarks.", "published_at": "2026-03-20", "relations": [], "review_status": "curated", "scope": "core", "summary": "Turns draft web-agent skills into structured executable artifacts that support deterministic checks, fault localization and local repair. VisualWebArena and MiniWoB experiments examine execution and reuse within the same benchmark family.", "topics": ["web-navigation", "planning", "robustness"], "updated_at": "2026-04-15", "url": "https://arxiv.org/abs/2603.20340"}
{"aliases": [], "authors": ["Shoubin Yu", "Lei Shu", "Antoine Yang", "Yao Fu", "Srinivas Sunkara", "Maria Wang", "Jindong Chen", "Mohit Bansal", "Boqing Gong"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "3e4ed7391969359c2594e8b96b8a3a559ef98d2c84a10c1a2c39b3bd962b58c5", "submitted_at": "2026-03-23T19:47:58Z", "updated_at": "2026-03-23T19:47:58Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2603.22529v1"}}, "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:2603.22529", "identifiers": {"arxiv": "2603.22529"}, "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/2603.22529"}], "name": "Ego2Web: A Web Agent Benchmark Grounded in Egocentric 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": "2026-03-23", "relations": [], "review_status": "curated", "scope": "core", "summary": "Ego2Web pairs first-person videos with online tasks that require agents to connect the user’s physical context to web actions. It also evaluates an automatic task judge against human judgments and analyzes failures in video understanding and web execution.", "topics": ["evaluation", "video", "web-navigation", "planning"], "updated_at": "2026-03-23", "url": "https://arxiv.org/abs/2603.22529"}
{"aliases": [], "authors": ["Akash Ghosh", "Tajamul Ashraf", "Rishu Kumar Singh", "Numan Saeed", "Sriparna Saha", "Xiuying Chen", "Salman Khan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "8fb6a294616e1be46c5da05f60cda7d43c05d8a1ce71116c5e30cb94507cd56f", "submitted_at": "2026-03-25T10:25:48Z", "updated_at": "2026-03-25T10:25:48Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2603.24157v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T12:31:16Z", "role": "primary", "sha256": "9289bc047b59e93f75703b72f5e92d9277e115238427519c343df550220e351b", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2602.19870%2C2602.22769%2C2602.23166%2C2602.23320%2C2603.00198%2C2603.00842%2C2603.00991%2C2603.01050%2C2603.01145%2C2603.02236%2C2603.02795%2C2603.03278%2C2603.04549%2C2603.06713%2C2603.07106%2C2603.07670%2C2603.07853%2C2603.08706%2C2603.09716%2C2603.10002%2C2603.10664%2C2603.11139%2C2603.11320%2C2603.12483%2C2603.13026%2C2603.13247%2C2603.15714%2C2603.16065%2C2603.16289%2C2603.16839%2C2603.17216%2C2603.17378%2C2603.19313%2C2603.19685%2C2603.21489%2C2603.22527%2C2603.23638%2C2603.24157%2C2603.24639%2C2603.27646%2C2603.28069%2C2603.28407%2C2603.29418%2C2603.29847%2C2604.00528%2C2604.00886%2C2604.01194%2C2604.01438%2C2604.01560%2C2604.02985&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:31:16Z", "role": "primary", "sha256": "9289bc047b59e93f75703b72f5e92d9277e115238427519c343df550220e351b", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2602.19870%2C2602.22769%2C2602.23166%2C2602.23320%2C2603.00198%2C2603.00842%2C2603.00991%2C2603.01050%2C2603.01145%2C2603.02236%2C2603.02795%2C2603.03278%2C2603.04549%2C2603.06713%2C2603.07106%2C2603.07670%2C2603.07853%2C2603.08706%2C2603.09716%2C2603.10002%2C2603.10664%2C2603.11139%2C2603.11320%2C2603.12483%2C2603.13026%2C2603.13247%2C2603.15714%2C2603.16065%2C2603.16289%2C2603.16839%2C2603.17216%2C2603.17378%2C2603.19313%2C2603.19685%2C2603.21489%2C2603.22527%2C2603.23638%2C2603.24157%2C2603.24639%2C2603.27646%2C2603.28069%2C2603.28407%2C2603.29418%2C2603.29847%2C2604.00528%2C2604.00886%2C2604.01194%2C2604.01438%2C2604.01560%2C2604.02985&max_results=50"}], "id": "arxiv:2603.24157", "identifiers": {"arxiv": "2603.24157"}, "kind": "paper", "last_verified_at": "2026-10-17T12:31:16Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2603.24157"}], "name": "CarePilot: A Multi-Agent Framework for Long-Horizon Computer Task Automation in Healthcare", "notes": "Reviewed against the complete retained official abstract. Experimental findings and release availability have not been independently reproduced or verified.", "published_at": "2026-03-25", "relations": [], "review_status": "curated", "scope": "core", "summary": "CareFlow benchmarks long healthcare-software workflows, while CarePilot coordinates action proposals, execution checks and corrective feedback using short- and long-term memory. Its actor consumes visual interfaces and system state to select semantic actions.", "topics": ["automation", "planning", "memory", "evaluation", "long-horizon", "workflow"], "updated_at": "2026-03-25", "url": "https://arxiv.org/abs/2603.24157"}
{"aliases": [], "authors": ["Rui Xie", "Zhi Gao", "Chenrui Shi", "Zirui Shang", "Lu Chen", "Qing Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "b388086aeb8ac47e386099e5cd68cb89bb03978ee7d5897950181035d0abb723", "submitted_at": "2026-03-27T10:33:08Z", "updated_at": "2026-06-30T04:47:42Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2603.26266v3"}}, "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:2603.26266", "identifiers": {"arxiv": "2603.26266"}, "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/2603.26266"}], "name": "GUIDE: Resolving Domain Bias in GUI Agents through Real-Time Web Video Retrieval and Plug-and-Play Annotation", "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": "2026-03-27", "relations": [], "review_status": "curated", "scope": "core", "summary": "GUIDE retrieves relevant software tutorials through subtitle-based video search, then infers planning and grounding knowledge from consecutive annotated keyframes. It inserts that knowledge into existing GUI-agent modules without parameter updates and evaluates the plug-in on OSWorld.", "topics": ["video", "planning", "grounding", "training-free"], "updated_at": "2026-06-30", "url": "https://arxiv.org/abs/2603.26266"}
{"aliases": [], "authors": ["Rongqian Chen", "Yu Li", "Zeyu Fang", "Sizhe Tang", "Weidong Cao", "Tian Lan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "238aac264fb21022f7c546ae3aa3f4ea104c381f5d318f6d962f22fb00095216", "submitted_at": "2026-04-06T20:39:30Z", "updated_at": "2026-05-28T08:13:55Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2604.05157v3"}}, "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:2604.05157", "identifiers": {"arxiv": "2604.05157"}, "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/2604.05157"}], "name": "IntentScore: Intent-Conditioned Action Evaluation 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": "2026-04-06", "relations": [], "review_status": "curated", "scope": "core", "summary": "Trains IntentScore on offline desktop GUI interactions to assess candidate actions using both their state relevance and planning intent. The reward model is evaluated as an action reranker for Agent S3 on OSWorld.", "topics": ["reward-models", "planning", "evaluation"], "updated_at": "2026-05-28", "url": "https://arxiv.org/abs/2604.05157"}
{"aliases": [], "authors": ["Zhaoyang Wang", "Qianhui Wu", "Xuchao Zhang", "Chaoyun Zhang", "Wenlin Yao", "Fazle Elahi Faisal", "Baolin Peng", "Si Qin", "Suman Nath", "Qingwei Lin", "Chetan Bansal", "Dongmei Zhang", "Saravan Rajmohan", "Jianfeng Gao", "Huaxiu Yao"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "bf709a3d7096bc07218206edb1accfdda64036a39ea770e637b3e073dd9a76c8", "submitted_at": "2026-04-14T21:48:15Z", "updated_at": "2026-08-31T07:02:14Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2604.13318v2"}}, "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:2604.13318", "identifiers": {"arxiv": "2604.13318"}, "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/2604.13318"}], "name": "WebXSkill: Skill Learning for Autonomous 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": "2026-04-14", "relations": [], "review_status": "curated", "scope": "core", "summary": "WebXSkill pairs parameterized action programs with step-level language guidance, mining reusable skills from synthetic browser-agent traces. A URL-based graph supports retrieval, and deployment alternates between direct program execution and guided native planning.", "topics": ["memory", "planning", "web-navigation", "automation"], "updated_at": "2026-08-31", "url": "https://arxiv.org/abs/2604.13318"}
{"aliases": [], "authors": ["Zhengxi Lu", "Fei Tang", "Guangyi Liu", "Kaitao Song", "Xu Tan", "Jin Ma", "Wenqi Zhang", "Weiming Lu", "Jun Xiao", "Yueting Zhuang", "Yongliang Shen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d6e81b448c9c22e2fd25dce519db3fdc0585b0265b94d448751d7a37ed910dfa", "submitted_at": "2026-04-15T12:55:46Z", "updated_at": "2026-04-15T12:55:46Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2604.13822v1"}}, "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:2604.13822", "identifiers": {"arxiv": "2604.13822"}, "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/2604.13822"}], "name": "UI-Copilot: Advancing Long-Horizon GUI Automation via Tool-Integrated Policy Optimization", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-04-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "UI-Copilot separates GUI execution from on-demand memory retrieval and numerical assistance. Its training procedure treats tool selection and multi-turn task execution as distinct optimization targets while keeping persistent observations outside transient context.", "topics": ["tool-integration", "memory", "planning", "long-horizon", "mobile-navigation"], "updated_at": "2026-04-15", "url": "https://arxiv.org/abs/2604.13822"}
{"aliases": [], "authors": ["Lingfeng Zhang", "Yongan Sun", "Jinpeng Hu", "Hui Ma", "Yang Ying", "Kuien Liu", "Zenglin Shi", "Meng Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "123577fbb4431f22bc18d2d32187744268f9d63b3149e9e1775fb44673771feb", "submitted_at": "2026-04-20T05:19:49Z", "updated_at": "2026-04-21T03:15:01Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2604.17821v2"}}, "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:2604.17821", "identifiers": {"arxiv": "2604.17821"}, "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/2604.17821"}], "name": "WebUncertainty: Dual-Level Uncertainty Driven Planning and Reasoning For Autonomous Web Agent", "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": "2026-04-20", "relations": [], "review_status": "curated", "scope": "core", "summary": "WebUncertainty adapts planning mode to task uncertainty and guides action-level tree search with confidence-derived uncertainty estimates. Its evaluations examine autonomous execution of dynamic, long-horizon web tasks on WebArena and WebVoyager.", "topics": ["planning", "web-navigation", "long-horizon", "robustness"], "updated_at": "2026-04-21", "url": "https://arxiv.org/abs/2604.17821"}
{"aliases": [], "authors": ["Zerui Qin", "Sheng Yue", "Xingyuan Hua", "Yongjian Fu", "Ju Ren"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "388eac2721929fc8211f40cc4edde61be0b6e6145fa60fbaa444ec80822fcc36", "submitted_at": "2026-05-12T15:48:44Z", "updated_at": "2026-05-12T15:48:44Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2605.12294v1"}}, "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:2605.12294", "identifiers": {"arxiv": "2605.12294"}, "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/2605.12294"}], "name": "Executable Agentic Memory for GUI Agent", "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": "Executable Agentic Memory stores GUI routines in a knowledge graph built through state-aware exploration and action-group mining. A learned value model guides tree search over this executable memory for retrieval-based planning, with AndroidWorld and efficiency evaluation.", "topics": ["memory", "planning", "long-horizon", "efficiency"], "updated_at": "2026-05-12", "url": "https://arxiv.org/abs/2605.12294"}
{"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": ["Julien Piet", "Annabella Chow", "Yiwei Hou", "Muxi Lyu", "Sylvie Venuto", "Jinhao Zhu", "Raluca Ada Popa", "David Wagner"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "572be49190554bbfb6accac4c87ef91608f88e90d06250f229db2922e19fcd56", "submitted_at": "2026-05-14T02:48:57Z", "updated_at": "2026-05-14T02:48:57Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2605.14290v1"}}, "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:2605.14290", "identifiers": {"arxiv": "2605.14290"}, "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/2605.14290"}], "name": "Web Agents Should Adopt the Plan-Then-Execute Paradigm", "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. Directly analyzes browser-agent prompt-injection boundaries and WebArena task structure, but advocates typed website operations rather than demonstrating a new GUI-control policy.", "published_at": "2026-05-14", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "Argues for compiling web tasks into predefined execution graphs before exposing an agent to untrusted runtime content. Its WebArena task analysis identifies typed, semantically complete website tools as a prerequisite and distinguishes this interface proposal from reactive click-level control.", "topics": ["planning", "safety-security", "tool-integration"], "updated_at": "2026-05-14", "url": "https://arxiv.org/abs/2605.14290"}
{"aliases": [], "authors": ["Yichao Liu", "Huawen Shen", "Liu Yu", "Shiyu Liu", "Zeyu Chen", "Yu Zhou"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "fd85c0d0a8b6bc5c687bc301b1b84d6a019c5713012890e0a23c08ae2d83a1e6", "submitted_at": "2026-05-15T02:27:41Z", "updated_at": "2026-05-15T02:27:41Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2605.15542v1"}}, "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:2605.15542", "identifiers": {"arxiv": "2605.15542"}, "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/2605.15542"}], "name": "DRS-GUI: Dynamic Region Search for Training-Free GUI Grounding", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-05-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "DRS-GUI searches screenshot regions through focus, shift and scatter operations selected by a tree-search planner. A region-quality reward guides this training-free perceptual search before final GUI grounding.", "topics": ["grounding", "training-free", "planning", "screen-understanding"], "updated_at": "2026-05-15", "url": "https://arxiv.org/abs/2605.15542"}
{"aliases": [], "authors": ["Jingjing Liu", "Ziye Huang", "Zihao Cheng", "Zeming Liu", "Jiahong Wu", "Yuhang Guo", "Kehai Chen", "Yunhong Wang", "Haifeng Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "017909a697c6c661d3486d960042f4d691f70a660a60083e70512de34690cb51", "submitted_at": "2026-05-18T08:36:04Z", "updated_at": "2026-05-18T08:36:04Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2605.18048v1"}}, "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:2605.18048", "identifiers": {"arxiv": "2605.18048"}, "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/2605.18048"}], "name": "DocOS: Towards Proactive Document-Guided Actions in 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": "DocOS benchmarks agents that search online documentation and translate procedural instructions into GUI actions in interactive web environments. It separates failures to retrieve useful procedures from failures to ground retrieved instructions into executable actions.", "topics": ["web-navigation", "grounding", "evaluation", "planning"], "updated_at": "2026-05-18", "url": "https://arxiv.org/abs/2605.18048"}
{"aliases": [], "authors": ["Caleb Winston", "Ron Yifeng Wang", "Azalia Mirhoseini", "Christos Kozyrakis"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "9b73bf7776e280a9040b1d7d1e29957913723253014c1dfe84f64916b2a08e5e", "submitted_at": "2026-05-20T17:54:27Z", "updated_at": "2026-05-29T17:54:05Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2605.21470v2"}}, "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:2605.21470", "identifiers": {"arxiv": "2605.21470"}, "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/2605.21470"}], "name": "Agent JIT Compilation for Latency-Optimizing Web Agent Planning and Scheduling", "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": "2026-05-20", "relations": [], "review_status": "curated", "scope": "core", "summary": "Agent JIT compilation turns browser task descriptions into validated executable plans containing model calls, tools and parallel work. A cost-based scheduler and pre/postcondition protocol target latency and incorrect tool use in web-agent execution.", "topics": ["planning", "efficiency", "tool-integration", "web-navigation"], "updated_at": "2026-05-29", "url": "https://arxiv.org/abs/2605.21470"}
{"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": ["Yuxiang Chai", "Han Xiao", "Xinyu Fu", "Jinpeng Chen", "Rui Liu", "Hongsheng Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "461790f351a2816004027cc58d07f79ee7c670ed5064c59b1651a4efb84ff2bf", "submitted_at": "2026-05-28T07:49:09Z", "updated_at": "2026-05-28T07:49:09Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2605.29534v1"}}, "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:2605.29534", "identifiers": {"arxiv": "2605.29534"}, "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/2605.29534"}], "name": "UI-KOBE: Knowledge-Oriented Behavior Exploration for Lightweight Graph-Guided 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": "2026-05-28", "relations": [], "review_status": "curated", "scope": "core", "summary": "UI-KOBE explores mobile applications to build graphs of UI states and executable transitions. A lightweight agent then matches its current screenshot to the graph and chooses a transition, local action, completion signal or fallback action.", "topics": ["mobile-navigation", "planning", "state-tracking", "memory", "efficiency"], "updated_at": "2026-05-28", "url": "https://arxiv.org/abs/2605.29534"}
{"aliases": [], "authors": ["Jing Yu Koh", "Ruslan Salakhutdinov", "Daniel Fried"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "acc6e8af676b7de5393469d379746e674ccd662f570399622d0d5b1b0e882939", "submitted_at": "2026-06-01T01:29:36Z", "updated_at": "2026-06-01T01:29:36Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.01533v1"}}, "environments": ["desktop", "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:2606.01533", "identifiers": {"arxiv": "2606.01533"}, "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/2606.01533"}], "name": "Multi-Agent Computer Use", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-06-01", "relations": [], "review_status": "curated", "scope": "core", "summary": "MACU represents computer-use objectives as a dependency graph whose manager dispatches parallel subagents and revises tasks as new information arrives. Desktop and web experiments examine coordination, information handoff and task-completion time.", "topics": ["planning", "long-horizon", "workflow", "state-tracking", "efficiency"], "updated_at": "2026-06-01", "url": "https://arxiv.org/abs/2606.01533"}
{"aliases": [], "authors": ["Jiaxi Li", "Ke Deng", "Yun Wang", "Jingyuan Huang", "Yucheng Shi", "Qiaoyu Tan", "Jin Lu", "Ninghao Liu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6d891956bde9b8672859fcfcd2d54dfd3e4720964c25d77cc28bc1200ef3a56c", "submitted_at": "2026-06-03T03:11:50Z", "updated_at": "2026-06-03T03:11:50Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.04391v1"}}, "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:2606.04391", "identifiers": {"arxiv": "2606.04391"}, "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/2606.04391"}], "name": "Online Skill Learning for Web Agents via State-Grounded Dynamic Retrieval", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-06-03", "relations": [], "review_status": "curated", "scope": "core", "summary": "State-Grounded Dynamic Retrieval extracts reusable sub-procedures from completed web trajectories and represents them as paired text and code. Skills are retrieved against both the current webpage state and the task goal at intermediate execution steps.", "topics": ["web-navigation", "memory", "interaction-history", "planning", "workflow"], "updated_at": "2026-06-03", "url": "https://arxiv.org/abs/2606.04391"}
{"aliases": [], "authors": ["Radeen Mostafa", "Sawradip Saha"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d29dec745923e652654521f65a32f919a99d4577992006efc1a5e469bf775d81", "submitted_at": "2026-06-08T12:18:22Z", "updated_at": "2026-06-08T12:18:22Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.09399v1"}}, "environments": ["web"], "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:2606.09399", "identifiers": {"arxiv": "2606.09399"}, "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/2606.09399"}], "name": "RunAgent SuperBrowser: A Theory of Autonomous Web Navigation Grounded in Human Browsing Behaviour", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-06-08", "relations": [], "review_status": "curated", "scope": "core", "summary": "SUPERBROWSER combines screenshot-based target proposals with separate orchestration, planning and action roles. A bounded ledger and context-eviction process retain recent actions, task facts and checkpoints during browser execution.", "topics": ["web-navigation", "grounding", "planning", "memory", "efficiency"], "updated_at": "2026-06-08", "url": "https://arxiv.org/abs/2606.09399"}
{"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": ["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": ["Wenyi Wu", "Sibo Zhu", "Kun Zhou", "Aayush Salvi", "Zixuan Song", "Biwei Huang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "40daa1b180a0c2eaba7b907a0f61b8dd8819cb3033b9da118162330d97cfecf7", "submitted_at": "2026-07-13T10:48:25Z", "updated_at": "2026-07-13T10:48:25Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2607.11388v1"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:09Z", "role": "primary", "sha256": "65cc3856c70a0fb267f9996872e077de10e9f54821ab1ffb1d82f0a92fa11dbf", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2607.05277%2C2607.06118%2C2607.06306%2C2607.07820%2C2607.11185%2C2607.11388%2C2607.12625%2C2607.12640%2C2607.13027%2C2607.14277%2C2607.14443%2C2607.14548%2C2607.15193%2C2607.15550%2C2607.17050%2C2607.18046%2C2607.19865%2C2607.21557%2C2607.22798%2C2607.23263%2C2607.24112%2C2607.24167%2C2607.25904%2C2607.26041%2C2607.27155%2C2607.28074%2C2607.28227%2C2607.28609%2C2607.29320%2C2608.01964%2C2608.02352%2C2608.02643%2C2608.03270%2C2608.03327%2C2608.04530%2C2608.04741%2C2608.05587%2C2608.05784%2C2608.05891%2C2608.06065%2C2608.06171%2C2608.06477%2C2608.09278%2C2608.09654%2C2608.10366%2C2608.10692%2C2608.11588%2C2608.12429%2C2608.15930%2C2608.17310&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:09Z", "role": "primary", "sha256": "65cc3856c70a0fb267f9996872e077de10e9f54821ab1ffb1d82f0a92fa11dbf", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?id_list=2607.05277%2C2607.06118%2C2607.06306%2C2607.07820%2C2607.11185%2C2607.11388%2C2607.12625%2C2607.12640%2C2607.13027%2C2607.14277%2C2607.14443%2C2607.14548%2C2607.15193%2C2607.15550%2C2607.17050%2C2607.18046%2C2607.19865%2C2607.21557%2C2607.22798%2C2607.23263%2C2607.24112%2C2607.24167%2C2607.25904%2C2607.26041%2C2607.27155%2C2607.28074%2C2607.28227%2C2607.28609%2C2607.29320%2C2608.01964%2C2608.02352%2C2608.02643%2C2608.03270%2C2608.03327%2C2608.04530%2C2608.04741%2C2608.05587%2C2608.05784%2C2608.05891%2C2608.06065%2C2608.06171%2C2608.06477%2C2608.09278%2C2608.09654%2C2608.10366%2C2608.10692%2C2608.11588%2C2608.12429%2C2608.15930%2C2608.17310&max_results=50"}], "id": "arxiv:2607.11388", "identifiers": {"arxiv": "2607.11388"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:09Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2607.11388"}], "name": "StructAgent: Harness Long-horizon Digital Agents with Unified Causal Structure", "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": "2026-07-13", "relations": [], "review_status": "curated", "scope": "core", "summary": "StructAgent maintains compact task state and regulates long-horizon execution through verified transitions and progress checkpoints. Its workflow supports evidence-based completion and targeted recovery, with desktop computer-use evaluation and a separate Minecraft extension.", "topics": ["state-tracking", "planning", "long-horizon", "robustness"], "updated_at": "2026-07-13", "url": "https://arxiv.org/abs/2607.11388"}
{"aliases": [], "authors": ["Yunxin Li", "Jinchao Li", "Shibo Su", "Zhenran Xu", "Chenrui Zhao", "Tongshu Bian", "Xiaoman Liang", "Meishan Zhang", "Baotian Hu", "Min Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "281de2bb6e45a58efd047735accc2049deab0f8815931c86abbd38b86b6d23bf", "submitted_at": "2026-07-14T11:04:01Z", "updated_at": "2026-07-15T04:41:47Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2607.12625v2"}}, "environments": ["desktop", "android", "ios", "harmonyos", "windows"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:09Z", "role": "primary", "sha256": "65cc3856c70a0fb267f9996872e077de10e9f54821ab1ffb1d82f0a92fa11dbf", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2607.05277%2C2607.06118%2C2607.06306%2C2607.07820%2C2607.11185%2C2607.11388%2C2607.12625%2C2607.12640%2C2607.13027%2C2607.14277%2C2607.14443%2C2607.14548%2C2607.15193%2C2607.15550%2C2607.17050%2C2607.18046%2C2607.19865%2C2607.21557%2C2607.22798%2C2607.23263%2C2607.24112%2C2607.24167%2C2607.25904%2C2607.26041%2C2607.27155%2C2607.28074%2C2607.28227%2C2607.28609%2C2607.29320%2C2608.01964%2C2608.02352%2C2608.02643%2C2608.03270%2C2608.03327%2C2608.04530%2C2608.04741%2C2608.05587%2C2608.05784%2C2608.05891%2C2608.06065%2C2608.06171%2C2608.06477%2C2608.09278%2C2608.09654%2C2608.10366%2C2608.10692%2C2608.11588%2C2608.12429%2C2608.15930%2C2608.17310&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:09Z", "role": "primary", "sha256": "65cc3856c70a0fb267f9996872e077de10e9f54821ab1ffb1d82f0a92fa11dbf", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?id_list=2607.05277%2C2607.06118%2C2607.06306%2C2607.07820%2C2607.11185%2C2607.11388%2C2607.12625%2C2607.12640%2C2607.13027%2C2607.14277%2C2607.14443%2C2607.14548%2C2607.15193%2C2607.15550%2C2607.17050%2C2607.18046%2C2607.19865%2C2607.21557%2C2607.22798%2C2607.23263%2C2607.24112%2C2607.24167%2C2607.25904%2C2607.26041%2C2607.27155%2C2607.28074%2C2607.28227%2C2607.28609%2C2607.29320%2C2608.01964%2C2608.02352%2C2608.02643%2C2608.03270%2C2608.03327%2C2608.04530%2C2608.04741%2C2608.05587%2C2608.05784%2C2608.05891%2C2608.06065%2C2608.06171%2C2608.06477%2C2608.09278%2C2608.09654%2C2608.10366%2C2608.10692%2C2608.11588%2C2608.12429%2C2608.15930%2C2608.17310&max_results=50"}], "id": "arxiv:2607.12625", "identifiers": {"arxiv": "2607.12625"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:09Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2607.12625"}], "name": "KnowAct-GUIClaw: Know Deeply, Act Perfectly, Personal GUI Assistant with Self-Evolving Memory and Skill", "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": "2026-07-14", "relations": [], "review_status": "curated", "scope": "core", "summary": "KnowAct-GUIClaw combines a host planner with pluggable GUI subagents, attributable experience memory and an evolving skill library. It uses user profiles and feedback for task decomposition and evaluates transfer across Android, iOS, HarmonyOS and Windows.", "topics": ["memory", "planning", "interaction-history", "mobile-navigation"], "updated_at": "2026-07-15", "url": "https://arxiv.org/abs/2607.12625"}
{"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": ["Madhumitha Venkatesan", "Shicheng Wen", "Jiajing Guo", "Jorge Piazentin Ono", "Liu Ren", "Dongyu Liu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "1179c657434761187f3fa3bed99f46530a69878578b7a74c68febae20e64d6b3", "submitted_at": "2026-07-16T16:48:39Z", "updated_at": "2026-07-16T16:48:39Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2607.15193v1"}}, "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"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%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:2607.15193", "identifiers": {"arxiv": "2607.15193"}, "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/2607.15193"}], "name": "Plover: Steering GUI Agents through Plan-Centric Interaction", "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": "Plover exposes GUI-agent plans and replanning as persistent artifacts that users can inspect and edit during execution. Its planner–executor design supports localized correction through plan edits, language guidance and screenshot-grounded interventions, studied through failure repair and workflow scenarios.", "topics": ["planning", "human-in-the-loop", "observability", "workflow"], "updated_at": "2026-07-16", "url": "https://arxiv.org/abs/2607.15193"}
{"aliases": [], "authors": ["Zedong Yu", "Qianxing Li", "Zhi Gao", "Liuyu Xiang", "Chenrui Shi", "Yang Liu", "Huiming Wu", "Yujie Wei", "Yuhao Fei", "Yubo Fu", "Zhaofeng He"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "078ff9a56b769923d15a98581685a3e2f47aac7d898ea731999cabaec3918fa7", "submitted_at": "2026-07-17T12:48:03Z", "updated_at": "2026-07-17T12:48:03Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2607.22689v1"}}, "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:2607.22689", "identifiers": {"arxiv": "2607.22689"}, "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/2607.22689"}], "name": "Beyond Sequential Interaction: Benchmarking Parallel Execution and Coordination 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": "2026-07-17", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces ParaGUIBench, with tasks and efficiency measurements for coordinated GUI agents operating separate desktop instances over a shared file system. Its ParaGUI planner-worker system decomposes tasks and dispatches concurrent subtasks for comparison with serial execution.", "topics": ["evaluation", "planning", "efficiency", "long-horizon"], "updated_at": "2026-07-17", "url": "https://arxiv.org/abs/2607.22689"}
{"aliases": [], "authors": ["Guangyi Liu", "Huan Zhao", "Quanming Yao"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "57fa0a3eaad7525991656989ed3011101026362acec43895f32cf6e24be2f4ee", "submitted_at": "2026-07-27T08:48:01Z", "updated_at": "2026-07-27T08:48:01Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2607.24167v1"}}, "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:2607.24167", "identifiers": {"arxiv": "2607.24167"}, "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/2607.24167"}], "name": "Falsifiable Commitment Planning for Self-Correcting 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": "2026-07-27", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces FCPAgent, which attaches confirming and falsifying evidence to each skill-grounded plan step. A plan–test–repair loop checks actions and observations, then localizes contradictions so execution, skills, or plans can be revised at the appropriate level.", "topics": ["planning", "robustness", "self-reflection"], "updated_at": "2026-07-27", "url": "https://arxiv.org/abs/2607.24167"}
{"aliases": [], "authors": ["Zihan Dong", "Rui Qian", "Qishi Zhan", "Dongshen Peng", "Kaixin Li", "Yu Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "463cf35eb57b3d38dda6ce768e6f08d93b1c16d6abba0d790760d846cd56fa2c", "submitted_at": "2026-07-30T15:50:10Z", "updated_at": "2026-07-30T15:50:10Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2607.28399v1"}}, "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:2607.28399", "identifiers": {"arxiv": "2607.28399"}, "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/2607.28399"}], "name": "Why Are GUI Agents Correct but Late? Decode on the Decision-Time Critical Path, Tested with Pre-Compiled Policy Trees", "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 Adaptive Anticipatory Policy Trees, which precompute guarded actions and deadlines during idle periods so transient GUI events can be handled without new text generation. Paired experiments examine when this prepared execution helps and when reactive control remains preferable.", "topics": ["efficiency", "planning", "robustness"], "updated_at": "2026-07-30", "url": "https://arxiv.org/abs/2607.28399"}
{"aliases": [], "authors": ["Yashaswi Malla", "Sandra Siby"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "60c93c90e6d4026452810a866a0c49d57afdec15ad5ced30606c925a8e83c8f7", "submitted_at": "2026-07-31T18:27:12Z", "updated_at": "2026-07-31T18:27:12Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2608.00202v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-18T10:54:12Z", "role": "primary", "sha256": "5f5bae19420f79231516d4bcf604ea18d4f930fc7790cde5a3e7fcb8d41a1126", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2508.19461%2C2508.19493%2C2509.21782%2C2509.25140%2C2510.03485%2C2510.07809%2C2510.14276%2C2510.27140%2C2510.27210%2C2511.13248%2C2512.08417%2C2601.07263%2C2602.13516%2C2603.00476%2C2603.11862%2C2603.18762%2C2603.23791%2C2604.00986%2C2604.16515%2C2604.16548%2C2604.24348%2C2605.05509%2C2605.08876%2C2605.20833%2C2605.24785%2C2607.00333%2C2607.18659%2C2607.25255%2C2608.00202%2C2608.08939%2C2608.24913%2C2608.29596&max_results=32"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:54:12Z", "role": "primary", "sha256": "5f5bae19420f79231516d4bcf604ea18d4f930fc7790cde5a3e7fcb8d41a1126", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2508.19461%2C2508.19493%2C2509.21782%2C2509.25140%2C2510.03485%2C2510.07809%2C2510.14276%2C2510.27140%2C2510.27210%2C2511.13248%2C2512.08417%2C2601.07263%2C2602.13516%2C2603.00476%2C2603.11862%2C2603.18762%2C2603.23791%2C2604.00986%2C2604.16515%2C2604.16548%2C2604.24348%2C2605.05509%2C2605.08876%2C2605.20833%2C2605.24785%2C2607.00333%2C2607.18659%2C2607.25255%2C2608.00202%2C2608.08939%2C2608.24913%2C2608.29596&max_results=32"}], "id": "arxiv:2608.00202", "identifiers": {"arxiv": "2608.00202"}, "kind": "paper", "last_verified_at": "2026-10-18T10:54:12Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2608.00202"}], "name": "From Monoliths to Swarms: A Study of Attack Surface Evolution in the Transition to Multi-Agent Web Systems", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-07-31", "relations": [], "review_status": "curated", "scope": "core", "summary": "Categorizes structural attack surfaces that emerge when single-agent browser systems are decomposed into multi-agent web workflows.", "topics": ["safety-security", "web-navigation", "planning"], "updated_at": "2026-07-31", "url": "https://arxiv.org/abs/2608.00202"}
{"aliases": [], "authors": ["Xiaokai Yan", "Jingtao Ding", "Yong Li", "Zhiwen Yu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e2660e01921a78bed7280b34f35a1b587e62339c2e280d6294bbb6d6e666786e", "submitted_at": "2026-08-14T09:36:59Z", "updated_at": "2026-08-14T09:36:59Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2608.14132v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-18T10:00:40.283922+00:00", "role": "primary", "sha256": "ddd5bd966c7430bd9d4fd0269e7d3d46d8839fa5bfa57d601b97b4dc735cfc9a", "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+OR+ti%3A%22mobile+agent%22+OR+ti%3A%22UI+agent%22+OR+ti%3A%22desktop+agent%22+OR+ti%3A%22screen+grounding%22%29+AND+submittedDate%3A%5B200001010000+TO+202610180000%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:00:40.283922+00:00", "role": "primary", "sha256": "ddd5bd966c7430bd9d4fd0269e7d3d46d8839fa5bfa57d601b97b4dc735cfc9a", "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+OR+ti%3A%22mobile+agent%22+OR+ti%3A%22UI+agent%22+OR+ti%3A%22desktop+agent%22+OR+ti%3A%22screen+grounding%22%29+AND+submittedDate%3A%5B200001010000+TO+202610180000%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:05:22.211464+00:00", "role": "primary", "sha256": "8d6aa8484ff4a7cbaac589c2c220407ebf6ecd1e216015a248125abdb0f2ed2d", "supports": ["summary", "scope", "environments", "notes"], "url": "https://arxiv.org/html/2608.14132v1"}], "id": "arxiv:2608.14132", "identifiers": {"arxiv": "2608.14132"}, "kind": "paper", "last_verified_at": "2026-10-18T10:00:40.283922+00:00", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2608.14132"}], "name": "Act2Intention: A Benchmark For Developing Active Mobile Agents Through Inferring User Intention from GUI Actions", "notes": "The benchmark combines real and generated trajectory material. Opportune intervention timing and real-user acceptance are stated limitations. Section 4.2.2 specifies Android emulator instances for generated action trajectories. No results were independently reproduced. Basis: complete retained abstract and source-reviewed sections of the exact-version paper (3 Task definition; 4.1 Raw data collection; 4.2 Trajectory construction; 7 Limitations and future directions); not an entire-paper reading or reproduction claim.", "published_at": "2026-08-14", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces intention understanding, prediction, and confirmed execution tasks over mobile GUI trajectories, with a proactive agent combining personalized intention models and experience-guided actions.", "topics": ["human-in-the-loop", "planning", "mobile-navigation", "evaluation"], "updated_at": "2026-08-14", "url": "https://arxiv.org/abs/2608.14132"}
{"aliases": [], "authors": ["Weiming Li", "Helen Paik", "Yulei Sui"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "af6b60c6a39a990e950cf4b98f905f2d3aa5ae62d4055630c25a650508ec8bf8", "submitted_at": "2026-08-26T13:20:27Z", "updated_at": "2026-09-25T23:54:04Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2608.25777v2"}}, "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:2608.25777", "identifiers": {"arxiv": "2608.25777"}, "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/2608.25777"}], "name": "LocalLSTC: A Long Short-Term Control Architecture for Locally Deployed 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": "2026-08-26", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces LocalLSTC, a training-free control architecture that maintains persistent task state and uses it to form short execution commitments. Outcomes feed back into that state for progress checks, recovery, and termination in locally deployed desktop agents.", "topics": ["state-tracking", "planning", "training-free"], "updated_at": "2026-09-25", "url": "https://arxiv.org/abs/2608.25777"}
{"aliases": [], "authors": ["Junxuan Li", "Zijun Liu", "Ziyi Huang", "Peng Li", "Yuzhou Liu", "Ming Yan", "Yang Liu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "bbf657b2228bad25e668e3c9d7ec20c0ecab403d34b7e0b5608d350c46b8e50f", "submitted_at": "2026-08-29T14:45:09Z", "updated_at": "2026-08-29T14:45:09Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2608.29305v1"}}, "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:2608.29305", "identifiers": {"arxiv": "2608.29305"}, "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/2608.29305"}], "name": "Learning Simple Test-Time Environments for LLM 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": "2026-08-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces Test-Time Environment Decomposition, which uses trial interaction to break complex web observations into simpler modules. A label-free adaptation procedure learns from these sub-environments and studies whether their gains compose in full web tasks.", "topics": ["planning", "robustness", "web-navigation"], "updated_at": "2026-08-29", "url": "https://arxiv.org/abs/2608.29305"}
{"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": ["Zhaoyuan Huang", "Tianjie Ju", "Pengzhou Cheng", "Zheng Wu", "Yansi Li", "Chuanbiao Song", "Jun Lan", "Huijia Zhu", "Weiqiang Wang", "Zhuosheng Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a59f7d54845de6efebf6365a188e61e41fdf31e9ebe894b1d01ce7b86d6be4ec", "submitted_at": "2026-09-03T06:48:41Z", "updated_at": "2026-09-03T06:48:41Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.03438v1"}}, "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", "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:2609.03438", "identifiers": {"arxiv": "2609.03438"}, "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/2609.03438"}], "name": "Do GUI Agents Know When Not to Act? Enabling Conflict-Aware Termination for Multimodal 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": "2026-09-03", "relations": [], "review_status": "curated", "scope": "core", "summary": "CONFLICTGUI evaluates termination under instruction conflicts and mismatches between instructions and interface context. CONFLICTGUARD couples feasibility checks with conditional action modulation to reduce unnecessary execution while tracking normal-task behavior.", "topics": ["safety-security", "evaluation", "robustness", "planning"], "updated_at": "2026-09-03", "url": "https://arxiv.org/abs/2609.03438"}
{"aliases": [], "authors": ["Zixiang Chen", "Yuheng Lu", "Zihao Cheng", "Zeming Liu", "Jizeng Bai", "Ziye Huang", "Zhiyin Lin", "Zihan Li", "Yuhang Guo", "Yunhong Wang", "Haifeng Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "5ec17516bce87a8e349983acac14415a6ceeb12c167fce57aa37d6157cd3cd83", "submitted_at": "2026-09-09T16:59:59Z", "updated_at": "2026-09-09T16:59:59Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.10451v1"}}, "environments": ["desktop", "android", "windows", "linux"], "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:2609.10451", "identifiers": {"arxiv": "2609.10451"}, "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/2609.10451"}], "name": "JarvisGUI: Towards Cross-Device GUI Agents with Dynamic Task Composition", "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": "2026-09-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "JarvisGUI composes typed input-output task transformations into dynamic workflows spanning Android, Windows and Ubuntu devices. Its virtual-environment benchmark tests transfer of intermediate results, shared state and cross-device dependencies.", "topics": ["evaluation", "cross-app", "planning", "long-horizon", "state-tracking"], "updated_at": "2026-09-09", "url": "https://arxiv.org/abs/2609.10451"}
{"aliases": [], "authors": ["Yan Zhang", "Pei Fu", "Daiqing Wu", "Huawen Shen", "Ruoceng Zhang", "Shaojie Zhang", "Jiahui Yang", "Yu Zhou", "Can Ma", "Zhenbo Luo", "Jian Luan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d184b111aa2a9ad65dd57391b72bcef369c2bcd01ceb8e4f00f7a2320db856d8", "submitted_at": "2026-09-22T07:02:35Z", "updated_at": "2026-09-22T07:02:35Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.25769v1"}}, "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:2609.25769", "identifiers": {"arxiv": "2609.25769"}, "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/2609.25769"}], "name": "Towards Omni-dimensional GUI Agent Navigation with Masked Trajectory Prediction", "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": "2026-09-22", "relations": [], "review_status": "curated", "scope": "core", "summary": "MaP unifies GUI-navigation objectives by masking and predicting components of multi-turn trajectories. Role-aware adapters address heterogeneous training data, and evaluations compare the approach with direct task-mixture training across navigation benchmarks.", "topics": ["imitation-learning", "action-prediction", "state-tracking", "planning"], "updated_at": "2026-09-22", "url": "https://arxiv.org/abs/2609.25769"}
{"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": ["Linghua Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "b387c1cfe77f6e06477415944b3923189a1f0ca30be0680f01815cb65b0067bc", "submitted_at": "2026-09-24T17:30:32Z", "updated_at": "2026-09-24T17:30:32Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.30186v1"}}, "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:2609.30186", "identifiers": {"arxiv": "2609.30186"}, "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/2609.30186"}], "name": "Jev-Mobile: Jev as an Executor for 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": "2026-09-24", "relations": [], "review_status": "curated", "scope": "core", "summary": "Separates occasional visual-language planning from frequent lightweight action selection over an accessibility-tree-based action space.", "topics": ["mobile-navigation", "efficiency", "planning"], "updated_at": "2026-09-24", "url": "https://arxiv.org/abs/2609.30186"}
{"aliases": [], "authors": ["Chuhan Zhang", "Qi Xie", "Ziyue Wang", "Jianing Yin", "Yunfan Zhou", "Dazhen Deng", "Yingcai Wu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "f091cab38c932e08d21afa936dfe241c93c0b2139aba1d1ce7a6d1513ad5da1a", "submitted_at": "2026-09-26T08:58:50Z", "updated_at": "2026-09-26T08:58:50Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.32385v1"}}, "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:2609.32385", "identifiers": {"arxiv": "2609.32385"}, "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/2609.32385"}], "name": "DashAct: A Progressive Diagnostic Benchmark for GUI Agents in Interactive Dashboard Analysis", "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": "Diagnoses dashboard-agent failures by progressively restoring interaction context, next-action support, and visual target information within the same task.", "topics": ["evaluation", "grounding", "planning"], "updated_at": "2026-09-26", "url": "https://arxiv.org/abs/2609.32385"}
{"aliases": [], "authors": ["Fuyao Zhang", "Xuan Wang", "Zherui Li", "Jiaming Zhang", "Longtao Huang", "Wei Yang Bryan Lim"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "563661718e6e64d62f0601bbc5c214828219b9495a729ae685ba1ae9599332f3", "submitted_at": "2026-09-27T07:15:57Z", "updated_at": "2026-09-27T07:15:57Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.33304v1"}}, "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:2609.33304", "identifiers": {"arxiv": "2609.33304"}, "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/2609.33304"}], "name": "Relevance Does Not Imply Applicability: Experience Activation for Personal 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": "2026-09-27", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces ExpActivator, which matches the current GUI state to relevant historical states before supplying past actions as guidance. A separate activation rule decides whether a recurring user intent is warranted by the current time and situation or should be withheld.", "topics": ["memory", "planning", "training-free"], "updated_at": "2026-09-27", "url": "https://arxiv.org/abs/2609.33304"}
{"aliases": [], "authors": ["Bingo Zhang", "Haochuan Lu", "Zongjie Li", "Genjian Li", "Ari Yu Zhang", "Chaozheng Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "f81fd07262dccf6783729e4336fbd77835f2fcdea030eef8dc2272e7ef82b8fc", "submitted_at": "2026-09-28T09:49:25Z", "updated_at": "2026-09-28T09:49:25Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.34769v1"}}, "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:2609.34769", "identifiers": {"arxiv": "2609.34769"}, "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/2609.34769"}], "name": "LongPuzzleBench: Evaluating GUI Agents on Long-Horizon Visual Puzzles", "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. Agents execute real GUI moves in six puzzle games; the tasks diagnose computer-control planning but remain game-based rather than general productivity workflows.", "published_at": "2026-09-28", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "LongPuzzleBench evaluates coherent long-horizon planning on persistent visual puzzle boards through native GUI actions. It compares GUI-only and code-assisted agents and uses controlled hints and memory diagnostics to examine decisions that preserve or eliminate future solutions.", "topics": ["evaluation", "planning", "long-horizon", "state-tracking"], "updated_at": "2026-09-28", "url": "https://arxiv.org/abs/2609.34769"}
{"aliases": [], "authors": ["Ruozhao Yang", "Mingfei Cheng", "Xiaofei Xie"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a9bc3b57957e6c4f1af399379c4275366563b94a8b8fe01ec83e54715cfc8d11", "submitted_at": "2026-09-28T11:53:24Z", "updated_at": "2026-09-28T11:53:24Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.34974v1"}}, "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:2609.34974", "identifiers": {"arxiv": "2609.34974"}, "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/2609.34974"}], "name": "Before Acting, Change the State: Prospective State Intervention for Web Agents under Deceptive Interfaces", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-09-28", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces Veer, a runtime defense that changes task-relevant web state when an otherwise valid proposed action would produce an unauthorized consequence. It plans a prospective intervention and grounds and verifies its execution before the base agent continues.", "topics": ["safety-security", "planning", "robustness"], "updated_at": "2026-09-28", "url": "https://arxiv.org/abs/2609.34974"}
{"aliases": [], "authors": ["Dhananjay Ashok", "Shantanu Agarwal", "Vivek Datla", "Jonathan May", "Alfy Samuel"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "35ded32e7bac9a875a0fa41efad43fed5fb0aab54fade5cfc2a0628b1e666587", "submitted_at": "2026-10-01T22:30:08Z", "updated_at": "2026-10-01T22:30:08Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.02542v1"}, "license_review": {"document": {"evidence_urls": ["https://arxiv.org/html/2610.02542v1"], "reason": "The retained version-specific arXiv HTML infobox explicitly labels this paper License: CC BY 4.0. This document declaration does not establish code, training-data or model-weight licenses."}}}, "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": "2026-10-20T15:08:56Z", "role": "primary", "sha256": "d8c22c6b3f07d4156c38af9dd7ca7c06a5d0a11889a7350d821a020e21604590", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "licenses.document"], "url": "https://arxiv.org/html/2610.02542v1"}], "id": "arxiv:2610.02542", "identifiers": {"arxiv": "2610.02542"}, "kind": "paper", "last_verified_at": "2026-10-20T15:08:56Z", "licenses": {"code": null, "data": null, "document": {"basis": "The retained version-specific arXiv HTML infobox explicitly labels this paper License: CC BY 4.0. This document declaration does not establish code, training-data or model-weight licenses.", "declared": "CC-BY-4.0", "retrieved_at": "2026-10-20T15:08:56Z", "url": "https://arxiv.org/html/2610.02542v1"}, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2610.02542"}], "name": "How To Train Your World Model: Fine-tuning vs RAG for LM-based World Modeling", "notes": "Adjacent cross-domain study with WebShop as its web evaluation. WebShop uses text-rendered search/product pages and an explicit valid-action menu, not screenshot control or live purchases. A frozen policy proposes five actions; a world model predicts their next observations and the policy scores them before one is executed in the simulator. Utility is normalized environment reward, while fidelity uses a model judge; neither is a universal GUI task-success rate. Fine-tuning and retrieval use collected training experience; the hybrid masks same-episode retrieval, which does not establish general contamination freedom. The retained Atom and v1 HTML abstract wording differs; original bibliographic metadata and its abstract digest are preserved.", "published_at": "2026-10-01", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "Compares fine-tuned, retrieval-based and hybrid language-model world models for next-observation prediction and action selection in text environments including WebShop.", "topics": ["world-models", "planning", "evaluation", "web-navigation"], "updated_at": "2026-10-01", "url": "https://arxiv.org/abs/2610.02542"}
{"aliases": [], "authors": ["Giulio Zingrillo", "Hanna Foerster", "Ilia Shumailov", "Yiren Zhao", "Robert Mullins"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "7469484bb1186dd633b26e41dddc890008e893c303f93d2352f93c9d47fd78c2", "submitted_at": "2026-10-02T10:08:39Z", "updated_at": "2026-10-02T10:08:39Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.03089v1"}}, "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:2610.03089", "identifiers": {"arxiv": "2610.03089"}, "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/2610.03089"}], "name": "Securing Computer-Use Agents Against Branch Steering Attacks", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-10-02", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces STEER-Bench to study branch-steering attacks that manipulate untrusted web data toward hazardous paths already permitted by an agent’s plan. Its COBRA architecture combines trusted branching plans with advance constraints on the parameters and destinations each branch can use.", "topics": ["safety-security", "planning", "evaluation"], "updated_at": "2026-10-02", "url": "https://arxiv.org/abs/2610.03089"}
{"aliases": [], "authors": ["Ziyang Long", "Xinqi Li", "Lujing Xing", "Hsin-Jung Yang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "baacd8b4fad7b146bb9152a54d60a617fdbed5d0142c0c0768dacea2e7f18e60", "submitted_at": "2026-10-03T22:56:09Z", "updated_at": "2026-10-03T22:56:09Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.04800v1"}}, "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:2610.04800", "identifiers": {"arxiv": "2610.04800"}, "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/2610.04800"}], "name": "MedImageOSWorld: Benchmarking GUI Agents for Medical Image Consoles", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-10-03", "relations": [], "review_status": "curated", "scope": "core", "summary": "Evaluates GUI agents in simulated medical-imaging consoles, separately checking workflow completion and acquisition outcomes against hidden anatomical targets.", "topics": ["evaluation", "science", "planning"], "updated_at": "2026-10-03", "url": "https://arxiv.org/abs/2610.04800"}
{"aliases": [], "authors": ["Yudong Bai", "Yihong Chen", "Quanming Yao", "Yaqing Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2627960a5698d4c2f245e91ca5033be8c09e8a1ddea02490fcc1ed9f3d42217d", "submitted_at": "2026-10-08T11:12:57Z", "updated_at": "2026-10-08T11:12:57Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.11707v1"}}, "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:2610.11707", "identifiers": {"arxiv": "2610.11707"}, "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/2610.11707"}], "name": "DeltaReplay: Task-Relative Memory Reuse for 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": "2026-10-08", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces DeltaReplay, which stores mobile GUI trajectories as transition graphs and decides at each step whether to reuse an action, substitute task-specific parameters, or defer to the base agent. This supports reuse of partially matching experiences without rewriting stored memory.", "topics": ["memory", "mobile-navigation", "planning"], "updated_at": "2026-10-08", "url": "https://arxiv.org/abs/2610.11707"}
{"aliases": [], "authors": ["Seungjun Yu", "Hojun Choi", "Seojeong Park", "Jaeyo Shin", "Hyunjung Shim"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "191e88c19124ddaddc2fed2f0e3f8e197a4c0c7f8adbc3997c056d3333754042", "submitted_at": "2026-10-09T18:45:11Z", "updated_at": "2026-10-09T18:45:11Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.14228v1"}}, "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:2610.14228", "identifiers": {"arxiv": "2610.14228"}, "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/2610.14228"}], "name": "Latent World Action Model 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": "2026-10-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "Replaces future-screenshot generation with a shared latent representation for GUI action selection and action-conditioned transition prediction. An action-aware contrastive objective encourages the transition representation to retain discriminative action-induced changes.", "topics": ["world-models", "planning", "efficiency"], "updated_at": "2026-10-09", "url": "https://arxiv.org/abs/2610.14228"}
{"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": ["Zirui Shang", "Yang Liu", "Xin Shu", "Zhi Gao", "Xinxiao Wu", "Lifeng Fan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "46ea5a1a350ccfcc12d22ab3783e05e3c580719962831561b01aeb3fcc6f6c28", "submitted_at": "2026-10-13T04:37:18Z", "updated_at": "2026-10-13T04:37:18Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.17710v1"}}, "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:2610.17710", "identifiers": {"arxiv": "2610.17710"}, "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/2610.17710"}], "name": "Adaptive-Group Process Rewards 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": "2026-10-13", "relations": [], "review_status": "curated", "scope": "core", "summary": "Groups GUI trajectories by subgoal and combines action-level and group-level assessments into intermediate rewards for training agents with sparse task outcomes.", "topics": ["reinforcement-learning", "long-horizon", "planning"], "updated_at": "2026-10-13", "url": "https://arxiv.org/abs/2610.17710"}
{"aliases": [], "authors": ["Hanzhang Zhou", "Long Li", "Quyu Kong", "Lei Wang", "Panrong Tong", "Long Chen", "Tianyu Xia", "Gongjie Zhang", "Xu Zhang", "Chen Liu", "Jianan Zhang", "Chenglin Cai", "Zhengwen Zeng", "Steven Hoi", "Yue Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "beab15a2bf8a6aca19d219dbe0a047e8c6d6bd84a580df33633b5d366c175a0e", "submitted_at": "2026-10-15T16:51:35Z", "updated_at": "2026-10-15T16:51:35Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.22200v1"}}, "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:2610.22200", "identifiers": {"arxiv": "2610.22200"}, "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/2610.22200"}], "name": "OpenGCUA: Toward a Fully Open Stack for Generalist Computer Use", "notes": "Editorial summary based on the primary paper abstract; experiments and released artifacts have not been independently reproduced.", "published_at": "2026-10-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces an open generalist computer-use stack combining visual and programmatic interaction, with model-controlled observation and task-state decisions.", "topics": ["tool-integration", "planning"], "updated_at": "2026-10-15", "url": "https://arxiv.org/abs/2610.22200"}
{"aliases": [], "authors": ["Qingni Wang", "Changdae Oh", "Yuzhe Yang", "Sharon Li", "Jiachen Yang", "Ang Li", "Xin Eric Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6a249a56a61ebb14b456500f9b245efe7b72997af5d5f515e5cc7707fe68b018", "submitted_at": "2026-10-15T17:40:12Z", "updated_at": "2026-10-15T17:40:12Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.22280v1"}}, "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:2610.22280", "identifiers": {"arxiv": "2610.22280"}, "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/2610.22280"}], "name": "LEASH: Uncertainty-Aware Weak-to-Strong Escalation and Oversight for Computer-Use Agents", "notes": "Editorial summary based on the primary paper abstract; experiments and released artifacts have not been independently reproduced.", "published_at": "2026-10-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "Uses weak-agent uncertainty to decide when to escalate to a stronger model and monitor its subsequent task execution.", "topics": ["efficiency", "planning"], "updated_at": "2026-10-15", "url": "https://arxiv.org/abs/2610.22280"}
{"aliases": [], "authors": ["Umar Iqbal"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "077dd623c980813e83197a38b6545161a3d2fc842d99fe291c53372b9770acd2", "submitted_at": "2026-10-15T22:02:45Z", "updated_at": "2026-10-15T22:02:45Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.22697v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-20T05:52:01Z", "role": "primary", "sha256": "63e37cd26bfe69455a3fa835f3d6ba4cfcc5bb7dc91f90505920d37b1e08f0b6", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2610.19384%2C2610.22672%2C2610.22697%2C2610.23303%2C2610.23307%2C2610.24729&max_results=6"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T13:38:46.444001Z", "role": "primary", "sha256": "cb6a8a9c0cc47be21b36b3d1831b26c6189d8f2e4af0d7e26d643c9c2a65452f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/html/2610.22697v1"}], "id": "arxiv:2610.22697", "identifiers": {"arxiv": "2610.22697"}, "kind": "paper", "last_verified_at": "2026-10-20T13:38:46.444001Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2610.22697"}], "name": "What It Is, Not What It Says: Securing Agentic Browsing Interactions with Web Semantics", "notes": "Evaluation instructs the model to simulate compromise and measures planner outputs and permission appearance. Utility uses plan matching and LLM-judged element suitability; no steps are executed in that assessment. Assumes an honest first party, trusted browser/harness and untampered instrumentation; multi-page and availability attacks are outside scope. Parallel latency reduction is an unimplemented estimate. No independent reproduction.", "published_at": "2026-10-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "Intact records webpage element provenance and capabilities, isolates planning per element, and uses a browser harness to constrain actions and request consent for third-party interactions.", "topics": ["safety-security", "web-navigation", "planning", "human-in-the-loop"], "updated_at": "2026-10-15", "url": "https://arxiv.org/abs/2610.22697"}
{"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": [], "details": {"component_path": "Mobile-Agent-E", "license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/LICENSE"], "reason": "Reviewed repository license declaration for code; third-party dependencies, data, weights and executable environment images are separate."}}, "repository_tree_sha": "11cea575561fb7800b5fb6b6cafa56f7a91de11f", "repository_url": "https://github.com/X-PLUG/MobileAgent"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:27:15Z", "role": "primary", "sha256": "f8515fbce73ea0ef3d1d558d4567bb949244e6384e61c956ba3ddbc27294f0ae", "supports": ["name", "kind", "environments", "scope", "notes", "relations"], "url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/Mobile-Agent-E/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T15:37:03Z", "role": "primary", "sha256": "d403251020124bebcd9f180f800c8ea9f18a3f117c82fcf04c3cbe9ffe0e35f9", "supports": ["identity"], "url": "https://api.github.com/repos/X-PLUG/MobileAgent"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:25:21Z", "role": "primary", "sha256": "7293efadde7c96b64ff8483b2157241e17d0b0fbfff4839de9c7b11485b07cbb", "supports": ["identity", "release_status", "details.repository_tree_sha", "relations"], "url": "https://api.github.com/repos/X-PLUG/MobileAgent/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["summary", "topics"], "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": "2026-10-16T06:42:03Z", "role": "primary", "sha256": "961819a89dacd1f4f161e1d2ebf616b538bf095a2eecfa3ef40b7ccec633fac4", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/LICENSE"}], "id": "component:x-plug/mobileagent/mobile-agent-e", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-18T04:27:15Z", "licenses": {"code": {"basis": "Reviewed repository license declaration for code; third-party dependencies, data, weights and executable environment images are separate.", "declared": "MIT", "retrieved_at": "2026-10-16T06:42:03Z", "url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/X-PLUG/MobileAgent/tree/main/Mobile-Agent-E"}], "name": "Mobile-Agent-E code component", "notes": "The inspected README explicitly supports Android only, with experiments on Samsung Galaxy A15; host-machine macOS/Linux instructions do not imply control of those operating systems. Agent code, launch scripts and benchmark definitions are present. The root MIT code license does not establish terms for the separately published task dataset or external models.", "published_at": null, "relations": [{"evidence_url": "https://api.github.com/repos/X-PLUG/MobileAgent/git/trees/main?recursive=1", "target_id": "github:x-plug/mobileagent", "type": "component_of"}, {"evidence_url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/Mobile-Agent-E/README.md", "target_id": "arxiv:2501.11733", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Android automation component with hierarchical planning, reflection and persistent tips/shortcuts learned across task sequences.", "topics": ["planning", "memory", "self-reflection", "mobile-navigation"], "updated_at": null, "url": "https://github.com/X-PLUG/MobileAgent/tree/main/Mobile-Agent-E"}
{"aliases": [], "authors": [], "details": {"component_path": "PC-Agent", "license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/LICENSE"], "reason": "Reviewed repository license declaration for code; third-party dependencies, data, weights and executable environment images are separate."}}, "repository_tree_sha": "11cea575561fb7800b5fb6b6cafa56f7a91de11f", "repository_url": "https://github.com/X-PLUG/MobileAgent"}, "environments": ["desktop", "windows", "macos"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:27:15Z", "role": "primary", "sha256": "ca5b488f10c229b79cc6f9e1f32007d7d84c34366492a10b90d0d66622e77273", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations"], "url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/PC-Agent/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T15:37:03Z", "role": "primary", "sha256": "d403251020124bebcd9f180f800c8ea9f18a3f117c82fcf04c3cbe9ffe0e35f9", "supports": ["identity"], "url": "https://api.github.com/repos/X-PLUG/MobileAgent"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:25:21Z", "role": "primary", "sha256": "7293efadde7c96b64ff8483b2157241e17d0b0fbfff4839de9c7b11485b07cbb", "supports": ["identity", "release_status", "details.repository_tree_sha", "relations"], "url": "https://api.github.com/repos/X-PLUG/MobileAgent/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:42:03Z", "role": "primary", "sha256": "961819a89dacd1f4f161e1d2ebf616b538bf095a2eecfa3ef40b7ccec633fac4", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/LICENSE"}], "id": "component:x-plug/mobileagent/pc-agent", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-18T04:27:15Z", "licenses": {"code": {"basis": "Reviewed repository license declaration for code; third-party dependencies, data, weights and executable environment images are separate.", "declared": "MIT", "retrieved_at": "2026-10-16T06:42:03Z", "url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/X-PLUG/MobileAgent/tree/main/PC-Agent"}], "name": "PC-Agent (X-PLUG component)", "notes": "Scoped to PC-Agent in the MobileAgent monorepo and distinct from GAIR-NLP/PC-Agent. The source tree contains current and v1 implementations; the README states a March 2025 code update. Root MIT covers repository code; the separately linked PC-Eval tasks and application software are not assigned that license by this record.", "published_at": null, "relations": [{"evidence_url": "https://api.github.com/repos/X-PLUG/MobileAgent/git/trees/main?recursive=1", "target_id": "github:x-plug/mobileagent", "type": "component_of"}, {"evidence_url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/PC-Agent/README.md", "target_id": "arxiv:2502.14282", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Multi-agent desktop automation component for Windows and macOS, combining active perception, task decomposition and reflection for workflows spanning applications.", "topics": ["planning", "cross-app", "self-reflection"], "updated_at": null, "url": "https://github.com/X-PLUG/MobileAgent/tree/main/PC-Agent"}
{"aliases": [], "authors": [], "details": {}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:07Z", "role": "primary", "sha256": "b68b2cc26bfc6a654fec35f3c53c1b1334f3d617671673407d3ed3b59f3db973", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/ai-nikolai/StateAct/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:06Z", "role": "primary", "sha256": "38f8c08395c6e67ff9bf52b27e772667f193a172bd6f726507c57776ad3e5e5f", "supports": ["identity"], "url": "https://api.github.com/repos/ai-nikolai/StateAct"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:07Z", "role": "primary", "sha256": "97841e6eab1002bcea949e031f5abd3567482ebbcf03b991bd64643027c468bf", "supports": ["release_status"], "url": "https://api.github.com/repos/ai-nikolai/StateAct/git/trees/main?recursive=1"}], "id": "github:ai-nikolai/stateact", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-17T02:02:07Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/ai-nikolai/StateAct"}], "name": "StateAct", "notes": "The repository contains a WebShop runner as well as ALFWorld and TextCraft experiments. Inclusion concerns its web-shopping interaction component. No code license or downloadable trained checkpoint is declared in the inspected root metadata.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/ai-nikolai/StateAct/HEAD/README.md", "target_id": "arxiv:2410.02810", "type": "documented_by"}], "review_status": "curated", "scope": "adjacent", "summary": "State-tracking prompts and experiment runners for language agents, including a WebShop implementation alongside text-based environments.", "topics": ["planning", "state-tracking", "web-navigation"], "updated_at": null, "url": "https://github.com/ai-nikolai/StateAct"}
{"aliases": [], "authors": [], "details": {"availability": "Public Android automation source and setup instructions; external models/services required.", "license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/AkimotoAyako/VisionTasker/HEAD/LICENSE"], "reason": "Repository GNU GPL version 3 license; pretrained models and linked data retain separate terms."}}, "repository_revision": "d31b50e98bde28c0d1cfbc7d7ee2528d10b3fbb0"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:20:18Z", "role": "primary", "sha256": "816981b343628c53188fd27041d62f52854fcfe83c976a125b84545f5e010274", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations"], "url": "https://raw.githubusercontent.com/AkimotoAyako/VisionTasker/HEAD/readme.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:18:23Z", "role": "primary", "sha256": "94d8e95d302160e3c581015267dcfbb9f61c41ad1f04e470fcd71f5ee1ffa25d", "supports": ["identity"], "url": "https://api.github.com/repos/AkimotoAyako/VisionTasker"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:19:48Z", "role": "primary", "sha256": "54a1c08d44e0ad1d60bee57e3db2842518e014db3be8b08362e484315f3a6a6f", "supports": ["release_status", "details.repository_revision"], "url": "https://api.github.com/repos/AkimotoAyako/VisionTasker/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:23:03Z", "role": "primary", "sha256": "3972dc9744f6499f0f9b2dbf76696f2ae7ad8af9b23dde66d6af86c9dfb36986", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/AkimotoAyako/VisionTasker/HEAD/LICENSE"}], "id": "github:akimotoayako/visiontasker", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-17T11:20:18Z", "licenses": {"code": {"basis": "Repository GNU GPL version 3 license; pretrained models and linked data retain separate terms.", "declared": "GPL-3.0", "retrieved_at": "2026-10-17T11:23:03Z", "url": "https://raw.githubusercontent.com/AkimotoAyako/VisionTasker/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/AkimotoAyako/VisionTasker"}], "name": "VisionTasker", "notes": "The official lowercase readme.md and source tree expose automatic and step-by-step entry points. The documented setup uses a Windows host and an Android phone connected through ADB; it also requires separately obtained detection/CLIP models and model/OCR services. GPL-3.0 is recorded for the code only. External model download links were not used to infer weight licenses or verify a separate checkpoint release.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/AkimotoAyako/VisionTasker/HEAD/readme.md", "target_id": "arxiv:2312.11190", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "An Android automation implementation that translates screenshot-based UI understanding into language descriptions for step-by-step LLM planning and action execution.", "topics": ["mobile-navigation", "screen-understanding", "planning"], "updated_at": null, "url": "https://github.com/AkimotoAyako/VisionTasker"}
{"aliases": [], "authors": [], "details": {}, "environments": ["desktop", "windows"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T13:14:29Z", "role": "primary", "sha256": "281793fe6b875bd1a63591502345ac9523b1127acd9f41d5c2c88933a357e4f2", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/Alokia/COLA-demo/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["relations", "environments"], "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": "github:alokia/cola-demo", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-16T13:14:29Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/Alokia/COLA-demo"}], "name": "COLA demo", "notes": "The publisher describes this as demonstration code and states that the full implementation is not released.", "published_at": null, "relations": [{"evidence_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", "target_id": "arxiv:2503.09263", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Research demonstration of collaborative Windows GUI automation with dynamic task scheduling and recovery mechanisms.", "topics": ["automation", "planning", "human-in-the-loop"], "updated_at": null, "url": "https://github.com/Alokia/COLA-demo"}
{"aliases": [], "authors": [], "details": {}, "environments": ["desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:02Z", "role": "primary", "sha256": "fe475d3e92c42807b792c542bc5986bf74caa7d166aea860d368fbe776342a82", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/BAAI-Agents/Cradle/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T06:42:04Z", "role": "primary", "sha256": "2964861ad96d977da95d891f62d8f5d60873f4a36da0b51f052d9e98ef97d02f", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/BAAI-Agents/Cradle/HEAD/LICENSE"}], "id": "github:baai-agents/cradle", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-16T06:36:02Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "MIT", "retrieved_at": "2026-10-16T06:42:04Z", "url": "https://raw.githubusercontent.com/BAAI-Agents/Cradle/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/BAAI-Agents/Cradle"}], "name": "Cradle", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/BAAI-Agents/Cradle/HEAD/README.md", "target_id": "arxiv:2403.03186", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Framework in which foundation models control games and software from screenshots alone, acting through keyboard and mouse, shown on RDR2, Stardew Valley, Cities: Skylines and Dealer's Life 2.", "topics": ["planning", "memory"], "updated_at": null, "url": "https://github.com/BAAI-Agents/Cradle"}
{"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": ["cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:35:59Z", "role": "primary", "sha256": "0c4feb17b657c0c9a71d7e0c5e2142abbd19bb5cff22e6cd3b6b082644c98ee7", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/bytedance/UI-TARS/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/bytedance/UI-TARS/HEAD/LICENSE"}], "id": "github:bytedance/ui-tars", "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/bytedance/UI-TARS/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/bytedance/UI-TARS"}], "name": "UI-TARS", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/bytedance/UI-TARS/HEAD/README.md", "target_id": "arxiv:2501.12326", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Native GUI agent models; UI-TARS-1.5 adds reinforcement-learned reasoning before acting and an open 7B checkpoint, and UI-TARS-2 extends to games, code and tool use.", "topics": ["grounding", "planning"], "updated_at": null, "url": "https://github.com/bytedance/UI-TARS"}
{"aliases": [], "authors": [], "details": {"license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/caap-agent/caap-agent/HEAD/LICENSE"], "reason": "Reviewed repository license declaration for code; third-party dependencies, data, weights and executable environment images are separate. The inspected text does not establish -only versus -or-later for the project."}}, "repository_tree_sha": "9d65c0f30e3c119db533dd832863b4f3f7b74b18", "repository_url": "https://github.com/caap-agent/caap-agent"}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:27:14Z", "role": "primary", "sha256": "481db1288ffc769f48b19c83f0db4fb40a64dc1b05859f776b77f0774bfcdb90", "supports": ["name", "kind", "environments", "topics", "scope", "notes"], "url": "https://raw.githubusercontent.com/caap-agent/caap-agent/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T15:37:02Z", "role": "primary", "sha256": "8d518cad746591e82b54b63b79d0a1d47531a75f02c8baf2b560877fb03918ba", "supports": ["identity"], "url": "https://api.github.com/repos/caap-agent/caap-agent"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:25:24Z", "role": "primary", "sha256": "c511d94ce58e73a87c9647169b1199c0ed0029e6b1df271ad1fcd3ca86f3458b", "supports": ["identity", "release_status", "details.repository_tree_sha"], "url": "https://api.github.com/repos/caap-agent/caap-agent/git/trees/master?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:28Z", "role": "primary", "sha256": "be244bfae7f89ef9449b69d85aac44295eb7b5f7e5e1bbedd76df312696676a7", "supports": ["summary", "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"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:35:20Z", "role": "primary", "sha256": "5e6fa1bfd473724e67366c42761db27fe091ee00fdcfbe599b9dd5d5a9311d36", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/caap-agent/caap-agent/HEAD/LICENSE"}], "id": "github:caap-agent/caap-agent", "identifiers": {}, "kind": "agent", "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. The inspected text does not establish -only versus -or-later for the project.", "declared": "AGPL-3.0", "retrieved_at": "2026-10-18T04:35:20Z", "url": "https://raw.githubusercontent.com/caap-agent/caap-agent/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/caap-agent/caap-agent"}], "name": "CAAP Agent", "notes": "Code is released for the two named browser benchmarks. Visual-observer checkpoints are linked separately through Google Drive and were not downloaded or access-tested. Root LICENSE begins with GNU AGPL v3 and includes third-party notices; this code declaration is not transferred to bundled/generated data or external weights.", "published_at": null, "relations": [{"evidence_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", "target_id": "arxiv:2406.06947", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Screenshot-driven agent implementation for MiniWoB++ and WebShop, using a visual observer and context-aware prompts to issue mouse and keyboard actions.", "topics": ["planning", "screen-understanding", "web-navigation"], "updated_at": null, "url": "https://github.com/caap-agent/caap-agent"}
{"aliases": ["github:cooelf/auto-ui", "cooelf/Auto-UI", "https://github.com/cooelf/Auto-UI"], "authors": [], "details": {}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T10:42:12Z", "role": "primary", "sha256": "13f045fa84dc0ad68dcceea95fc7e0e34a45bcda866d589132351559b87a8e30", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/cooelf/Auto-GUI/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T10:48:41Z", "role": "primary", "sha256": "c71d239df91726fc519c6eb72d318ec65820627232b2f796219e87dcf35d0ab4", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/cooelf/Auto-GUI/HEAD/LICENSE"}, {"method": "github_repository_identity", "retrieved_at": "2026-10-17T14:13:59Z", "role": "primary", "sha256": "569f3c779e6a0b8210a764c9096e49e8a5199c107a64f7fcfb7a011257bf2a12", "supports": ["aliases"], "url": "https://api.github.com/repos/cooelf/Auto-UI"}, {"method": "github_repository_identity", "retrieved_at": "2026-10-17T14:14:00Z", "role": "primary", "sha256": "569f3c779e6a0b8210a764c9096e49e8a5199c107a64f7fcfb7a011257bf2a12", "supports": ["aliases"], "url": "https://api.github.com/repos/cooelf/Auto-GUI"}], "id": "github:cooelf/auto-gui", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-16T10:42:12Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "Apache-2.0", "retrieved_at": "2026-10-16T10:48:41Z", "url": "https://raw.githubusercontent.com/cooelf/Auto-GUI/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/cooelf/Auto-GUI"}], "name": "Auto-GUI / Auto-UI", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/cooelf/Auto-GUI/HEAD/README.md", "target_id": "arxiv:2309.11436", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Screenshot-based mobile agent implementation with training and inference code for chain-of-action prediction.", "topics": ["mobile-navigation", "planning"], "updated_at": null, "url": "https://github.com/cooelf/Auto-GUI"}
{"aliases": ["Droidrun"], "authors": [], "details": {"availability": "Published source; official release metadata inspected", "release_tag": "v0.7.0"}, "environments": ["android", "ios"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T03:59:53Z", "role": "primary", "sha256": "80052560e06db353277a55f3f8f6a4b5231aa6dc1376d5182559b72adcb5c3a1", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://raw.githubusercontent.com/droidrun/mobilerun/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:01:56Z", "role": "primary", "sha256": "e48b9154a58b4b7021a4707ca185af32b9dd81c828d53ba36a86f195c1720344", "supports": ["identity"], "url": "https://api.github.com/repos/droidrun/mobilerun"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T03:59:53Z", "role": "primary", "sha256": "7f40d42acd62b4405b5077a0f594a35aae73fb057589346afa5f6fb22d710a0c", "supports": ["release_status"], "url": "https://api.github.com/repos/droidrun/mobilerun/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:00:56Z", "role": "primary", "sha256": "f85c8b766edc107c455a86e363c56af3d1f7fce669b55377f4010982aa3a7e44", "supports": ["release_status"], "url": "https://api.github.com/repos/droidrun/mobilerun/releases/latest"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:00:57Z", "role": "primary", "sha256": "b3ebb19fda8286efa415d457529eafff41e1b1a27fe4e00d99f34e1559b6cdd9", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/droidrun/mobilerun/HEAD/LICENSE"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T03:59:53Z", "role": "primary", "sha256": "e48b9154a58b4b7021a4707ca185af32b9dd81c828d53ba36a86f195c1720344", "supports": ["identity", "aliases", "notes"], "url": "https://api.github.com/repos/droidrun/droidrun"}, {"method": "license_header", "retrieved_at": "2026-10-17T04:00:57Z", "role": "primary", "sha256": "b3ebb19fda8286efa415d457529eafff41e1b1a27fe4e00d99f34e1559b6cdd9", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/droidrun/mobilerun/HEAD/LICENSE"}], "id": "github:droidrun/mobilerun", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-17T03:59:53Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "MIT", "retrieved_at": "2026-10-17T04:00:57Z", "url": "https://raw.githubusercontent.com/droidrun/mobilerun/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/droidrun/mobilerun"}], "name": "Mobilerun (formerly Droidrun)", "notes": "The former droidrun/droidrun repository resolves to droidrun/mobilerun through the official API. Source and release v0.7.0 are public. Android and iOS setup paths differ, and the companion Portal app is a separate device bridge. MIT applies to framework code; model-provider and cloud-service terms are separate.", "published_at": null, "relations": [], "review_status": "curated", "scope": "core", "summary": "A framework that combines screenshots and accessibility state with agent tools for mobile-device tasks through command-line and Python interfaces.", "topics": ["mobile-navigation", "tool-integration", "planning"], "updated_at": null, "url": "https://github.com/droidrun/mobilerun"}
{"aliases": ["github:mobilegptsys/mobilegpt", "mobilegptsys/MobileGPT", "https://github.com/mobilegptsys/MobileGPT"], "authors": [], "details": {"availability": "Public Android client, Python server and manual offline-explorer source.", "license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/fclab-skku/MobileGPT/HEAD/LICENSE"], "reason": "Repository source license; linked benchmark data and external models retain separate terms."}}, "repository_revision": "7945048e6c05849aff1633744cbdf653c58bece2"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:32:07Z", "role": "primary", "sha256": "8683cff296d21c1a7a38bc372efaf004b8af644030fac77a4dc9d2b2e21d894b", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://raw.githubusercontent.com/fclab-skku/MobileGPT/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:30:55Z", "role": "primary", "sha256": "3d3ceb084d374b706842cff1f4ec986fde50c13cba141dc4619d5afe8b631d55", "supports": ["identity"], "url": "https://api.github.com/repos/mobilegptsys/MobileGPT"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:32:06Z", "role": "primary", "sha256": "2a8f30b3422a033b345c2054d55adff0836b7dac163e600583ba0db1b4423320", "supports": ["release_status", "details.repository_revision"], "url": "https://api.github.com/repos/fclab-skku/MobileGPT/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:24Z", "role": "primary", "sha256": "3aac6e9748769efff3c475ede0da192776daa03431d231b5f67e2aa50f8a3be9", "supports": ["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"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:34:10Z", "role": "primary", "sha256": "c71d239df91726fc519c6eb72d318ec65820627232b2f796219e87dcf35d0ab4", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/fclab-skku/MobileGPT/HEAD/LICENSE"}, {"method": "github_repository_identity", "retrieved_at": "2026-10-17T14:14:04Z", "role": "primary", "sha256": "3d3ceb084d374b706842cff1f4ec986fde50c13cba141dc4619d5afe8b631d55", "supports": ["aliases"], "url": "https://api.github.com/repos/mobilegptsys/MobileGPT"}, {"method": "github_repository_identity", "retrieved_at": "2026-10-17T14:14:06Z", "role": "primary", "sha256": "3d3ceb084d374b706842cff1f4ec986fde50c13cba141dc4619d5afe8b631d55", "supports": ["aliases"], "url": "https://api.github.com/repos/fclab-skku/MobileGPT"}], "id": "github:fclab-skku/mobilegpt", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-17T12:32:07Z", "licenses": {"code": {"basis": "Repository source license; linked benchmark data and external models retain separate terms.", "declared": "Apache-2.0", "retrieved_at": "2026-10-17T12:34:10Z", "url": "https://raw.githubusercontent.com/fclab-skku/MobileGPT/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/fclab-skku/MobileGPT"}], "name": "MobileGPT", "notes": "The old mobilegptsys repository resolves to fclab-skku/MobileGPT. Public Android client and Python server sources require Android SDK 33+, Accessibility Service setup and external service credentials. The README explicitly omits the random explorer and human-in-the-loop memory repair from the open-source version. Its benchmark description differs from the retained arXiv snapshot, so no dataset counts or reproduction claim are merged into this code record. External benchmark data was not approved as a separate release.", "published_at": null, "relations": [{"evidence_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", "target_id": "arxiv:2312.03003", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "An Android client and Python-server agent that learns reusable app subtasks through exploration, selection, derivation and memory recall.", "topics": ["mobile-navigation", "memory", "planning", "workflow"], "updated_at": null, "url": "https://github.com/fclab-skku/MobileGPT"}
{"aliases": [], "authors": [], "details": {}, "environments": ["desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:03Z", "role": "primary", "sha256": "178774927f003b5221d695fa78ded6fcc543aa8653c7b2dd8016418e6911afb7", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/GAIR-NLP/PC-Agent/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T06:42:04Z", "role": "primary", "sha256": "aee22091f60f10968afb7dc33dd8b728ea929e287c260ed82adffdec4866fcb5", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/GAIR-NLP/PC-Agent/HEAD/LICENSE"}], "id": "github:gair-nlp/pc-agent", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-16T06:36:03Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "MIT", "retrieved_at": "2026-10-16T06:42:04Z", "url": "https://raw.githubusercontent.com/GAIR-NLP/PC-Agent/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/GAIR-NLP/PC-Agent"}], "name": "PC-Agent", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/GAIR-NLP/PC-Agent/HEAD/README.md", "target_id": "arxiv:2412.17589", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Human-cognition-transfer framework for PC agents: PC Tracker logs human computer use, cognition completion turns logs into cognitive trajectories, and a planner works with a grounding agent.", "topics": ["planning", "cross-app"], "updated_at": null, "url": "https://github.com/GAIR-NLP/PC-Agent"}
{"aliases": [], "authors": [], "details": {"availability": "Public CoAT prompting/action-prediction and evaluation source.", "license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/IMNearth/CoAT/HEAD/README.md", "https://api.github.com/repos/IMNearth/CoAT/git/trees/main?recursive=1"], "reason": "Apache-2.0 badge alone does not establish component-scoped source terms; the inspected non-truncated tree has no license file."}}, "repository_revision": "9ab1a9f0e76581c50c84f8ebc6ada1a2084a0c71"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:30:53Z", "role": "primary", "sha256": "7a9fb905b64f2760a137cd267a5e38dd343df1821eda8172b606ab7500d27731", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations", "licenses.code"], "url": "https://raw.githubusercontent.com/IMNearth/CoAT/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:30:53Z", "role": "primary", "sha256": "d0c5056c9a74913f84a4f809bf01d9fd9ea325ac0a91b4c4a0a51220b2da9d68", "supports": ["identity"], "url": "https://api.github.com/repos/IMNearth/CoAT"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:32:04Z", "role": "primary", "sha256": "29d84ed12dc25035fd539604b06f6cc113080b389edeb9aa605cfa55c5f28151", "supports": ["release_status", "details.repository_revision", "licenses.code"], "url": "https://api.github.com/repos/IMNearth/CoAT/git/trees/main?recursive=1"}], "id": "github:imnearth/coat", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-17T12:30:53Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/IMNearth/CoAT"}], "name": "CoAT research implementation", "notes": "The tree contains run_coat.py and coat/agent.py, model.py and evaluate.py, with API-backed model configuration. This verifies a released annotation/prediction implementation, not a live phone controller or a released checkpoint. AiTZ screens/annotations are linked through external Drive/Baidu bundles whose complete download inventory was not verified here. The README has an Apache-2.0 badge, but no license file or explicit component grant is present; code/data/weight rights remain unestablished.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/IMNearth/CoAT/HEAD/README.md", "target_id": "arxiv:2403.02713", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Research scripts that generate screen descriptions, action reasoning and expected outcomes, then predict Android GUI actions using the resulting Chain-of-Action-Thought context.", "topics": ["planning", "mobile-navigation", "action-prediction"], "updated_at": null, "url": "https://github.com/IMNearth/CoAT"}
{"aliases": [], "authors": [], "details": {}, "environments": ["cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:35:59Z", "role": "primary", "sha256": "5703c892aaf99d250bf59edf52d359560a80d443f2be13c7aa76af0c1a7f1d67", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/inclusionAI/UI-Venus/HEAD/README.md"}], "id": "github:inclusionai/ui-venus", "identifiers": {}, "kind": "model", "last_verified_at": "2026-10-16T06:35:59Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/inclusionAI/UI-Venus"}], "name": "UI-Venus", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/inclusionAI/UI-Venus/HEAD/README.md", "target_id": "arxiv:2508.10833", "type": "documented_by"}, {"evidence_url": "https://raw.githubusercontent.com/inclusionAI/UI-Venus/HEAD/README.md", "target_id": "arxiv:2602.09082", "type": "documented_by"}, {"evidence_url": "https://raw.githubusercontent.com/inclusionAI/UI-Venus/HEAD/README.md", "target_id": "arxiv:2609.00028", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "GUI agent model family; UI-Venus-2, at 9B and 27B, acts on mobile, web and desktop in one end-to-end agent, trained with keypoint-based multi-judge task verification.", "topics": ["grounding", "planning"], "updated_at": null, "url": "https://github.com/inclusionAI/UI-Venus"}
{"aliases": [], "authors": [], "details": {"availability": "Public memory, prompting, browser-experiment and fine-tuning source.", "license_conflict": "README MIT badge versus NTUITIVE agreement; the agreement contains inconsistent distribution provisions.", "license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/ltzheng/synapse/HEAD/README.md", "https://raw.githubusercontent.com/ltzheng/Synapse/HEAD/LICENSE"], "reason": "Unresolved conflict between README MIT badge and NTUITIVE non-commercial LICENSE, including inconsistent redistribution clauses. Keep permissions unknown rather than choose an interpretation."}}, "repository_revision": "08c3a252ff158822e50e9c109c1cd49dc2f6147d"}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:30:54Z", "role": "primary", "sha256": "ed5416d91665017fca4b354016b7c2c748acf32d7723483fecb5387a22a1fdff", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "licenses.code"], "url": "https://raw.githubusercontent.com/ltzheng/synapse/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:30:54Z", "role": "primary", "sha256": "79a238a6f486d2a8970baba199f0968d28f14d42922c86ce71753eb687e0e9c6", "supports": ["identity"], "url": "https://api.github.com/repos/ltzheng/synapse"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:32:05Z", "role": "primary", "sha256": "785a7163dac9f41b1d0ca97f8dd9311b6df6a09f3a5163438dd64db226a4ceaf", "supports": ["release_status", "details.repository_revision"], "url": "https://api.github.com/repos/ltzheng/Synapse/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["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"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:34:09Z", "role": "primary", "sha256": "cc4bce2274289d2b924fcc3c185c68b8e7c539365c5f8c72be865883244088ec", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/ltzheng/Synapse/HEAD/LICENSE"}], "id": "github:ltzheng/synapse", "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/ltzheng/Synapse"}], "name": "Synapse", "notes": "Training, memory-building and evaluation scripts for MiniWoB++ and Mind2Web are present. External benchmark data, model/API credentials and legacy dependencies are still required. License remains unknown: the README shows MIT, while LICENSE is an NTUITIVE non-commercial agreement that both permits some distribution and separately says distribution is prohibited. No interpretation resolves those contradictions, and no commercial or non-commercial permission is asserted. This admission confirms the released GUI implementation, not a licensing grant.", "published_at": null, "relations": [{"evidence_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", "target_id": "arxiv:2306.07863", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "A computer-control research implementation that retrieves example trajectories from memory, abstracts task-relevant state and prompts an agent to navigate browser tasks.", "topics": ["memory", "planning", "web-navigation", "imitation-learning"], "updated_at": null, "url": "https://github.com/ltzheng/Synapse"}
{"aliases": [], "authors": [], "details": {"availability": "Public conversational planning, candidate-generation and training/evaluation source.", "license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/magicgh/Self-MAP/HEAD/LICENSE"], "reason": "Project source LICENSE; separate dataset license is not folded into the code slot."}}, "repository_revision": "6cf779682b70dae597626f96dfa5130b202689bc"}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:30:55Z", "role": "primary", "sha256": "693197cb807e3ab4d5afda9de35dd51342b4feb18e68b8ccfcdaa370399dbeb3", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations"], "url": "https://raw.githubusercontent.com/magicgh/self-map/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:30:54Z", "role": "primary", "sha256": "82c425841c94bf129c4fbd1e70029c581aa1488d5f85cdd23aa0f292092146f0", "supports": ["identity"], "url": "https://api.github.com/repos/magicgh/self-map"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:32:05Z", "role": "primary", "sha256": "8d844f227c457321640231ab3a4953457e3c6cfeab8d8f4803f8aa01dc15a8f8", "supports": ["release_status", "details.repository_revision"], "url": "https://api.github.com/repos/magicgh/Self-MAP/git/trees/main?recursive=1"}, {"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"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:34:10Z", "role": "primary", "sha256": "f271c479b19a620600d18fc2265224205bf71483b969672d3f6acd9bc552b3f5", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/magicgh/Self-MAP/HEAD/LICENSE"}], "id": "github:magicgh/self-map", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-17T12:30:55Z", "licenses": {"code": {"basis": "Project source LICENSE; separate dataset license is not folded into the code slot.", "declared": "MIT", "retrieved_at": "2026-10-17T12:34:10Z", "url": "https://raw.githubusercontent.com/magicgh/Self-MAP/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/magicgh/Self-MAP"}], "name": "Self-MAP", "notes": "The source tree provides candidate-generation and action-prediction training/evaluation scripts. Its documented workflow requires separately downloaded MT-Mind2Web data and configured model/API paths. The README cites the ACL publication by DOI; a retained original arXiv abstract explicitly names Self-MAP and MT-Mind2Web and supplies the catalog paper relation without a new bibliography entity. MIT is the code license; the separate data artifact declares CC-BY-4.0.", "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:2402.15057", "type": "documented_by"}, {"evidence_url": "https://raw.githubusercontent.com/magicgh/self-map/HEAD/README.md", "target_id": "url:bdfb50330deb97893ce4", "type": "uses"}], "review_status": "curated", "scope": "core", "summary": "A conversational web-agent implementation with candidate ranking and memory-assisted action planning, including optional memory simplification, refinement and self-reflection modules.", "topics": ["web-navigation", "memory", "self-reflection", "planning"], "updated_at": null, "url": "https://github.com/magicgh/Self-MAP"}
{"aliases": [], "authors": [], "details": {}, "environments": ["desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:35:59Z", "role": "primary", "sha256": "b00a072dfde61f774db6acf94fcde9bd32337b1b4f16a178998378d842bbbe4d", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/meituan/EvoCUA/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T06:42:00Z", "role": "primary", "sha256": "a8d1d88a0f9cce3cdbff73b03a80dc5e5d69e3ba2443946c268815a550c9e0ec", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/meituan/EvoCUA/HEAD/LICENSE"}], "id": "github:meituan/evocua", "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/meituan/EvoCUA/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/meituan/EvoCUA"}], "name": "EvoCUA", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/meituan/EvoCUA/HEAD/README.md", "target_id": "arxiv:2601.15876", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Computer-use model and training resources from the EvoCUA project.", "topics": ["reinforcement-learning", "planning"], "updated_at": null, "url": "https://github.com/meituan/EvoCUA"}
{"aliases": [], "authors": [], "details": {}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:00Z", "role": "primary", "sha256": "ce11bdfe8354eb9c3b8a985dd5ddd21f88f94fb5fb9d00c9b01a76226f0b0a36", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/microsoft/fara/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T06:42:00Z", "role": "primary", "sha256": "98a96c38494df4963951e32b81ec4effebe4c8a812ab8d9e7d185d59e86fe2fc", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/microsoft/fara/HEAD/LICENSE"}], "id": "github:microsoft/fara", "identifiers": {}, "kind": "model", "last_verified_at": "2026-10-16T06:36:00Z", "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/microsoft/fara/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/microsoft/fara"}], "name": "Fara", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/microsoft/fara/HEAD/README.md", "target_id": "arxiv:2511.19663", "type": "documented_by"}, {"evidence_url": "https://raw.githubusercontent.com/microsoft/fara/HEAD/README.md", "target_id": "arxiv:2606.20785", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Microsoft browser-use agent models that act from screenshots by predicting click coordinates without accessibility trees; Fara1.5 comes in 4B, 9B and 27B, succeeding Fara-7B.", "topics": ["planning", "web-navigation"], "updated_at": null, "url": "https://github.com/microsoft/fara"}
{"aliases": [], "authors": [], "details": {}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T17:18:21Z", "role": "primary", "sha256": "8aaf8feff6e1d5a533c01afd8afc0cfc0700f0c1c73ba47949b56c11fb8ab5be", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/microsoft/Magma/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-17T17:18:56Z", "role": "primary", "sha256": "c2cfccb812fe482101a8f04597dfc5a9991a6b2748266c47ac91b6a5aae15383", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/microsoft/Magma/HEAD/LICENSE"}], "id": "github:microsoft/magma", "identifiers": {}, "kind": "model", "last_verified_at": "2026-10-17T17:18:21Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "MIT", "retrieved_at": "2026-10-17T17:18:56Z", "url": "https://raw.githubusercontent.com/microsoft/Magma/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/microsoft/Magma"}], "name": "Magma", "notes": "", "published_at": null, "relations": [], "review_status": "curated", "scope": "core", "summary": "Microsoft foundation model for multimodal agents, pretrained with Set-of-Mark and Trace-of-Mark objectives on UI, robotics and video data and fine-tunable for web and mobile UI navigation.", "topics": ["grounding", "planning", "mobile-navigation"], "updated_at": null, "url": "https://github.com/microsoft/Magma"}
{"aliases": [], "authors": [], "details": {}, "environments": ["windows", "desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:02Z", "role": "primary", "sha256": "40d9a4010ab122ab50242a0cd7a5320e2fd16b3e53494d988eecdbd94fd8e6e8", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/microsoft/UFO/HEAD/README.md"}], "id": "github:microsoft/ufo", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-16T06:36:02Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/microsoft/UFO"}], "name": "UFO", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/microsoft/UFO/HEAD/README.md", "target_id": "arxiv:2402.07939", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Windows agent framework for application and desktop automation.", "topics": ["planning", "cross-app"], "updated_at": null, "url": "https://github.com/microsoft/UFO"}
{"aliases": [], "authors": [], "details": {"license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/MobileLLM/AgentProg/HEAD/LICENSE"], "reason": "Reviewed repository license declaration for code; third-party dependencies, data, weights and executable environment images are separate."}}, "repository_tree_sha": "ef5e2998ac2defa86e0dc4c13e7e00d595e1e7cf", "repository_url": "https://github.com/MobileLLM/AgentProg"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:27:13Z", "role": "primary", "sha256": "becc4961d93f0170fab29120afa4b09b2999f2cd0e74b13ccb7f28cb9623761b", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations"], "url": "https://raw.githubusercontent.com/MobileLLM/AgentProg/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T15:37:01Z", "role": "primary", "sha256": "d14eecf7c041933cefecc03b62bbea45c1617a183fd8c4ec1334e22a644a7f67", "supports": ["identity"], "url": "https://api.github.com/repos/MobileLLM/AgentProg"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:25:22Z", "role": "primary", "sha256": "dd1cf9cd0aed31af63d7046ffa98d1be2c36258706951173ad8c28a64fcf9f7a", "supports": ["identity", "release_status", "details.repository_tree_sha"], "url": "https://api.github.com/repos/MobileLLM/AgentProg/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:35:19Z", "role": "primary", "sha256": "751123fccb2fd04b5c08b8c712a68a8bc69f56391593711efbfa533eaebbdb10", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/MobileLLM/AgentProg/HEAD/LICENSE"}], "id": "github:mobilellm/agentprog", "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": "Apache-2.0", "retrieved_at": "2026-10-18T04:35:19Z", "url": "https://raw.githubusercontent.com/MobileLLM/AgentProg/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/MobileLLM/AgentProg"}], "name": "AgentProg", "notes": "Public tree contains the agentprog package, CLI, program/context-management code and AndroidWorld evaluation integration. README announces AW-Extend, but this record is the framework rather than a separately audited benchmark release. External model services and Android device/emulator setup are required. Apache-2.0 covers code, not external model weights, apps or VM images.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/MobileLLM/AgentProg/HEAD/README.md", "target_id": "arxiv:2512.10371", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Android GUI-agent framework that keeps long-running task context in structured programs, variables and control flow, with a CLI and AndroidWorld integration.", "topics": ["long-horizon", "memory", "planning", "state-tracking"], "updated_at": null, "url": "https://github.com/MobileLLM/AgentProg"}
{"aliases": [], "authors": [], "details": {}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:03:20Z", "role": "primary", "sha256": "2fb7c0d70ccfa93605d5250a8f6d5182324f0c53d70e1c35ba18cb00f4ada548", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/night-chen/WorkForceAgent-R1/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:03:19Z", "role": "primary", "sha256": "0ad3ef1d900200e3ecab86cd44f79038000e8748fb285cdd5f32882455d2e36e", "supports": ["identity"], "url": "https://api.github.com/repos/night-chen/WorkForceAgent-R1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:05:39Z", "role": "primary", "sha256": "521d065188a6354416c1aafe88a0eb34486f2c2ec011c1cdf8ed008760dd9c55", "supports": ["release_status"], "url": "https://api.github.com/repos/night-chen/WorkForceAgent-R1/git/trees/main?recursive=1"}], "id": "github:night-chen/workforceagent-r1", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-17T02:03:20Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/night-chen/WorkForceAgent-R1"}], "name": "WorkForceAgent-R1", "notes": "The release contains WorkArena preprocessing and training source. The inspected root has no explicit project license; dependencies retain their own terms. Although the README discusses trained 7B/14B models, it does not provide a verified checkpoint release, so this is cataloged only as a framework.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/night-chen/WorkForceAgent-R1/HEAD/README.md", "target_id": "arxiv:2505.22942", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Web-task reward functions, WorkArena data processing and GRPO training configurations for reasoning-oriented language agents.", "topics": ["reinforcement-learning", "web-navigation", "planning"], "updated_at": null, "url": "https://github.com/night-chen/WorkForceAgent-R1"}
{"aliases": [], "authors": [], "details": {}, "environments": ["desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:02Z", "role": "primary", "sha256": "2248b2048b9166ce0348b24e81b2587fdf373a958fd86ab870ba85825425b02f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/niuzaisheng/ScreenAgent/HEAD/README.md"}], "id": "github:niuzaisheng/screenagent", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-16T06:36:02Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/niuzaisheng/ScreenAgent"}], "name": "ScreenAgent", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/niuzaisheng/ScreenAgent/HEAD/README.md", "target_id": "arxiv:2402.07945", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Environment and agent in which a vision-language model controls real desktops over VNC with mouse and keyboard in plan, act and reflect stages, plus a hand-annotated dataset of daily computer tasks.", "topics": ["planning"], "updated_at": null, "url": "https://github.com/niuzaisheng/ScreenAgent"}
{"aliases": [], "authors": [], "details": {"availability": "Public Android executor, agents, prompts and setup documentation.", "license_review": {"code": {"evidence_urls": ["https://api.github.com/repos/OpenDFM/MobA/git/trees/main?recursive=1"], "reason": "No repository source-license declaration established from the current public tree."}}, "repository_revision": "05cd237d80fcd23ad0dd6096cdb144117ecb3471"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:30:53Z", "role": "primary", "sha256": "49c096e5ee6a87205497d693b5248817ea98cad0f7e5254793660007ef4292f7", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations"], "url": "https://raw.githubusercontent.com/OpenDFM/MobA/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:30:53Z", "role": "primary", "sha256": "0ecd1833a1b6840441769e1cc26a55268f2de7beb61bc3e783a851dd7cff45f1", "supports": ["identity"], "url": "https://api.github.com/repos/OpenDFM/MobA"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:32:04Z", "role": "primary", "sha256": "302f324fce2884c5948cf62c75f2f839cd40c301faff6ccf4b6eb5dd29f953a3", "supports": ["release_status", "details.repository_revision", "licenses.code"], "url": "https://api.github.com/repos/OpenDFM/MobA/git/trees/main?recursive=1"}], "id": "github:opendfm/moba", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-17T12:30:53Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/OpenDFM/MobA"}], "name": "MobA", "notes": "The repository releases a configurable Python executor and Android ADB integration; its setup documents a Windows host, but the controlled interface is Android. The source is marked under continuing development and requires external model credentials. The inspected repository has no license file; the separately hosted MobBench data declaration is not transferred to this code.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/OpenDFM/MobA/HEAD/README.md", "target_id": "arxiv:2410.13757", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "An Android agent that combines global task planning and memory with a local action executor, using reflection to revise mobile interaction steps.", "topics": ["mobile-navigation", "memory", "planning", "self-reflection"], "updated_at": null, "url": "https://github.com/OpenDFM/MobA"}
{"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": ["desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:02Z", "role": "primary", "sha256": "b840368441695a54137216c2ba6bc99a8d35df75ad4196a4f0f0a1f5b5a70a9d", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/OS-Copilot/OS-Copilot/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T06:42:04Z", "role": "primary", "sha256": "3704700ddeaef20ec20cad39ed105ea518b316bc80c04dbcf2e69c3dc768a34f", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/OS-Copilot/OS-Copilot/HEAD/LICENSE"}], "id": "github:os-copilot/os-copilot", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-16T06:36:02Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "MIT", "retrieved_at": "2026-10-16T06:42:04Z", "url": "https://raw.githubusercontent.com/OS-Copilot/OS-Copilot/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/OS-Copilot/OS-Copilot"}], "name": "OS-Copilot", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/OS-Copilot/OS-Copilot/HEAD/README.md", "target_id": "arxiv:2402.07456", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Open-source library for generalist agents acting on an operating system's web, terminals, files and third-party apps, with the FRIDAY assistant, which takes custom tools and self-learns Excel tasks.", "topics": ["planning", "tool-integration"], "updated_at": null, "url": "https://github.com/OS-Copilot/OS-Copilot"}
{"aliases": [], "authors": [], "details": {}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T10:43:11Z", "role": "primary", "sha256": "6602cb260b52ae5b2b93db231a51a38d0463c78d8e71321620e05b4fa31caacb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/OSU-NLP-Group/WebDreamer/HEAD/README.md"}], "id": "github:osu-nlp-group/webdreamer", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-16T10:43:11Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/OSU-NLP-Group/WebDreamer"}], "name": "WebDreamer", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/OSU-NLP-Group/WebDreamer/HEAD/README.md", "target_id": "arxiv:2411.06559", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "World-model and planning components that simulate web-action outcomes, with linked training data and model checkpoints.", "topics": ["world-models", "planning"], "updated_at": null, "url": "https://github.com/OSU-NLP-Group/WebDreamer"}
{"aliases": [], "authors": [], "details": {}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T13:14:29Z", "role": "primary", "sha256": "75f99d9388aee3e0b124ebdb5f33e08fbe8ec999ce1e9eba5cd71c21e60522e0", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/PathOnAI/LiteWebAgent/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T13:29:25Z", "role": "primary", "sha256": "ec6ef142e3900e9f5555009b825387e41da67a77a24ee72ef6963c99b065e6eb", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/PathOnAIOrg/LiteWebAgent/HEAD/LICENSE"}], "id": "github:pathonaiorg/litewebagent", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-16T13:14:29Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "Apache-2.0", "retrieved_at": "2026-10-16T13:29:25Z", "url": "https://raw.githubusercontent.com/PathOnAIOrg/LiteWebAgent/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/PathOnAIOrg/LiteWebAgent"}], "name": "LiteWebAgent", "notes": "The repository is maintained by the PathOnAI.org research community; its README explicitly disclaims affiliation with a for-profit company.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/PathOnAI/LiteWebAgent/HEAD/README.md", "target_id": "arxiv:2503.02950", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Modular web-agent framework with browser control, planning, workflow memory, and tree-search components, plus web and extension interfaces.", "topics": ["web-navigation", "planning", "memory", "tool-integration"], "updated_at": null, "url": "https://github.com/PathOnAIOrg/LiteWebAgent"}
{"aliases": [], "authors": [], "details": {"availability": "Public agent and browser-environment source; legacy API compatibility is unverified.", "license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/posgnu/rci-agent/HEAD/LICENSE"], "reason": "Repository code license; not a license for the remote language-model service."}}, "repository_revision": "31bf737922fa0337d03367a844de5cf1aff4ac52"}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:14:12Z", "role": "primary", "sha256": "6a101c867546365d21b48527781bd3af324075929067fb465c6105872070f692", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations", "details"], "url": "https://raw.githubusercontent.com/posgnu/rci-agent/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:14:11Z", "role": "primary", "sha256": "b826fc1747a1bbe38a8df8da567a68f79936509c5325ec76a683706118c526f5", "supports": ["identity"], "url": "https://api.github.com/repos/posgnu/rci-agent"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:14:11Z", "role": "primary", "sha256": "3dcb8cc06679490315f0f3636bb2795e2f717b77d4156ad0bf2429cc36e3273c", "supports": ["release_status"], "url": "https://api.github.com/repos/posgnu/rci-agent/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:14:12Z", "role": "primary", "sha256": "ffa8b2a603abd9ede8eb686e87c4a433110f1be47d7cc98df4977b7fb2ee72fc", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/posgnu/rci-agent/HEAD/LICENSE"}], "id": "github:posgnu/rci-agent", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-17T09:14:12Z", "licenses": {"code": {"basis": "Repository code license; not a license for the remote language-model service.", "declared": "MIT", "retrieved_at": "2026-10-17T09:14:12Z", "url": "https://raw.githubusercontent.com/posgnu/rci-agent/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/posgnu/rci-agent"}], "name": "RCI Agent", "notes": "The public release includes the agent, prompting utilities and a bundled MiniWoB++ environment wrapper. Its README documents older hosted model names and an OpenAI-key setup; contemporary endpoint compatibility was not tested. The repository notes that its MiniWoB++ dependency is not officially supported on Windows. MIT applies to repository code; no new trained weights are claimed.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/posgnu/rci-agent/HEAD/README.md", "target_id": "arxiv:2303.17491", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "A MiniWoB++ agent that uses explicit and implicit recursive criticism-and-improvement prompts to refine action plans and ground actions to browser state.", "topics": ["web-navigation", "planning", "self-reflection", "training-free"], "updated_at": null, "url": "https://github.com/posgnu/rci-agent"}
{"aliases": [], "authors": [], "details": {}, "environments": ["cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:00Z", "role": "primary", "sha256": "5fd1399c88bdf406e7689765ccf0599b3b7e0ddbed082fa92bffdd7b120bd41a", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/Reallm-Labs/InfiGUI-R1/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/Reallm-Labs/InfiGUI-R1/HEAD/LICENSE"}], "id": "github:reallm-labs/infigui-r1", "identifiers": {}, "kind": "model", "last_verified_at": "2026-10-16T06:36:00Z", "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/Reallm-Labs/InfiGUI-R1/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/Reallm-Labs/InfiGUI-R1"}], "name": "InfiGUI-R1", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/Reallm-Labs/InfiGUI-R1/HEAD/README.md", "target_id": "arxiv:2504.14239", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "3B GUI agent trained in two stages: distilling spatial reasoning from teacher models, then reinforcement learning that rewards accurate sub-goals and trains error recovery.", "topics": ["reinforcement-learning", "planning"], "updated_at": null, "url": "https://github.com/Reallm-Labs/InfiGUI-R1"}
{"aliases": [], "authors": [], "details": {}, "environments": ["desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:02Z", "role": "primary", "sha256": "9c43f97560c3781b28f97c894ff65bbea7121c3df8c183ddf15bb38842bd1e39", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/showlab/assistgui/HEAD/README.md"}], "id": "github:showlab/assistgui", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-16T06:36:02Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/showlab/assistgui"}], "name": "AssistGUI", "notes": "", "published_at": null, "relations": [], "review_status": "curated", "scope": "core", "summary": "LLM-based assistant for automating work in desktop productivity software by watching the screen and issuing mouse and keyboard commands; its repository says the code will be released later.", "topics": ["planning"], "updated_at": null, "url": "https://github.com/showlab/assistgui"}
{"aliases": [], "authors": [], "details": {}, "environments": ["desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:02Z", "role": "primary", "sha256": "9ea381fdb78eb326d988ea384b6f5074ef0ff4c0d5fab555a148eded385e88f1", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/simular-ai/Agent-S/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T06:42:03Z", "role": "primary", "sha256": "43070e2d4e532684de521b885f385d0841030efa2b1a20bafb76133a5e1379c1", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/simular-ai/Agent-S/HEAD/LICENSE"}], "id": "github:simular-ai/agent-s", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-16T06:36:02Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "Apache-2.0", "retrieved_at": "2026-10-16T06:42:03Z", "url": "https://raw.githubusercontent.com/simular-ai/Agent-S/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/simular-ai/Agent-S"}], "name": "Agent S", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/simular-ai/Agent-S/HEAD/README.md", "target_id": "arxiv:2410.08164", "type": "documented_by"}, {"evidence_url": "https://raw.githubusercontent.com/simular-ai/Agent-S/HEAD/README.md", "target_id": "arxiv:2504.00906", "type": "documented_by"}, {"evidence_url": "https://raw.githubusercontent.com/simular-ai/Agent-S/HEAD/README.md", "target_id": "organization:simular", "type": "published_by"}], "review_status": "curated", "scope": "core", "summary": "Simular's open-source framework for computer-use agents that click, type and scroll through desktop and web apps on macOS, Windows and Linux; its third generation, Agent S3, reports 72.6% on OSWorld.", "topics": ["planning", "memory"], "updated_at": null, "url": "https://github.com/simular-ai/Agent-S"}
{"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/sqzhang-lazy/D-PoT/HEAD/README.md"], "reason": "No project-wide code license is established by the inspected README/API/tree; third-party code presence does not establish a license for this package."}}, "repository_tree_sha": "fc764f19802f4f59afa635bfeec5a50e4c3d5465", "repository_url": "https://github.com/sqzhang-lazy/D-PoT"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:27:14Z", "role": "primary", "sha256": "c9c21102db512270079666531dbab71c1ab2ec2073add16be56f8ab3c6fe8abf", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "licenses.code"], "url": "https://raw.githubusercontent.com/sqzhang-lazy/D-PoT/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T15:37:02Z", "role": "primary", "sha256": "d62cfecbfccd120e53a7fc60e8d4b90b157e998cc5ddac467514c30c0cae1363", "supports": ["identity"], "url": "https://api.github.com/repos/sqzhang-lazy/D-PoT"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:25:24Z", "role": "primary", "sha256": "f75444a0252df5d6ec1ab70d805584cb490cdc38efa995aae55ffbde43ab5fb1", "supports": ["identity", "release_status", "details.repository_tree_sha"], "url": "https://api.github.com/repos/sqzhang-lazy/D-PoT/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:35Z", "role": "primary", "sha256": "b3e67984f59a1b05ccd946eb54e2d37b0bb78ab0ca796fbb6d544793dafea3e0", "supports": ["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": "github:sqzhang-lazy/d-pot", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-18T04:27:14Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/sqzhang-lazy/D-PoT"}], "name": "D-PoT evaluation code", "notes": "README describes AITW test sampling and preprocessing reused from Auto-GUI, followed by prediction/action-extraction/evaluation scripts. This is not claimed as a ready-to-run interactive phone controller. No project-level license declaration appears in the inspected README/API/non-truncated tree; code, dataset and model rights remain unestablished here.", "published_at": null, "relations": [{"evidence_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", "target_id": "arxiv:2410.00467", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Offline prompting and evaluation code for Dynamic Planning of Thoughts on Android-in-the-Wild, including action extraction and comparison with reference actions.", "topics": ["planning", "action-prediction", "evaluation"], "updated_at": null, "url": "https://github.com/sqzhang-lazy/D-PoT"}
{"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": ["android", "cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:02Z", "role": "primary", "sha256": "ca402ffd3b05f72a6d72f47de52c962f92e410749a77939044f3301d4330af16", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T06:42:03Z", "role": "primary", "sha256": "961819a89dacd1f4f161e1d2ebf616b538bf095a2eecfa3ef40b7ccec633fac4", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/LICENSE"}], "id": "github:x-plug/mobileagent", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-16T06:36:02Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "MIT", "retrieved_at": "2026-10-16T06:42:03Z", "url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/X-PLUG/MobileAgent"}], "name": "Mobile-Agent", "notes": "This repository contains multiple Mobile-Agent generations. Platform coverage differs by version; use the corresponding paper and implementation instructions.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/README.md", "target_id": "arxiv:2401.16158", "type": "documented_by"}, {"evidence_url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/README.md", "target_id": "arxiv:2406.01014", "type": "documented_by"}, {"evidence_url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/README.md", "target_id": "arxiv:2508.15144", "type": "documented_by"}, {"evidence_url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/README.md", "target_id": "arxiv:2602.16855", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Mobile device agent family using visual interface interaction.", "topics": ["mobile-navigation", "planning"], "updated_at": null, "url": "https://github.com/X-PLUG/MobileAgent"}
{"aliases": [], "authors": [], "details": {}, "environments": ["cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:00Z", "role": "primary", "sha256": "2dd9f36b818c1b7b6bd772ba8575f6c0ebab6d9e0d307edcc2f196f3e4ba502e", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/xlang-ai/aguvis/HEAD/README.md"}], "id": "github:xlang-ai/aguvis", "identifiers": {}, "kind": "model", "last_verified_at": "2026-10-16T06:36:00Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/xlang-ai/aguvis"}], "name": "Aguvis", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/xlang-ai/aguvis/HEAD/README.md", "target_id": "arxiv:2412.04454", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Vision-based GUI agent models with unified action representations.", "topics": ["grounding", "planning"], "updated_at": null, "url": "https://github.com/xlang-ai/aguvis"}
{"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": {"platform_constraints": {"harmonyos": "The README limits device debugging support to HarmonyOS 4 or earlier."}}, "environments": ["android", "harmonyos"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T13:14:29Z", "role": "primary", "sha256": "02e231c68ffc37c829789eabc508a220f3d99a33ee8bf676e53c999a0ff54021", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://raw.githubusercontent.com/Yuqi-Zhou/CHOP/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T13:29:25Z", "role": "primary", "sha256": "1a9f9af82764fdfc07dd8c2ec02e66cbb5c7c2cfc899b0e7b53b26e2a774ddf9", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/Yuqi-Zhou/CHOP/HEAD/LICENSE"}], "id": "github:yuqi-zhou/chop", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-16T13:14:29Z", "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/Yuqi-Zhou/CHOP/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/Yuqi-Zhou/CHOP"}], "name": "CHOP", "notes": "The README says the complete CHOP-ZH dataset remains unreleased. Android and HarmonyOS compatibility is publisher-documented and has not been runtime-tested here.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/Yuqi-Zhou/CHOP/HEAD/README.md", "target_id": "arxiv:2503.03743", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Mobile automation framework that uses reusable basis subtasks to constrain and refine planning between user goals and executable actions.", "topics": ["mobile-navigation", "planning", "workflow"], "updated_at": null, "url": "https://github.com/Yuqi-Zhou/CHOP"}
{"aliases": [], "authors": [], "details": {"availability": "Public preprocessing/refinement code, offline action-matching evaluation, and multipart annotation archives; original images and a fine-tuned model are separate.", "license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/ZrW00/GUIPivot/b02e8efdbe02ab82169a2c54d724e33b4037330e/LICENSE"], "reason": "Repository GNU GPL version 3 license for code; no data or pretrained-weight license inferred."}}, "repository_revision": "b02e8efdbe02ab82169a2c54d724e33b4037330e"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-18T12:02:55.267942+00:00", "role": "primary", "sha256": "77384ff1dfceac74f2e60af56b016a815963481f9e41a29915698777f55e828e", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations"], "url": "https://raw.githubusercontent.com/ZrW00/GUIPivot/b02e8efdbe02ab82169a2c54d724e33b4037330e/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T11:25:27+00:00", "role": "primary", "sha256": "3d57dc43db78620832e41416e4bfb7ae445afe782e5538d85a9784a8f1e5e4eb", "supports": ["identity"], "url": "https://api.github.com/repos/ZrW00/GUIPivot"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T12:02:54.310430+00:00", "role": "primary", "sha256": "a4d3ddbb2b2f4a936c9bc1a1c388b9870dbc5d2c61ff6841a1082b884f5c54e9", "supports": ["release_status", "details.repository_revision", "details.availability", "notes"], "url": "https://api.github.com/repos/ZrW00/GUIPivot/git/trees/b02e8efdbe02ab82169a2c54d724e33b4037330e?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T12:03:28.611747+00:00", "role": "primary", "sha256": "3972dc9744f6499f0f9b2dbf76696f2ae7ad8af9b23dde66d6af86c9dfb36986", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/ZrW00/GUIPivot/b02e8efdbe02ab82169a2c54d724e33b4037330e/LICENSE"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T12:02:36.501726+00:00", "role": "primary", "sha256": "25e5282e7d85bca133fc77e294cd95d90e4ae65b0d8796cd7c49f53a633b4893", "supports": ["summary", "environments", "notes"], "url": "https://arxiv.org/html/2503.00401v2"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T12:03:30.911159+00:00", "role": "primary", "sha256": "9b5b196e065be2ebc05169a68710b8a6e1bd3b6eb663c68bb449234dbb7a45af", "supports": ["notes", "details.availability"], "url": "https://raw.githubusercontent.com/ZrW00/GUIPivot/b02e8efdbe02ab82169a2c54d724e33b4037330e/data/dataset_info.json"}], "id": "github:zrw00/guipivot", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-18T12:02:55.267942+00:00", "licenses": {"code": {"basis": "Repository GNU GPL version 3 license for code; no data or pretrained-weight license inferred.", "declared": "GPL-3.0", "retrieved_at": "2026-10-18T12:03:28.611747+00:00", "url": "https://raw.githubusercontent.com/ZrW00/GUIPivot/b02e8efdbe02ab82169a2c54d724e33b4037330e/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/ZrW00/GUIPivot"}], "name": "GUIPivot", "notes": "The repository releases preprocessing/refinement/evaluation code and multipart ShareGPT-format JSON annotations. Evaluation matches recorded single-step action predictions on AndroidControl and AITZ; it does not run live device tasks. The README links a UIBERT training-JSON path that is not a standalone file in this revision; annotations are packaged in data/GUI.z01, data/GUI.z02 and data/GUI.zip. Original UIBERT/OS-Atlas, AITZ and AndroidControl images must be obtained separately. This entry does not claim a released trained checkpoint or transfer the GPL-3.0 code license to third-party datasets and images.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/ZrW00/GUIPivot/b02e8efdbe02ab82169a2c54d724e33b4037330e/README.md", "target_id": "arxiv:2503.00401", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Query-inference training tools connecting screenshot grounding with GUI action reasoning, including preprocessing, query refinement and offline single-step evaluation on AndroidControl and AITZ.", "topics": ["grounding", "planning", "imitation-learning"], "updated_at": null, "url": "https://github.com/ZrW00/GUIPivot"}
{"aliases": [], "authors": [], "details": {"disabled": false, "download_url": "https://huggingface.co/Hcompany/Holo2-4B/tree/main", "gated": false, "hub_id": "Hcompany/Holo2-4B"}, "environments": ["web", "desktop", "cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:07:19Z", "role": "primary", "sha256": "32232445026dc93113beb6b23d3e43da44d6288a67501f8f0023a78a43efb100", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/Hcompany/Holo2-4B/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:07:18Z", "role": "primary", "sha256": "bb870c78f3075002ae42a7d1f392fda048fa31a58c865290b2a9b4a1137c6dd3", "supports": ["identity", "release_status", "details", "licenses.weights"], "url": "https://huggingface.co/api/models/Hcompany/Holo2-4B"}, {"method": "huggingface_card", "retrieved_at": "2026-10-17T04:07:19Z", "role": "primary", "sha256": "32232445026dc93113beb6b23d3e43da44d6288a67501f8f0023a78a43efb100", "supports": ["licenses.weights"], "url": "https://huggingface.co/Hcompany/Holo2-4B/raw/main/README.md"}], "id": "url:16a51d9e725ad5a5c629", "identifiers": {}, "kind": "model", "last_verified_at": "2026-10-17T04:07:19Z", "licenses": {"code": null, "data": null, "document": null, "weights": {"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/Hcompany/Holo2-4B/raw/main/README.md"}}, "links": [{"rel": "primary", "url": "https://huggingface.co/Hcompany/Holo2-4B"}], "name": "Holo2 4B", "notes": "Two public safetensors shards are present and this 4B model card declares Apache-2.0. The same card assigns different terms to the 30B family member, so the 4B declaration is not generalized to the family. Scores and proprietary training data are not treated as independently verified or publicly released.", "published_at": null, "relations": [], "review_status": "curated", "scope": "core", "summary": "A Qwen3-VL-based checkpoint for navigation and multi-step GUI actions across web, desktop and mobile interfaces.", "topics": ["grounding", "planning", "action-prediction"], "updated_at": null, "url": "https://huggingface.co/Hcompany/Holo2-4B"}
{"aliases": [], "authors": [], "details": {"artifact_type": "model-checkpoint", "disabled": false, "download_url": "https://huggingface.co/meituan/EvoCUA-8B-20260105/tree/main", "gated": false, "hub_id": "meituan/EvoCUA-8B-20260105", "languages": ["en", "zh"], "modalities": [], "size_category_reported": null}, "environments": ["desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:50:49Z", "role": "primary", "sha256": "3fd1e50729e97746fec57c7d5da6f093d76fb9174e74c3f2c630fed5486fe995", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/api/models/meituan/EvoCUA-8B-20260105"}, {"method": "huggingface_api", "retrieved_at": "2026-10-16T06:50:49Z", "role": "primary", "sha256": "3fd1e50729e97746fec57c7d5da6f093d76fb9174e74c3f2c630fed5486fe995", "supports": ["details", "licenses.weights"], "url": "https://huggingface.co/api/models/meituan/EvoCUA-8B-20260105"}], "id": "url:2a2682f86dcb8ebc9a80", "identifiers": {}, "kind": "model", "last_verified_at": "2026-10-16T06:50:49Z", "licenses": {"code": null, "data": null, "document": null, "weights": {"basis": "publisher model/dataset card", "declared": "apache-2.0", "retrieved_at": "2026-10-16T06:50:49Z", "url": "https://huggingface.co/meituan/EvoCUA-8B-20260105"}}, "links": [{"rel": "primary", "url": "https://huggingface.co/meituan/EvoCUA-8B-20260105"}], "name": "EvoCUA 8B 20260105", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/api/models/meituan/EvoCUA-8B-20260105", "target_id": "github:meituan/evocua", "type": "variant_of"}], "review_status": "curated", "scope": "core", "summary": "EvoCUA computer-use agent checkpoint (20260105 release) with about 8.8B BF16 parameters in the Qwen3-VL architecture, tagged for OSWorld and for English and Chinese.", "topics": ["reinforcement-learning", "planning"], "updated_at": null, "url": "https://huggingface.co/meituan/EvoCUA-8B-20260105"}
{"aliases": [], "authors": [], "details": {"disabled": false, "download_url": "https://huggingface.co/xTRam1/plan-and-act-planner-70b/tree/main", "gated": false, "hub_id": "xTRam1/plan-and-act-planner-70b", "license_review": {"weights": {"evidence_urls": ["https://huggingface.co/xTRam1/plan-and-act-planner-70b/raw/main/README.md"], "reason": "The model card declares MIT in its metadata header (license: mit), beside a template comment asking the publisher to update it if needed. It is recorded as the publisher's declaration and does not override the Llama-3.3-70B-Instruct base model terms."}}}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:28Z", "role": "primary", "sha256": "4ef2d84d2b5d9d2972907e6135ea49f1ef088beaf6ef6042dc7d8fb4297a9a3a", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details", "licenses.weights"], "url": "https://huggingface.co/xTRam1/plan-and-act-planner-70b/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:28Z", "role": "primary", "sha256": "d2f757790ded81244e7c9b9e7350ec9fdc304609e387a95c7918db705bda535c", "supports": ["identity", "release_status", "details", "licenses.weights"], "url": "https://huggingface.co/api/models/xTRam1/plan-and-act-planner-70b"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:02:06Z", "role": "primary", "sha256": "feee31aa4d0801cb364578099a13e4f9cdcaa4eb5fbf4312b3fc3f8edd479a14", "supports": ["environments"], "url": "https://raw.githubusercontent.com/SqueezeAILab/plan-and-act/HEAD/README.md"}], "id": "url:3932d69c6b6f9b2a7ba8", "identifiers": {}, "kind": "model", "last_verified_at": "2026-10-17T02:02:28Z", "licenses": {"code": null, "data": null, "document": null, "weights": {"basis": "publisher model/dataset card metadata; third-party assets retain their own terms", "declared": "mit", "retrieved_at": "2026-10-17T02:02:28Z", "url": "https://huggingface.co/xTRam1/plan-and-act-planner-70b/raw/main/README.md"}}, "links": [{"rel": "primary", "url": "https://huggingface.co/xTRam1/plan-and-act-planner-70b"}], "name": "Plan-and-Act Planner 70B", "notes": "The Hub manifest contains 17 safetensors shards and tokenizer/configuration files. Card metadata declares MIT and Llama-3.3-70B-Instruct as the base, but retains template comments asking the publisher to update those fields if needed. These are recorded as publisher declarations, without claiming they override upstream model terms.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/xTRam1/plan-and-act-planner-70b/raw/main/README.md", "target_id": "arxiv:2503.09572", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "The planner component of Plan-and-Act, generating structured high-level plans that an executor turns into web actions.", "topics": ["planning", "web-navigation", "long-horizon"], "updated_at": null, "url": "https://huggingface.co/xTRam1/plan-and-act-planner-70b"}
{"aliases": [], "authors": [], "details": {"availability": "Ungated mixture/post-training JSON and image-archive manifests; some per-domain images and training code remain pending.", "dataset_profile": {"counts": [], "evidence_urls": ["https://huggingface.co/datasets/hkust-nlp/GUIMid/raw/main/README.md", "https://raw.githubusercontent.com/hkust-nlp/GUIMid/HEAD/README.md"], "formats": ["JSON", "tar.gz image archives"], "modalities": ["text", "images"], "notes": "Do not treat Hub size bands or the inconsistent introductory task count as a sample total. Some constituent-domain image packages remain forthcoming.", "splits": ["train configurations for named mid-training domains"], "supervision": ["mid-training instruction supervision", "GUI post-training trajectories"]}, "gated": false, "hub_id": "hkust-nlp/GUIMid", "license_review": {"data": {"evidence_urls": ["https://huggingface.co/datasets/hkust-nlp/GUIMid/raw/main/README.md"], "reason": "Publisher dataset-card declaration for the released packaging; original source datasets retain their own terms."}}, "release_revision": "b6a360815c843b5733674191b6e6d7fa68b2ea8c"}, "environments": ["web", "android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:23:07Z", "role": "primary", "sha256": "1cdf702c819ecf370f4191c6c71fae7b579cfbaf766908b42156964a69eddacc", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations", "details.dataset_profile", "licenses.data"], "url": "https://huggingface.co/datasets/hkust-nlp/GUIMid/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:23:07Z", "role": "primary", "sha256": "ededa2b3c8f21af8188152b028758c800db76cdbb8461ba82b8574cab375914e", "supports": ["identity", "release_status", "details.hub_id", "details.gated", "details.release_revision"], "url": "https://huggingface.co/api/datasets/hkust-nlp/GUIMid"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:18:19Z", "role": "primary", "sha256": "31b836da66c6907ff9a75ebc7f06da0db67e7d9fe9ca92d0c770f5f15f90a39e", "supports": ["identity"], "url": "https://api.github.com/repos/hkust-nlp/GUIMid"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:18:19Z", "role": "primary", "sha256": "d9f63a03088c37f7145725c7a182100d1593fceadd26a9bf817d434ca8d6297e", "supports": ["notes", "relations"], "url": "https://raw.githubusercontent.com/hkust-nlp/GUIMid/HEAD/README.md"}], "id": "url:422010da7a5c845a18f6", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-17T11:23:07Z", "licenses": {"code": null, "data": {"basis": "Publisher dataset-card declaration for the released packaging; original source datasets retain their own terms.", "declared": "mit", "retrieved_at": "2026-10-17T11:23:07Z", "url": "https://huggingface.co/datasets/hkust-nlp/GUIMid/raw/main/README.md"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/hkust-nlp/GUIMid"}], "name": "GUIMid training data", "notes": "The public manifest includes GUIMid.json, mavis.tar.gz, GUI_trajectory.json, traj.tar.gz and domain-specific JSON files. The card says image data for some individual domains is still forthcoming and its overview inconsistently says nine tasks while later enumerating seven vision-language plus four language-only domains; no aggregate task/sample count is inferred. Training code remains marked coming soon in the repository. MIT is the publisher dataset declaration, not a relicensing of all upstream content.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/hkust-nlp/GUIMid/raw/main/README.md", "target_id": "arxiv:2504.10127", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "A released GUI-agent mid-training mixture and supporting domain datasets, together with a post-training trajectory file and image archives for web and mobile interaction.", "topics": ["imitation-learning", "planning", "screen-understanding"], "updated_at": null, "url": "https://huggingface.co/datasets/hkust-nlp/GUIMid"}
{"aliases": ["OpenCUA 72B Preview", "OpenCUA-72B-preview", "url:8c792416d297293422cf", "xlangai/OpenCUA-72B-preview", "https://huggingface.co/xlangai/OpenCUA-72B-preview"], "authors": [], "details": {"availability": "Public ungated checkpoint with 82 safetensors shards and model index.", "base_model": "Qwen/Qwen2.5-VL-72B-Instruct", "disabled": false, "download_url": "https://huggingface.co/xlangai/OpenCUA-72B/tree/main", "gated": false, "hub_id": "xlangai/OpenCUA-72B", "license_review": {"weights": {"evidence_urls": ["https://huggingface.co/xlangai/OpenCUA-72B/raw/main/README.md"], "reason": "Publisher model card explicitly declares MIT for the released model; other code and training-data scopes are not generalized."}}, "release_revision": "3cc4af5337a1d0010f934284df3c482797266547"}, "environments": ["desktop", "linux", "windows", "macos"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:14:12Z", "role": "primary", "sha256": "ebe551af1008a2307d16cf68715fc1861f8c501c4981660dbe08f691cf73d787", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations", "details", "licenses.weights"], "url": "https://huggingface.co/xlangai/OpenCUA-72B/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:14:12Z", "role": "primary", "sha256": "ff0ee40df2f44a2b3aa1b8117c31752013a40fdb6a85d66cdc596e096583ac5d", "supports": ["identity", "release_status", "details"], "url": "https://huggingface.co/api/models/xlangai/OpenCUA-72B"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:15:35Z", "role": "primary", "sha256": "ff0ee40df2f44a2b3aa1b8117c31752013a40fdb6a85d66cdc596e096583ac5d", "supports": ["identity", "notes"], "url": "https://huggingface.co/api/models/xlangai/OpenCUA-72B-preview"}, {"method": "huggingface_repository_identity", "retrieved_at": "2026-10-17T19:15:04Z", "role": "primary", "sha256": "ff0ee40df2f44a2b3aa1b8117c31752013a40fdb6a85d66cdc596e096583ac5d", "supports": ["aliases"], "url": "https://huggingface.co/api/models/xlangai/OpenCUA-72B-preview"}, {"method": "huggingface_repository_identity", "retrieved_at": "2026-10-17T19:15:05Z", "role": "primary", "sha256": "ff0ee40df2f44a2b3aa1b8117c31752013a40fdb6a85d66cdc596e096583ac5d", "supports": ["aliases"], "url": "https://huggingface.co/api/models/xlangai/OpenCUA-72B"}], "id": "url:4a0b1692589594745a3c", "identifiers": {}, "kind": "model", "last_verified_at": "2026-10-17T09:14:12Z", "licenses": {"code": null, "data": null, "document": null, "weights": {"basis": "Publisher model card explicitly declares MIT for the released model; other code and training-data scopes are not generalized.", "declared": "mit", "retrieved_at": "2026-10-17T09:14:12Z", "url": "https://huggingface.co/xlangai/OpenCUA-72B/raw/main/README.md"}}, "links": [{"rel": "primary", "url": "https://huggingface.co/xlangai/OpenCUA-72B"}], "name": "OpenCUA 72B", "notes": "The former OpenCUA-72B-preview API redirects to the canonical xlangai/OpenCUA-72B repository. The public manifest lists 82 safetensors weight shards plus an index and custom model/tokenizer code. The card declares MIT for the released model and describes absolute coordinates after smart resizing. Weight files and remote code were not downloaded or executed.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/xlangai/OpenCUA-72B/raw/main/README.md", "target_id": "github:xlang-ai/opencua", "type": "variant_of"}, {"evidence_url": "https://huggingface.co/xlangai/OpenCUA-72B/raw/main/README.md", "target_id": "arxiv:2508.09123", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "A Qwen2.5-VL-based computer-use checkpoint that combines screenshot history and reflective reasoning to predict executable desktop actions.", "topics": ["grounding", "planning", "action-prediction", "self-reflection"], "updated_at": null, "url": "https://huggingface.co/xlangai/OpenCUA-72B"}
{"aliases": [], "authors": [], "details": {"artifact_type": "model-checkpoint", "disabled": false, "download_url": "https://huggingface.co/ByteDance-Seed/UI-TARS-1.5-7B/tree/main", "gated": false, "hub_id": "ByteDance-Seed/UI-TARS-1.5-7B", "languages": ["en"], "modalities": [], "size_category_reported": null}, "environments": ["cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:50:49Z", "role": "primary", "sha256": "73b8964fb8ae2c25009dd4621549a3549d357c4c5c766ef7ed441acaef7efbc9", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/api/models/ByteDance-Seed/UI-TARS-1.5-7B"}, {"method": "huggingface_api", "retrieved_at": "2026-10-16T06:50:49Z", "role": "primary", "sha256": "73b8964fb8ae2c25009dd4621549a3549d357c4c5c766ef7ed441acaef7efbc9", "supports": ["details", "licenses.weights"], "url": "https://huggingface.co/api/models/ByteDance-Seed/UI-TARS-1.5-7B"}], "id": "url:4d32ad3682ae2d71c178", "identifiers": {}, "kind": "model", "last_verified_at": "2026-10-16T06:50:49Z", "licenses": {"code": null, "data": null, "document": null, "weights": {"basis": "publisher model/dataset card", "declared": "apache-2.0", "retrieved_at": "2026-10-16T06:50:49Z", "url": "https://huggingface.co/ByteDance-Seed/UI-TARS-1.5-7B"}}, "links": [{"rel": "primary", "url": "https://huggingface.co/ByteDance-Seed/UI-TARS-1.5-7B"}], "name": "UI-TARS 1.5 7B", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/api/models/ByteDance-Seed/UI-TARS-1.5-7B", "target_id": "github:bytedance/ui-tars", "type": "variant_of"}], "review_status": "curated", "scope": "core", "summary": "UI-TARS 1.5 GUI-agent checkpoint in the Qwen2.5-VL architecture with about 8.3B parameters stored in FP32, released under Apache-2.0 for English image-text-to-text use in Transformers.", "topics": ["grounding", "planning"], "updated_at": null, "url": "https://huggingface.co/ByteDance-Seed/UI-TARS-1.5-7B"}
{"aliases": [], "authors": [], "details": {}, "environments": ["cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:38:27Z", "role": "primary", "sha256": "de16b4fd611982cfaa160cb0fbdc1b0f0559b90983e452964527704a6de4e72d", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://docs.anthropic.com/en/docs/agents-and-tools/tool-use/computer-use-tool"}], "id": "url:57f17b7d0c18b2d7c8bf", "identifiers": {}, "kind": "model", "last_verified_at": "2026-10-16T06:38:27Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://docs.anthropic.com/en/docs/agents-and-tools/tool-use/computer-use-tool"}], "name": "Claude Computer Use", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://docs.anthropic.com/en/docs/agents-and-tools/tool-use/computer-use-tool", "target_id": "organization:anthropic", "type": "published_by"}], "review_status": "curated", "scope": "core", "summary": "Claude tool capability for screenshot-based computer interaction.", "topics": ["planning", "grounding"], "updated_at": null, "url": "https://docs.anthropic.com/en/docs/agents-and-tools/tool-use/computer-use-tool"}
{"aliases": [], "authors": [], "details": {"artifact_type": "model-checkpoint", "disabled": false, "download_url": "https://huggingface.co/meituan/EvoCUA-32B-20260105/tree/main", "gated": false, "hub_id": "meituan/EvoCUA-32B-20260105", "languages": ["en", "zh"], "modalities": [], "size_category_reported": null}, "environments": ["desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:50:49Z", "role": "primary", "sha256": "f42d9537a05f2562b57e5d508f95c3d29a9107d06574318c5715c7da5fb2a86d", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/api/models/meituan/EvoCUA-32B-20260105"}, {"method": "huggingface_api", "retrieved_at": "2026-10-16T06:50:49Z", "role": "primary", "sha256": "f42d9537a05f2562b57e5d508f95c3d29a9107d06574318c5715c7da5fb2a86d", "supports": ["details", "licenses.weights"], "url": "https://huggingface.co/api/models/meituan/EvoCUA-32B-20260105"}], "id": "url:60a95f34072aeb2771cc", "identifiers": {}, "kind": "model", "last_verified_at": "2026-10-16T06:50:49Z", "licenses": {"code": null, "data": null, "document": null, "weights": {"basis": "publisher model/dataset card", "declared": "apache-2.0", "retrieved_at": "2026-10-16T06:50:49Z", "url": "https://huggingface.co/meituan/EvoCUA-32B-20260105"}}, "links": [{"rel": "primary", "url": "https://huggingface.co/meituan/EvoCUA-32B-20260105"}], "name": "EvoCUA 32B 20260105", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/api/models/meituan/EvoCUA-32B-20260105", "target_id": "github:meituan/evocua", "type": "variant_of"}], "review_status": "curated", "scope": "core", "summary": "Larger EvoCUA computer-use agent checkpoint (20260105 release) with about 33B BF16 parameters in the Qwen3-VL architecture, tagged for OSWorld and for English and Chinese.", "topics": ["reinforcement-learning", "planning"], "updated_at": null, "url": "https://huggingface.co/meituan/EvoCUA-32B-20260105"}
{"aliases": ["url:5f4cb613bda51b45d9c7", "inclusionAI/UI-Venus-2-9B", "https://huggingface.co/inclusionAI/UI-Venus-2-9B"], "authors": [], "details": {"artifact_type": "model-checkpoint", "disabled": false, "download_url": "https://huggingface.co/inclusionAI/UI-Venus-2-9b/tree/main", "gated": false, "hub_id": "inclusionAI/UI-Venus-2-9B", "languages": [], "modalities": [], "size_category_reported": null}, "environments": ["cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:50:49Z", "role": "primary", "sha256": "0097be2c0e5c3f3c8aace10de835b0d7c1075531cc7823287c3f5488aebaea08", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/api/models/inclusionAI/UI-Venus-2-9b"}, {"method": "huggingface_api", "retrieved_at": "2026-10-16T06:50:49Z", "role": "primary", "sha256": "0097be2c0e5c3f3c8aace10de835b0d7c1075531cc7823287c3f5488aebaea08", "supports": ["details", "licenses.weights"], "url": "https://huggingface.co/api/models/inclusionAI/UI-Venus-2-9b"}, {"method": "huggingface_repository_identity", "retrieved_at": "2026-10-17T19:15:02Z", "role": "primary", "sha256": "3ec7929329e8e1b09f31dee4a3e67128d68cddc95220a5944f4cd6102dae327e", "supports": ["aliases"], "url": "https://huggingface.co/api/models/inclusionAI/UI-Venus-2-9B"}, {"method": "huggingface_repository_identity", "retrieved_at": "2026-10-17T19:15:03Z", "role": "primary", "sha256": "3ec7929329e8e1b09f31dee4a3e67128d68cddc95220a5944f4cd6102dae327e", "supports": ["aliases"], "url": "https://huggingface.co/api/models/inclusionAI/UI-Venus-2-9b"}], "id": "url:611df37e5f3caa6befd7", "identifiers": {}, "kind": "model", "last_verified_at": "2026-10-16T06:50:49Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/inclusionAI/UI-Venus-2-9b"}], "name": "UI-Venus 2 9B", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/api/models/inclusionAI/UI-Venus-2-9b", "target_id": "github:inclusionai/ui-venus", "type": "variant_of"}], "review_status": "curated", "scope": "core", "summary": "UI-Venus 2 GUI-agent checkpoint fine-tuned from Qwen3.5-9B, with about 9.4B BF16 parameters in the Qwen3.5 multimodal architecture, loaded through Transformers for image-text-to-text.", "topics": ["grounding", "planning"], "updated_at": null, "url": "https://huggingface.co/inclusionAI/UI-Venus-2-9b"}
{"aliases": [], "authors": ["James Allen", "Nathanael Chambers", "George Ferguson", "Lucian Galescu", "Hyuckchul Jung", "Mary Swift", "William Taysom"], "details": {"publication": {"authors": ["James Allen", "Nathanael Chambers", "George Ferguson", "Lucian Galescu", "Hyuckchul Jung", "Mary Swift", "William Taysom"], "date_basis": "Supplied year precision: 2007. Papers section: AAAI-2007", "date_parts": [2007], "doi": null, "event_dates": null, "evidence_urls": ["https://aaai.org/Papers/AAAI/2007/AAAI07-240.pdf", "https://www.cs.rochester.edu/research/cisd/projects/plow/"], "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 ordered names follow the original published paper, without expanding initials or replacing names from secondary indexes. DOI and volume number are not supplied in the retained primary sources. No exact publication month/day is established.", "pages": "1514-1519", "pdf_url": "https://cdn.aaai.org/AAAI/2007/AAAI07-240.pdf", "publisher": null, "series": null, "title": "PLOW: A Collaborative Task Learning Agent", "type": "conference-paper", "url": "https://aaai.org/Papers/AAAI/2007/AAAI07-240.pdf", "venue": "Twenty-Second AAAI National Conference on Artificial Intelligence", "volume": null}}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:53:26Z", "role": "primary", "sha256": "795d275c9f85180561fd101652221d30fb98daca2c2a00faf2b57c397bdfd68d", "supports": ["details.publication.authors", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.title", "details.publication.url"], "url": "https://aaai.org/Papers/AAAI/2007/AAAI07-240.pdf"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:53:24Z", "role": "primary", "sha256": "c0b3571efd5f2e137e42b38107de29aca6668819bec84109912d514867d85b8c", "supports": ["details.publication.date_basis", "details.publication.date_parts", "details.publication.notes", "details.publication.type", "details.publication.venue"], "url": "https://www.cs.rochester.edu/research/cisd/projects/plow/"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:53:26Z", "role": "primary", "sha256": "795d275c9f85180561fd101652221d30fb98daca2c2a00faf2b57c397bdfd68d", "supports": ["details.publication.authors", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.title", "details.publication.url", "environments", "name", "scope", "summary", "topics"], "url": "https://aaai.org/Papers/AAAI/2007/AAAI07-240.pdf"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:53:24Z", "role": "primary", "sha256": "c0b3571efd5f2e137e42b38107de29aca6668819bec84109912d514867d85b8c", "supports": ["details.publication.date_basis", "details.publication.date_parts", "details.publication.notes", "details.publication.type", "details.publication.venue", "environments", "kind"], "url": "https://www.cs.rochester.edu/research/cisd/projects/plow/"}], "id": "url:628b5251757bb3888e89", "identifiers": {}, "kind": "paper", "last_verified_at": "2026-10-17T09:53:26Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://aaai.org/Papers/AAAI/2007/AAAI07-240.pdf"}, {"rel": "publication_pdf", "url": "https://cdn.aaai.org/AAAI/2007/AAAI07-240.pdf"}], "name": "PLOW: A Collaborative Task Learning Agent", "notes": "The annotation was reviewed against retained primary publication sources. Experiments have not been independently reproduced. The ordered names follow the original published paper, without expanding initials or replacing names from secondary indexes.", "published_at": null, "relations": [], "review_status": "curated", "scope": "core", "summary": "Describes PLOW, which learns executable web-task models from a session combining demonstrations, natural-language explanations, and dialogue. Browser instrumentation connects user actions with DOM structure to support task generalization.", "topics": ["web-navigation", "human-in-the-loop", "planning"], "updated_at": null, "url": "https://aaai.org/Papers/AAAI/2007/AAAI07-240.pdf"}
{"aliases": [], "authors": [], "details": {}, "environments": ["web", "desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:38:27Z", "role": "primary", "sha256": "14f573af0c4ea3e5922b8831b7f03724936963e293d7c0aaa3fccbb71dbb0cfa", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://openai.com/index/computer-using-agent/"}], "id": "url:74eeb7c9fb20674fe6e2", "identifiers": {}, "kind": "model", "last_verified_at": "2026-10-16T06:38:27Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://openai.com/index/computer-using-agent/"}], "name": "OpenAI CUA", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://openai.com/index/computer-using-agent/", "target_id": "organization:openai", "type": "published_by"}], "review_status": "curated", "scope": "core", "summary": "Computer-using model introduced with the Operator research preview.", "topics": ["planning", "grounding"], "updated_at": null, "url": "https://openai.com/index/computer-using-agent/"}
{"aliases": [], "authors": [], "details": {}, "environments": ["cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:38:27Z", "role": "primary", "sha256": "e7062cfb4c0948d3a2274843f9f493f34cb40dfd634371408086df6141a0e036", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://ai.google.dev/gemini-api/docs/computer-use"}], "id": "url:879d4e3e75e83582bde6", "identifiers": {}, "kind": "model", "last_verified_at": "2026-10-16T06:38:27Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://ai.google.dev/gemini-api/docs/computer-use"}], "name": "Gemini Computer Use", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://ai.google.dev/gemini-api/docs/computer-use", "target_id": "organization:google", "type": "published_by"}], "review_status": "curated", "scope": "core", "summary": "Gemini capability for generating interface actions from screenshots.", "topics": ["planning", "grounding"], "updated_at": null, "url": "https://ai.google.dev/gemini-api/docs/computer-use"}
{"aliases": [], "authors": [], "details": {"disabled": false, "download_url": "https://huggingface.co/Hcompany/Holo3-35B-A3B/tree/main", "gated": false, "hub_id": "Hcompany/Holo3-35B-A3B"}, "environments": ["web", "desktop", "cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:07:19Z", "role": "primary", "sha256": "0681f87b059110a4e5dff8955e9feca6c5ea118b326f729f9c0ed0fe49e42d03", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/Hcompany/Holo3-35B-A3B/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:07:18Z", "role": "primary", "sha256": "bc5324dd6ceea9a0cc6bf0639b99d24cdd172af3c5d247bf923fdce1dc11f582", "supports": ["identity", "release_status", "details", "licenses.weights"], "url": "https://huggingface.co/api/models/Hcompany/Holo3-35B-A3B"}, {"method": "huggingface_card", "retrieved_at": "2026-10-17T04:07:19Z", "role": "primary", "sha256": "0681f87b059110a4e5dff8955e9feca6c5ea118b326f729f9c0ed0fe49e42d03", "supports": ["licenses.weights"], "url": "https://huggingface.co/Hcompany/Holo3-35B-A3B/raw/main/README.md"}], "id": "url:a349aace3cfde25628a8", "identifiers": {}, "kind": "model", "last_verified_at": "2026-10-17T04:07:19Z", "licenses": {"code": null, "data": null, "document": null, "weights": {"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/Hcompany/Holo3-35B-A3B/raw/main/README.md"}}, "links": [{"rel": "primary", "url": "https://huggingface.co/Hcompany/Holo3-35B-A3B"}], "name": "Holo3 35B-A3B", "notes": "Fourteen public safetensors shards are present. The card declares Apache-2.0 for this release and describes a 35B-total/3B-active MoE architecture. No unpublished enterprise benchmark or training corpus is added solely from its performance claims.", "published_at": null, "relations": [], "review_status": "curated", "scope": "core", "summary": "A sparse Qwen3.5-based GUI-action checkpoint released for navigation and computer-use workflows.", "topics": ["grounding", "planning", "action-prediction"], "updated_at": null, "url": "https://huggingface.co/Hcompany/Holo3-35B-A3B"}
{"aliases": ["url:4338ec0f1a67bdb6a139", "microsoft/Fara-7B", "https://huggingface.co/microsoft/Fara-7B", "url:d416331cdf9730afd6b9", "microsoft/fara-7b", "https://huggingface.co/microsoft/fara-7b"], "authors": [], "details": {"artifact_type": "model-checkpoint", "disabled": false, "download_url": "https://huggingface.co/microsoft/Fara-7b/tree/main", "gated": false, "hub_id": "microsoft/Fara-7B", "languages": ["en"], "modalities": [], "size_category_reported": null}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:50:49Z", "role": "primary", "sha256": "92bfa3c1f34e7e76a35f72389a3b3dd9fdab5fcff0569c9fc8aabe331c703f07", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/api/models/microsoft/Fara-7b"}, {"method": "huggingface_api", "retrieved_at": "2026-10-16T06:50:49Z", "role": "primary", "sha256": "92bfa3c1f34e7e76a35f72389a3b3dd9fdab5fcff0569c9fc8aabe331c703f07", "supports": ["details", "licenses.weights"], "url": "https://huggingface.co/api/models/microsoft/Fara-7b"}, {"method": "huggingface_repository_identity", "retrieved_at": "2026-10-17T19:14:59Z", "role": "primary", "sha256": "cd1f83cd2ebf30d45501d1a31ecbe3bcb2039928d580c750733d2edb49916398", "supports": ["aliases"], "url": "https://huggingface.co/api/models/microsoft/Fara-7B"}, {"method": "huggingface_repository_identity", "retrieved_at": "2026-10-17T19:15:00Z", "role": "primary", "sha256": "cd1f83cd2ebf30d45501d1a31ecbe3bcb2039928d580c750733d2edb49916398", "supports": ["aliases"], "url": "https://huggingface.co/api/models/microsoft/Fara-7b"}, {"method": "huggingface_repository_identity", "retrieved_at": "2026-10-17T19:15:01Z", "role": "primary", "sha256": "cd1f83cd2ebf30d45501d1a31ecbe3bcb2039928d580c750733d2edb49916398", "supports": ["aliases"], "url": "https://huggingface.co/api/models/microsoft/fara-7b"}], "id": "url:a4da77ee63f879cc8e63", "identifiers": {}, "kind": "model", "last_verified_at": "2026-10-16T06:50:49Z", "licenses": {"code": null, "data": null, "document": null, "weights": {"basis": "publisher model/dataset card", "declared": "mit", "retrieved_at": "2026-10-16T06:50:49Z", "url": "https://huggingface.co/microsoft/Fara-7b"}}, "links": [{"rel": "primary", "url": "https://huggingface.co/microsoft/Fara-7b"}], "name": "Fara 7B", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/api/models/microsoft/Fara-7b", "target_id": "github:microsoft/fara", "type": "variant_of"}], "review_status": "curated", "scope": "core", "summary": "Microsoft's Fara checkpoint, a Qwen2.5-VL-architecture image-text-to-text model with about 8.3B BF16 parameters, English, MIT-licensed and linked to arXiv paper 2511.19663.", "topics": ["planning", "web-navigation"], "updated_at": null, "url": "https://huggingface.co/microsoft/Fara-7b"}
{"aliases": [], "authors": [], "details": {}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-20T12:51:44.325799+00:00", "role": "primary", "sha256": "1f0315e8758fe0edd97adb2c5253955dcdbe6d7bd8be393fe14dcd093483705f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://www.deeplearning.ai/courses/building-ai-browser-agents"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T13:41:10.403317+00:00", "role": "primary", "sha256": "0915d862638a308da6200d9603d668ab4d11bcabfa713ff0ccb82c7b21a5a92f", "supports": ["summary", "environments", "topics", "scope", "notes"], "url": "https://learn.deeplearning.ai/courses/building-ai-browser-agents/lesson/wwewm/building-a-simple-web-agent"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T13:42:56.029732+00:00", "role": "primary", "sha256": "20d17ca47d1d16826f8e2736d6639d0cf11e826ef28c788187e16adcac9d7561", "supports": ["summary", "environments", "topics", "scope", "notes"], "url": "https://learn.deeplearning.ai/courses/building-ai-browser-agents/lesson/lot5j/building-an-autonomous-web-agent"}], "id": "url:aa693ca3e8948623f31e", "identifiers": {}, "kind": "learning", "last_verified_at": "2026-10-20T12:51:44.325799+00:00", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://www.deeplearning.ai/courses/building-ai-browser-agents"}], "name": "Building AI Browser Agents", "notes": "Review covers the public syllabus and complete published transcripts of the simple-agent and autonomous-agent lessons, not the whole course or its video frames. The first uses Playwright and structured HTML extraction; the second uses a MultiOn session API and shows missed steps, continuations and clarifications. Hosted notebooks and helper files were not retrieved, and current service access was not tested. No account, enrollment, terms acceptance, installation or execution occurred. Course and example-code reuse terms remain unknown.", "published_at": null, "relations": [], "review_status": "curated", "scope": "core", "summary": "Browser-agent course with worked lessons on structured web extraction and multi-step browser interaction, followed by AgentQ search and self-correction concepts.", "topics": ["web-navigation", "planning"], "updated_at": null, "url": "https://www.deeplearning.ai/courses/building-ai-browser-agents"}
{"aliases": [], "authors": [], "details": {"disabled": false, "download_url": "https://huggingface.co/xlangai/OpenCUA-32B/tree/main", "gated": false, "hub_id": "xlangai/OpenCUA-32B"}, "environments": ["desktop", "linux", "windows", "macos"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:44:29Z", "role": "primary", "sha256": "151217eb96e4ab22a47c98766052e5cfd76497a2a9197cc125c0237f3fbc59f6", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/xlangai/OpenCUA-32B/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:44:29Z", "role": "primary", "sha256": "a3be2e1a6cd39ee89809baa9504813c4a6a9a03bbd916e82924cc0f89cb5da33", "supports": ["identity", "release_status", "details", "licenses.weights"], "url": "https://huggingface.co/api/models/xlangai/OpenCUA-32B"}, {"method": "huggingface_card", "retrieved_at": "2026-10-17T02:44:29Z", "role": "primary", "sha256": "151217eb96e4ab22a47c98766052e5cfd76497a2a9197cc125c0237f3fbc59f6", "supports": ["licenses.weights"], "url": "https://huggingface.co/xlangai/OpenCUA-32B/raw/main/README.md"}], "id": "url:ad7f176848fd84de64e2", "identifiers": {}, "kind": "model", "last_verified_at": "2026-10-17T02:44:29Z", "licenses": {"code": null, "data": null, "document": null, "weights": {"basis": "publisher model/dataset card metadata; third-party assets retain their own terms", "declared": "mit", "retrieved_at": "2026-10-17T02:44:29Z", "url": "https://huggingface.co/xlangai/OpenCUA-32B/raw/main/README.md"}}, "links": [{"rel": "primary", "url": "https://huggingface.co/xlangai/OpenCUA-32B"}], "name": "OpenCUA 32B", "notes": "The public manifest contains 64 safetensors shards. The card declares MIT for this model and names its Qwen2.5-VL base; upstream terms remain separate. This is the 32B release, distinct from the existing 72B preview. Family-level or 72B benchmark scores are not transferred to it.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/xlangai/OpenCUA-32B/raw/main/README.md", "target_id": "arxiv:2508.09123", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "A Qwen2.5-VL-based computer-use checkpoint that combines screenshot history with planning and grounded executable actions.", "topics": ["grounding", "planning", "self-reflection"], "updated_at": null, "url": "https://huggingface.co/xlangai/OpenCUA-32B"}
{"aliases": [], "authors": [], "details": {"disabled": false, "download_url": "https://huggingface.co/mPLUG/GUI-Owl-7B/tree/main", "gated": false, "hub_id": "mPLUG/GUI-Owl-7B"}, "environments": ["android", "desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:03:41Z", "role": "primary", "sha256": "a0e6dd76aee6f138c15f244f4e2e8ba650b8829295397fdbfa839a8edf852b57", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/mPLUG/GUI-Owl-7B/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:03:40Z", "role": "primary", "sha256": "4822eab7fd35e583d52db7369d7d929d69dbde66b27c9da3d370cad9ec06ec89", "supports": ["identity", "release_status", "details", "licenses.weights"], "url": "https://huggingface.co/api/models/mPLUG/GUI-Owl-7B"}, {"method": "huggingface_card", "retrieved_at": "2026-10-17T04:03:41Z", "role": "primary", "sha256": "a0e6dd76aee6f138c15f244f4e2e8ba650b8829295397fdbfa839a8edf852b57", "supports": ["licenses.weights"], "url": "https://huggingface.co/mPLUG/GUI-Owl-7B/raw/main/README.md"}], "id": "url:aeb32973694f91ad9920", "identifiers": {}, "kind": "model", "last_verified_at": "2026-10-17T04:03:41Z", "licenses": {"code": null, "data": null, "document": null, "weights": {"basis": "publisher model/dataset card metadata; third-party assets retain their own terms", "declared": "mit", "retrieved_at": "2026-10-17T04:03:41Z", "url": "https://huggingface.co/mPLUG/GUI-Owl-7B/raw/main/README.md"}}, "links": [{"rel": "primary", "url": "https://huggingface.co/mPLUG/GUI-Owl-7B"}], "name": "GUI-Owl 7B", "notes": "Five public safetensors shards are present. MIT is declared in this model card. This record identifies the original 7B release; other sizes and later GUI-Owl generations are not aliases or assumed to share its exact terms. The online demo is still marked forthcoming.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/mPLUG/GUI-Owl-7B/raw/main/README.md", "target_id": "arxiv:2508.15144", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "A Qwen2.5-VL-based GUI checkpoint released with Mobile-Agent-v3 for grounded actions and specialized roles in its agent framework.", "topics": ["grounding", "action-prediction", "planning"], "updated_at": null, "url": "https://huggingface.co/mPLUG/GUI-Owl-7B"}
{"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": [], "details": {}, "environments": ["cross-platform"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:38:27Z", "role": "primary", "sha256": "84ce0fd682367a6dac5cc5abc39d5c877556024bbf6ffdad30f76fc2ec08523f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://hcompany.ai/research/models"}], "id": "url:c0bd07d82ce25ad1d99d", "identifiers": {}, "kind": "model", "last_verified_at": "2026-10-16T06:38:27Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://hcompany.ai/research/models"}], "name": "Holo Model Family", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://hcompany.ai/research/models", "target_id": "organization:h", "type": "published_by"}], "review_status": "curated", "scope": "core", "summary": "H computer-use model family and its release-specific access options.", "topics": ["planning", "grounding"], "updated_at": null, "url": "https://hcompany.ai/research/models"}
{"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": {"base_model": "Qwen/Qwen2-VL-7B", "download_url": "https://huggingface.co/osunlp/Dreamer-7B/tree/main", "languages": ["en"]}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T11:10:45Z", "role": "primary", "sha256": "f15e05cc2718948e344bd1fb943402797ef2e3f2130a9ce2fd94d419210c0574", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/osunlp/Dreamer-7B/raw/main/README.md"}, {"method": "huggingface_card", "retrieved_at": "2026-10-16T11:10:45Z", "role": "primary", "sha256": "f15e05cc2718948e344bd1fb943402797ef2e3f2130a9ce2fd94d419210c0574", "supports": ["licenses.weights"], "url": "https://huggingface.co/osunlp/Dreamer-7B/raw/main/README.md"}], "id": "url:e3c1a2a2d556e4728953", "identifiers": {}, "kind": "model", "last_verified_at": "2026-10-16T11:10:45Z", "licenses": {"code": null, "data": null, "document": null, "weights": {"basis": "publisher model/dataset card metadata; third-party assets retain their own terms", "declared": "apache-2.0", "retrieved_at": "2026-10-16T11:10:45Z", "url": "https://huggingface.co/osunlp/Dreamer-7B/raw/main/README.md"}}, "links": [{"rel": "primary", "url": "https://huggingface.co/osunlp/Dreamer-7B"}], "name": "Dreamer-7B", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/osunlp/Dreamer-7B/raw/main/README.md", "target_id": "arxiv:2411.06559", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "WebDreamer model checkpoint that predicts the consequences of web actions from screenshots and textual descriptions.", "topics": ["world-models", "planning"], "updated_at": null, "url": "https://huggingface.co/osunlp/Dreamer-7B"}
{"aliases": [], "authors": [], "details": {"disabled": false, "download_url": "https://huggingface.co/Hcompany/Holo1-7B/tree/main", "gated": false, "hub_id": "Hcompany/Holo1-7B"}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:44:50Z", "role": "primary", "sha256": "a7e185380bb02150d761651420cccc9c4a57f67c70a5dec3b7f85af63fb8773b", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/Hcompany/Holo1-7B/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:44:50Z", "role": "primary", "sha256": "66076674ebbdcf9faeb408ebd137ac1283fb56b1011acd7037ebb73e09ba749d", "supports": ["identity", "release_status", "details", "licenses.weights"], "url": "https://huggingface.co/api/models/Hcompany/Holo1-7B"}, {"method": "huggingface_card", "retrieved_at": "2026-10-17T02:44:50Z", "role": "primary", "sha256": "a7e185380bb02150d761651420cccc9c4a57f67c70a5dec3b7f85af63fb8773b", "supports": ["licenses.weights"], "url": "https://huggingface.co/Hcompany/Holo1-7B/raw/main/README.md"}], "id": "url:f35cfcbaeb79663eabeb", "identifiers": {}, "kind": "model", "last_verified_at": "2026-10-17T02:44:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": {"basis": "publisher model/dataset card metadata; third-party assets retain their own terms", "declared": "apache-2.0", "retrieved_at": "2026-10-17T02:44:50Z", "url": "https://huggingface.co/Hcompany/Holo1-7B/raw/main/README.md"}}, "links": [{"rel": "primary", "url": "https://huggingface.co/Hcompany/Holo1-7B"}], "name": "Holo1 7B", "notes": "Four public safetensors shards and example navigation/localization files are present. The model card declares Apache-2.0. Reported Surfer-H results depend on the assembled policy/localizer/validator configuration and are not imported as standalone checkpoint scores.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/Hcompany/Holo1-7B/raw/main/README.md", "target_id": "arxiv:2506.02865", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "A Qwen2.5-VL-based action model released for the Surfer-H web-agent system, supporting policy, interface-localization and validation roles.", "topics": ["web-navigation", "grounding", "planning"], "updated_at": null, "url": "https://huggingface.co/Hcompany/Holo1-7B"}
{"aliases": [], "authors": [], "details": {"disabled": false, "download_url": "https://huggingface.co/OpenGVLab/ScaleCUA-7B/tree/main", "gated": false, "hub_id": "OpenGVLab/ScaleCUA-7B"}, "environments": ["web", "desktop", "android", "linux"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:45:20Z", "role": "primary", "sha256": "e45116466b8b9ef3e93992a1efca8fa5b921a8b041297933210c9f552767ffb8", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/OpenGVLab/ScaleCUA-7B/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:45:19Z", "role": "primary", "sha256": "bad2617fd64633212a30b78d99fb053a662378ab50b42722a04af4104ac2248f", "supports": ["identity", "release_status", "details", "licenses.weights"], "url": "https://huggingface.co/api/models/OpenGVLab/ScaleCUA-7B"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T01:54:55Z", "role": "primary", "sha256": "0538701c860afedd5d63c3839039699d8e5a14977d5c339b187ed13ad666c8fa", "supports": ["environments", "relations"], "url": "https://raw.githubusercontent.com/OpenGVLab/ScaleCUA/HEAD/README.md"}, {"method": "huggingface_card", "retrieved_at": "2026-10-17T02:45:20Z", "role": "primary", "sha256": "e45116466b8b9ef3e93992a1efca8fa5b921a8b041297933210c9f552767ffb8", "supports": ["licenses.weights"], "url": "https://huggingface.co/OpenGVLab/ScaleCUA-7B/raw/main/README.md"}], "id": "url:ff7b1873a3eead89607a", "identifiers": {}, "kind": "model", "last_verified_at": "2026-10-17T02:45:20Z", "licenses": {"code": null, "data": null, "document": null, "weights": {"basis": "publisher model/dataset card metadata; third-party assets retain their own terms", "declared": "apache-2.0", "retrieved_at": "2026-10-17T02:45:20Z", "url": "https://huggingface.co/OpenGVLab/ScaleCUA-7B/raw/main/README.md"}}, "links": [{"rel": "primary", "url": "https://huggingface.co/OpenGVLab/ScaleCUA-7B"}], "name": "ScaleCUA 7B", "notes": "Four public safetensors shards and model/tokenizer configuration are present. The card declares Apache-2.0. The model card’s Paper link points to the code repository, so the arXiv association uses the repository’s explicit citation rather than treating that link as a paper URL.", "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": "A Qwen2.5-VL-based computer-use checkpoint with direct grounding and agent action modes across desktop, mobile and web interfaces.", "topics": ["grounding", "action-prediction", "planning"], "updated_at": null, "url": "https://huggingface.co/OpenGVLab/ScaleCUA-7B"}