{"aliases": [], "authors": ["Toby Jia-Jun Li", "Marissa Radensky", "Justin Jia", "Kirielle Singarajah", "Tom M. Mitchell", "Brad A. Myers"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "4cef465f53abcbaad50c83f24e0ce54fb7f23390955e9f7e5592ec0b8287839e", "submitted_at": "2019-08-30T18:35:01Z", "updated_at": "2020-01-07T00:57:53Z", "version": 2, "versioned_url": "https://arxiv.org/abs/1909.00031v2"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T07:57:26Z", "role": "primary", "sha256": "3e89566cdb4521a8b3b0d99c514c3401645eda5b5c0de9720eb726b1ffe0a11f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=1511.03745%2C1602.02261%2C1809.04860%2C1811.06521%2C1811.07871%2C1812.03426%2C1901.11528%2C1909.00031%2C1912.03098%2C2003.00431%2C2004.14973%2C2006.16241%2C2007.04428%2C2010.09670%2C2010.12083%2C2012.12350%2C2104.08244%2C2105.11941%2C2106.09422%2C2106.13500%2C2107.05911%2C2108.03298%2C2108.05308%2C2110.10661%2C2201.04100%2C2201.11676%2C2203.10897%2C2204.05239%2C2205.10816%2C2206.08853%2C2206.12839%2C2206.15407%2C2208.06032%2C2208.08080%2C2209.14981%2C2210.01379%2C2210.03735%2C2210.08073%2C2210.10275%2C2210.11435%2C2211.07912%2C2211.09527%2C2212.13326%2C2301.10677%2C2301.13779%2C2302.12173%2C2303.00001%2C2303.11366%2C2303.12027%2C2304.03442&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:11:46Z", "role": "primary", "sha256": "ee42fc28e088432e79837641c310dbd2d72f260a2a01d86f5f7f8b9c1b544a39", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/abs/1909.00031v2"}], "id": "arxiv:1909.00031", "identifiers": {"arxiv": "1909.00031"}, "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/1909.00031"}], "name": "Interactive Task and Concept Learning from Natural Language Instructions and GUI Demonstrations", "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": "2019-08-30", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces PUMICE, an agent that learns mobile-app procedures and conditional concepts through natural-language conversation and GUI demonstrations, recursively resolving ambiguous instructions with users.", "topics": ["human-in-the-loop", "mobile-navigation", "imitation-learning"], "updated_at": "2020-01-07", "url": "https://arxiv.org/abs/1909.00031"}
{"aliases": [], "authors": ["Andrea Burns", "Deniz Arsan", "Sanjna Agrawal", "Ranjitha Kumar", "Kate Saenko", "Bryan A. Plummer"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "1687369e31b0ad3632381d4a46b1d555c2e2ed1d6504e9c1b63006dd29f3445e", "submitted_at": "2022-02-04T18:51:50Z", "updated_at": "2022-08-15T00:24:24Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2202.02312v3"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=1802.08802%2C1808.09132%2C2005.03776%2C2101.09465%2C2103.16057%2C2104.08560%2C2105.13231%2C2107.03374%2C2107.13731%2C2108.03353%2C2202.02312%2C2202.08137%2C2204.01691%2C2205.11029%2C2207.01206%2C2209.08199%2C2209.08655%2C2209.14927%2C2209.15099%2C2210.02663%2C2210.03629%2C2210.03945%2C2211.07615%2C2301.10165%2C2301.13280%2C2302.04761%2C2303.07280%2C2303.08774%2C2303.17491%2C2303.17580%2C2304.04370%2C2304.07061%2C2305.04091%2C2305.05662%2C2305.08144%2C2305.10601%2C2305.11854%2C2305.12487%2C2305.14257%2C2305.14325%2C2305.16504%2C2305.16653%2C2305.19308%2C2306.00245%2C2306.06070%2C2306.07863%2C2307.10088%2C2307.12856%2C2307.13854%2C2308.01990&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "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:2202.02312", "identifiers": {"arxiv": "2202.02312"}, "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/2202.02312"}], "name": "A Dataset for Interactive Vision-Language Navigation with Unknown Command Feasibility", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2022-02-04", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces MoTIF mobile-app interactions with feasible and infeasible commands, feasibility explanations, and follow-up questions for ambiguous requests.", "topics": ["mobile-navigation", "human-in-the-loop", "evaluation"], "updated_at": "2022-08-15", "url": "https://arxiv.org/abs/2202.02312"}
{"aliases": [], "authors": ["Liangtai Sun", "Xingyu Chen", "Lu Chen", "Tianle Dai", "Zichen Zhu", "Kai Yu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "fbe2a4ece7ef72dfa008740991a5c46d9ab4b9de73f3de02f5dabcee2a5c1aa5", "submitted_at": "2022-05-23T04:05:37Z", "updated_at": "2022-11-24T06:34:43Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2205.11029v2"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=1802.08802%2C1808.09132%2C2005.03776%2C2101.09465%2C2103.16057%2C2104.08560%2C2105.13231%2C2107.03374%2C2107.13731%2C2108.03353%2C2202.02312%2C2202.08137%2C2204.01691%2C2205.11029%2C2207.01206%2C2209.08199%2C2209.08655%2C2209.14927%2C2209.15099%2C2210.02663%2C2210.03629%2C2210.03945%2C2211.07615%2C2301.10165%2C2301.13280%2C2302.04761%2C2303.07280%2C2303.08774%2C2303.17491%2C2303.17580%2C2304.04370%2C2304.07061%2C2305.04091%2C2305.05662%2C2305.08144%2C2305.10601%2C2305.11854%2C2305.12487%2C2305.14257%2C2305.14325%2C2305.16504%2C2305.16653%2C2305.19308%2C2306.00245%2C2306.06070%2C2306.07863%2C2307.10088%2C2307.12856%2C2307.13854%2C2308.01990&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "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:2205.11029", "identifiers": {"arxiv": "2205.11029"}, "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/2205.11029"}], "name": "META-GUI: Towards Multi-modal Conversational Agents on Mobile GUI", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2022-05-23", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces GUI-based task-oriented dialogue and a dataset pairing mobile-interface operations with conversational responses.", "topics": ["mobile-navigation", "human-in-the-loop"], "updated_at": "2022-11-24", "url": "https://arxiv.org/abs/2205.11029"}
{"aliases": [], "authors": ["Tao Li", "Gang Li", "Jingjie Zheng", "Purple Wang", "Yang Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "15e57e7dd30e6f3ae8b13cbd4a190b39339f2bbf7a551a6a5df5190dc62b5434", "submitted_at": "2022-09-29T21:08:18Z", "updated_at": "2022-09-29T21:08:18Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2209.15099v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=1802.08802%2C1808.09132%2C2005.03776%2C2101.09465%2C2103.16057%2C2104.08560%2C2105.13231%2C2107.03374%2C2107.13731%2C2108.03353%2C2202.02312%2C2202.08137%2C2204.01691%2C2205.11029%2C2207.01206%2C2209.08199%2C2209.08655%2C2209.14927%2C2209.15099%2C2210.02663%2C2210.03629%2C2210.03945%2C2211.07615%2C2301.10165%2C2301.13280%2C2302.04761%2C2303.07280%2C2303.08774%2C2303.17491%2C2303.17580%2C2304.04370%2C2304.07061%2C2305.04091%2C2305.05662%2C2305.08144%2C2305.10601%2C2305.11854%2C2305.12487%2C2305.14257%2C2305.14325%2C2305.16504%2C2305.16653%2C2305.19308%2C2306.00245%2C2306.06070%2C2306.07863%2C2307.10088%2C2307.12856%2C2307.13854%2C2308.01990&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "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:2209.15099", "identifiers": {"arxiv": "2209.15099"}, "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/2209.15099"}], "name": "MUG: Interactive Multimodal Grounding on User 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": "2022-09-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "Studies iterative UI grounding in which users refine or correct an agent's interpretation over multiple conversational turns.", "topics": ["grounding", "human-in-the-loop"], "updated_at": "2022-09-29", "url": "https://arxiv.org/abs/2209.15099"}
{"aliases": [], "authors": ["Tinghe Ding"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "9b8aaa2bccafd7f41802e5b7c25bc98f78307f77aeb36b7c75c57ef0f8e583a1", "submitted_at": "2024-01-04T03:44:42Z", "updated_at": "2024-01-17T06:35:45Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2401.04124v3"}}, "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:2401.04124", "identifiers": {"arxiv": "2401.04124"}, "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/2401.04124"}], "name": "MobileAgent: enhancing mobile control via human-machine interaction and SOP integration", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-01-04", "relations": [], "review_status": "curated", "scope": "core", "summary": "Combines operating-procedure context with user interaction to guide mobile task automation and handle personalized or sensitive requests.", "topics": ["mobile-navigation", "workflow", "human-in-the-loop"], "updated_at": "2024-01-17", "url": "https://arxiv.org/abs/2401.04124"}
{"aliases": [], "authors": ["Minh Duc Vu", "Han Wang", "Zhuang Li", "Jieshan Chen", "Shengdong Zhao", "Zhenchang Xing", "Chunyang Chen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a1c95264a931b10a57ee4b364498e2d2795abaf9a3e89096a0f29d2408174693", "submitted_at": "2024-01-25T16:02:56Z", "updated_at": "2024-08-14T00:48:43Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2401.14268v3"}}, "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:2401.14268", "identifiers": {"arxiv": "2401.14268"}, "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/2401.14268"}], "name": "GPTVoiceTasker: Advancing Multi-step Mobile Task Efficiency Through Dynamic Interface Exploration and Learning", "notes": "Reviewed against the retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2024-01-25", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces GPTVoiceTasker, a mobile assistant that interprets user commands, executes device interactions, and learns from prior commands to streamline later task execution. The work includes a user study of task efficiency.", "topics": ["mobile-navigation", "human-in-the-loop", "memory"], "updated_at": "2024-08-14", "url": "https://arxiv.org/abs/2401.14268"}
{"aliases": [], "authors": ["Tian Huang", "Chun Yu", "Weinan Shi", "Zijian Peng", "David Yang", "Weiqi Sun", "Yuanchun Shi"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6b2f357e166bddc2bbb5db4ec102721fc12fd27ba45d953193be0354fa5c7bce", "submitted_at": "2024-04-03T05:32:05Z", "updated_at": "2025-03-18T09:59:25Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2404.02475v2"}}, "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:2404.02475", "identifiers": {"arxiv": "2404.02475"}, "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/2404.02475"}], "name": "Prompt2Task: Automating UI Tasks on Smartphones from Textual Prompts", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-04-03", "relations": [], "review_status": "curated", "scope": "core", "summary": "Prompt2Task converts textual goals or procedures into smartphone UI automation using agents for intent interpretation, external-information handling, and execution. User feedback and accumulated task knowledge support subsequent task generation and operation.", "topics": ["mobile-navigation", "human-in-the-loop", "workflow"], "updated_at": "2025-03-18", "url": "https://arxiv.org/abs/2404.02475"}
{"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": ["Segev Shlomov", "Avi Yaeli", "Sami Marreed", "Sivan Schwartz", "Netanel Eder", "Offer Akrabi", "Sergey Zeltyn"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "4caa87e0b88d9375a23ca412ed19b3806f67712ca4ac222a39746e08ee80b07d", "submitted_at": "2024-07-22T14:41:32Z", "updated_at": "2024-08-18T09:02:43Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2407.15673v3"}}, "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:2407.15673", "identifiers": {"arxiv": "2407.15673"}, "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/2407.15673"}], "name": "IDA: Breaking Barriers in No-code UI Automation Through Large Language Models and Human-Centric Design", "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": "IDA helps business users create web UI automation through guided demonstrations, a semantic programming model, and a teaching-oriented interface. A prototype user study examines automation creation with enterprise applications by users without programming backgrounds.", "topics": ["human-in-the-loop", "automation", "workflow", "industry"], "updated_at": "2024-08-18", "url": "https://arxiv.org/abs/2407.15673"}
{"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": ["Faria Huq", "Zora Zhiruo Wang", "Frank F. Xu", "Tianyue Ou", "Shuyan Zhou", "Jeffrey P. Bigham", "Graham Neubig"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "9b14b88273749ac54ebbe3694cbe4196640da8f81ba2faf77b0e6bbd79a060f6", "submitted_at": "2025-01-28T00:56:53Z", "updated_at": "2026-02-27T05:35:06Z", "version": 4, "versioned_url": "https://arxiv.org/abs/2501.16609v4"}, "publication": {"authors": ["Faria Huq", "Zora Zhiruo Wang", "Frank F. Xu", "Tianyue Ou", "Shuyan Zhou", "Jeffrey P. Bigham", "Graham Neubig"], "date_basis": "The ACL Anthology citation_publication_date is 2025/4, and its visible Month/Year fields state April 2025. Only month precision is supplied; no day is inferred from conference dates.", "date_parts": [2025, 4], "doi": "10.18653/v1/2025.naacl-demo.17", "event_dates": null, "evidence_urls": ["https://arxiv.org/abs/2501.16609v4", "https://aclanthology.org/2025.naacl-demo.17/"], "identity_basis": "The retained arXiv v4 abstract page explicitly links this DOI. The ACL publication has the same complete title and the same seven authors in the same order, and describes the same CowPilot human-agent navigation framework. This attaches the proceedings edition to the existing arXiv identity.", "issue": null, "notes": "The April 2025 proceedings edition is distinct from the retained arXiv v4, last revised on 2026-02-27. The publisher abstract has minor wording and typesetting differences from the retained arXiv abstract; neither its text nor its date replaces the original arXiv snapshot or annotation. Exact publication day, numeric volume, issue and series are not established.", "pages": "163-172", "pdf_url": "https://aclanthology.org/2025.naacl-demo.17.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "CowPilot: A Framework for Autonomous and Human-Agent Collaborative Web Navigation", "type": "conference-paper", "url": "https://aclanthology.org/2025.naacl-demo.17/", "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": "2026-10-17T07:57:44Z", "role": "primary", "sha256": "9ca057ee6538002c739e58a7b3f86fc51a434d085a33e5590f0720e4f27c6b24", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2501.06497%2C2501.06994%2C2501.07834%2C2501.07927%2C2501.11233%2C2501.15570%2C2501.16403%2C2501.16609%2C2501.17391%2C2502.00372%2C2502.00392%2C2502.00964%2C2502.00988%2C2502.01562%2C2502.01839%2C2502.03723%2C2502.05174%2C2502.05227%2C2502.08489%2C2502.11221%2C2502.12130%2C2502.14496%2C2502.16587%2C2502.16786%2C2502.16796%2C2502.18356%2C2502.18512%2C2502.20383%2C2502.20390%2C2502.20582%2C2503.06378%2C2503.08026%2C2503.10613%2C2503.10809%2C2503.11579%2C2503.12228%2C2503.12461%2C2503.13843%2C2503.14495%2C2503.15426%2C2503.16036%2C2503.16248%2C2503.16975%2C2503.17671%2C2503.18102%2C2503.21889%2C2503.21949%2C2503.23803%2C2504.01328%2C2504.01377&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:19:57Z", "role": "primary", "sha256": "8c39a7773a9f2b58950aa22d6924b00ed32ab5b2b42302590dcec88da255a71f", "supports": ["details.publication.doi", "details.publication.identity_basis", "details.publication.notes"], "url": "https://arxiv.org/abs/2501.16609v4"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:44:12Z", "role": "primary", "sha256": "410da7f4b7a79e56efc5f88014a078da0feb8d2ae3ef3d36a04dce5be29d4250", "supports": ["details.publication.type", "details.publication.title", "details.publication.authors", "details.publication.date_parts", "details.publication.date_basis", "details.publication.url", "details.publication.pdf_url", "details.publication.doi", "details.publication.venue", "details.publication.publisher", "details.publication.pages", "details.publication.identity_basis", "details.publication.notes"], "url": "https://aclanthology.org/2025.naacl-demo.17/"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:19:57Z", "role": "primary", "sha256": "8c39a7773a9f2b58950aa22d6924b00ed32ab5b2b42302590dcec88da255a71f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/abs/2501.16609v4"}], "id": "arxiv:2501.16609", "identifiers": {"arxiv": "2501.16609"}, "kind": "paper", "last_verified_at": "2026-10-17T11:19:57Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2501.16609"}, {"rel": "publication", "url": "https://aclanthology.org/2025.naacl-demo.17/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2025.naacl-demo.17.pdf"}], "name": "CowPilot: A Framework for Autonomous and Human-Agent Collaborative Web Navigation", "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": "2025-01-28", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces CowPilot, a web-navigation framework where agents suggest actions and users can pause, reject, override, or resume them. It evaluates autonomous and collaborative use through task success and human effort.", "topics": ["web-navigation", "human-in-the-loop", "evaluation"], "updated_at": "2026-02-27", "url": "https://arxiv.org/abs/2501.16609"}
{"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": ["Piero A. Bonatti", "John Domingue", "Anna Lisa Gentile", "Andreas Harth", "Olaf Hartig", "Aidan Hogan", "Katja Hose", "Ernesto Jimenez-Ruiz", "Deborah L. McGuinness", "Chang Sun", "Ruben Verborgh", "Jesse Wright"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "4c778da33baa88916fbcb3e02a33f760dccea8b193d25087df9b56cd42c3c248", "submitted_at": "2025-01-31T12:26:27Z", "updated_at": "2025-01-31T12:26:27Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2503.15515v1"}}, "environments": ["web", "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:2503.15515", "identifiers": {"arxiv": "2503.15515"}, "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/2503.15515"}], "name": "Towards Computer-Using Personal 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": "2025-01-31", "relations": [], "review_status": "curated", "scope": "core", "summary": "Proposes computer-using personal agents connected to user-controlled data repositories, discussing privacy, interoperability, and collaborative tasks.", "topics": ["safety-security", "memory", "human-in-the-loop"], "updated_at": "2025-01-31", "url": "https://arxiv.org/abs/2503.15515"}
{"aliases": [], "authors": ["Pengzhou Cheng", "Zheng Wu", "Zongru Wu", "Aston Zhang", "Zhuosheng Zhang", "Gongshen Liu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "aa6201528b13c6a1aae8dcd745f4f5d995e79982ef7c54a1962e2f9035233293", "submitted_at": "2025-02-26T12:31:16Z", "updated_at": "2025-07-14T15:56:44Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2503.16465v3"}}, "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:2503.16465", "identifiers": {"arxiv": "2503.16465"}, "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/2503.16465"}], "name": "OS-Kairos: Adaptive Interaction for MLLM-Powered 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": "2025-02-26", "relations": [], "review_status": "curated", "scope": "core", "summary": "Uses step-level confidence estimates to decide when a GUI agent should act autonomously or request human intervention in complex situations.", "topics": ["human-in-the-loop", "safety-security", "mobile-navigation"], "updated_at": "2025-07-14", "url": "https://arxiv.org/abs/2503.16465"}
{"aliases": [], "authors": ["Noam Kahlon", "Guy Rom", "Anatoly Efros", "Filippo Galgani", "Omri Berkovitch", "Sapir Caduri", "William E. Bishop", "Oriana Riva", "Ido Dagan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "8a49d662eddf84690395b025e50f3378c3be2a3fc8abe02bd7202c6ff879bbe3", "submitted_at": "2025-03-25T10:46:08Z", "updated_at": "2025-03-25T10:46:08Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2503.19537v1"}}, "environments": ["android"], "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:2503.19537", "identifiers": {"arxiv": "2503.19537"}, "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/2503.19537"}], "name": "Agent-Initiated Interaction in Phone UI 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": "2025-03-25", "relations": [], "review_status": "curated", "scope": "core", "summary": "Defines when phone agents should request user interaction and introduces AndroidInteraction annotations for deciding whether and how to engage the user.", "topics": ["human-in-the-loop", "mobile-navigation", "evaluation"], "updated_at": "2025-03-25", "url": "https://arxiv.org/abs/2503.19537"}
{"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": ["Suyu Ye", "Haojun Shi", "Darren Shih", "Hyokun Yun", "Tanya Roosta", "Tianmin Shu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "268adf347e1f224f434e6dba838873c99a4050686cae1495643d9638d840d809", "submitted_at": "2025-04-14T17:36:46Z", "updated_at": "2025-12-01T15:22:40Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2504.10445v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:42Z", "role": "primary", "sha256": "afe4e75ffb8a5c44f4c088a90ffa5f53553036f514aa768f5aee0c5aa38d731e", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2503.16788%2C2503.17709%2C2503.18492%2C2503.19537%2C2503.20978%2C2503.21620%2C2503.23434%2C2504.01382%2C2504.07981%2C2504.10445%2C2504.10458%2C2504.11257%2C2504.11281%2C2504.12679%2C2504.13805%2C2504.13865%2C2504.14239%2C2504.18575%2C2504.19838%2C2504.20464%2C2505.10924%2C2505.11891%2C2505.12299%2C2505.12632%2C2505.13227%2C2505.13887%2C2505.13909%2C2505.14289%2C2505.15810%2C2505.16282%2C2505.16827%2C2505.18829%2C2505.19897%2C2505.21496%2C2505.21499%2C2505.21936%2C2505.22648%2C2505.23762%2C2505.24838%2C2506.00618%2C2506.01391%2C2506.01968%2C2506.02456%2C2506.03143%2C2506.03569%2C2506.04135%2C2506.04614%2C2506.07672%2C2506.08012%2C2506.08972&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:42Z", "role": "primary", "sha256": "afe4e75ffb8a5c44f4c088a90ffa5f53553036f514aa768f5aee0c5aa38d731e", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2503.16788%2C2503.17709%2C2503.18492%2C2503.19537%2C2503.20978%2C2503.21620%2C2503.23434%2C2504.01382%2C2504.07981%2C2504.10445%2C2504.10458%2C2504.11257%2C2504.11281%2C2504.12679%2C2504.13805%2C2504.13865%2C2504.14239%2C2504.18575%2C2504.19838%2C2504.20464%2C2505.10924%2C2505.11891%2C2505.12299%2C2505.12632%2C2505.13227%2C2505.13887%2C2505.13909%2C2505.14289%2C2505.15810%2C2505.16282%2C2505.16827%2C2505.18829%2C2505.19897%2C2505.21496%2C2505.21499%2C2505.21936%2C2505.22648%2C2505.23762%2C2505.24838%2C2506.00618%2C2506.01391%2C2506.01968%2C2506.02456%2C2506.03143%2C2506.03569%2C2506.04135%2C2506.04614%2C2506.07672%2C2506.08012%2C2506.08972&max_results=50"}], "id": "arxiv:2504.10445", "identifiers": {"arxiv": "2504.10445"}, "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/2504.10445"}], "name": "RealWebAssist: A Benchmark for Long-Horizon Web Assistance with Real-World Users", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-04-14", "relations": [], "review_status": "curated", "scope": "core", "summary": "RealWebAssist evaluates web assistance through sequences of instructions collected from real users across multiple websites. It tests intent interpretation, user-specific routines, long-term context, and visual grounding when instructions are ambiguous or evolve during interaction.", "topics": ["evaluation", "long-horizon", "human-in-the-loop", "grounding"], "updated_at": "2025-12-01", "url": "https://arxiv.org/abs/2504.10445"}
{"aliases": [], "authors": ["Chaoran Chen", "Zhiping Zhang", "Bingcan Guo", "Shang Ma", "Ibrahim Khalilov", "Simret A Gebreegziabher", "Yanfang Ye", "Ziang Xiao", "Yaxing Yao", "Tianshi Li", "Toby Jia-Jun Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "48acf3279b854d45ec076e630e5a84601879fc9ff894c5d368adb59d03bd3de6", "submitted_at": "2025-04-15T15:21:09Z", "updated_at": "2025-04-15T15:21:09Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2504.11281v1"}}, "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", "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=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2504.11281", "identifiers": {"arxiv": "2504.11281"}, "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/2504.11281"}], "name": "The Obvious Invisible Threat: LLM-Powered GUI Agents' Vulnerability to Fine-Print Injections", "notes": "Reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2025-04-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "Characterizes six forms of GUI-content injection and studies their effects on GUI agents and human users using adversarial webpages. The analysis examines privacy failures and the limits of human oversight.", "topics": ["safety-security", "human-in-the-loop", "evaluation"], "updated_at": "2025-04-15", "url": "https://arxiv.org/abs/2504.11281"}
{"aliases": [], "authors": ["Guangyi Liu", "Pengxiang Zhao", "Liang Liu", "Zhiming Chen", "Yuxiang Chai", "Shuai Ren", "Hao Wang", "Shibo He", "Wenchao Meng"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2bbe6c4969a9557546752235809d25912bff96a97001fade30102d8d2c751054", "submitted_at": "2025-04-18T17:13:34Z", "updated_at": "2025-04-18T17:13:34Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2504.13805v1"}}, "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:2504.13805", "identifiers": {"arxiv": "2504.13805"}, "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/2504.13805"}], "name": "LearnAct: Few-Shot Mobile GUI Agent with a Unified Demonstration Benchmark", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-04-18", "relations": [], "review_status": "curated", "scope": "core", "summary": "LearnAct parses human demonstrations into reusable knowledge, retrieves relevant examples, and uses them during mobile GUI task execution. The accompanying LearnGUI benchmark includes 2,252 offline tasks and 101 online tasks for studying adaptation from demonstrations.", "topics": ["mobile-navigation", "memory", "human-in-the-loop", "evaluation"], "updated_at": "2025-04-18", "url": "https://arxiv.org/abs/2504.13805"}
{"aliases": [], "authors": ["Chaoran Chen", "Zhiping Zhang", "Ibrahim Khalilov", "Bingcan Guo", "Simret A Gebreegziabher", "Yanfang Ye", "Ziang Xiao", "Yaxing Yao", "Tianshi Li", "Toby Jia-Jun Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d2a55e347bc10e5571464924d9301ff2c32d5e9f1f990ca91725ec8c9f6da594", "submitted_at": "2025-04-24T20:51:20Z", "updated_at": "2025-06-05T12:40:53Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2504.17934v2"}}, "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", "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=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2504.17934", "identifiers": {"arxiv": "2504.17934"}, "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/2504.17934"}], "name": "Toward a Human-Centered Evaluation Framework for Trustworthy LLM-Powered GUI Agents", "notes": "Reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced. This is a position paper proposing an evaluation framework; the abstract does not report a new benchmark or empirical validation of the proposed framework.", "published_at": "2025-04-24", "relations": [], "review_status": "curated", "scope": "core", "summary": "Presents a position paper on privacy and security evaluation for GUI agents, identifying risk categories and challenges in involving human evaluators. It proposes risk-aware assessment and in-context consent as design directions.", "topics": ["evaluation", "safety-security", "human-in-the-loop"], "updated_at": "2025-06-05", "url": "https://arxiv.org/abs/2504.17934"}
{"aliases": [], "authors": ["Shuning Zhang", "Jingruo Chen", "Zhiqi Gao", "Jiajing Gao", "Xin Yi", "Hewu Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "63f4a505645fec16eac6b59923a6c1979c53376f19c6a4ea53c9a5d32a45e3d4", "submitted_at": "2025-05-15T00:42:51Z", "updated_at": "2025-05-16T07:01:20Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2505.09875v2"}}, "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", "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=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2505.09875", "identifiers": {"arxiv": "2505.09875"}, "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/2505.09875"}], "name": "Characterizing Unintended Consequences in Human-GUI Agent Collaboration for Web Browsing", "notes": "Reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2025-05-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "Studies unintended consequences of collaborative web browsing with GUI agents through social-media analysis and interviews. It organizes observed failures, user impacts, and mitigation practices around planning, interface interaction, feedback, privacy, and trust.", "topics": ["human-in-the-loop", "evaluation", "safety-security"], "updated_at": "2025-05-16", "url": "https://arxiv.org/abs/2505.09875"}
{"aliases": [], "authors": ["Omar Shaikh", "Shardul Sapkota", "Shan Rizvi", "Eric Horvitz", "Joon Sung Park", "Diyi Yang", "Michael S. Bernstein"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d986ee6aa62b11759c366fa9f6a01a1a0b12eae715970eacd8053dbbb6622fe5", "submitted_at": "2025-05-16T04:00:31Z", "updated_at": "2025-09-21T00:03:34Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2505.10831v3"}}, "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", "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=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2505.10831", "identifiers": {"arxiv": "2505.10831"}, "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/2505.10831"}], "name": "Creating General User Models from Computer Use", "notes": "Reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2025-05-16", "relations": [], "review_status": "curated", "scope": "core", "summary": "Proposes general user models that turn computer-use observations into confidence-weighted propositions and revise them as new context arrives. Demonstrations use this memory to contextualize assistants, manage notifications, and support proactive actions across apps.", "topics": ["memory", "human-in-the-loop", "cross-app"], "updated_at": "2025-09-21", "url": "https://arxiv.org/abs/2505.10831"}
{"aliases": [], "authors": ["Weikai Xu", "Zhizheng Jiang", "Yuxuan Liu", "Pengzhi Gao", "Wei Liu", "Jian Luan", "Yuanchun Li", "Yunxin Liu", "Bin Wang", "Bo An"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "5ca1ebfdd5a96e146d37539294462c7fb67bf20deaf77307de37aa3bb1dde082", "submitted_at": "2025-05-17T07:58:34Z", "updated_at": "2026-02-02T06:55:10Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2505.11891v3"}}, "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:2505.11891", "identifiers": {"arxiv": "2505.11891"}, "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/2505.11891"}], "name": "Mobile-Bench-v2: A More Realistic and Comprehensive Benchmark for VLM-based Mobile Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-05-17", "relations": [], "review_status": "curated", "scope": "core", "summary": "Mobile-Bench-v2 evaluates mobile GUI agents using multiple acceptable offline action paths, noisy interfaces with ads or pop-ups, and ambiguous instructions with scripted clarification exchanges. These splits separate step-level execution, robustness, and proactive interaction.", "topics": ["evaluation", "mobile-navigation", "robustness", "human-in-the-loop"], "updated_at": "2026-02-02", "url": "https://arxiv.org/abs/2505.11891"}
{"aliases": [], "authors": ["Thong Q. Nguyen", "Shubhang Desai", "Raja Hasnain Anwar", "Firoz Shaik", "Vishwas Suryanarayanan", "Vishal Chowdhary"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "401ca18cadbf38045fa20985354988571b69e2280b1a090aff0e128cdf3c70e0", "submitted_at": "2025-06-03T07:25:49Z", "updated_at": "2025-08-07T21:48:01Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2506.02539v3"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "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=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2506.02539", "identifiers": {"arxiv": "2506.02539"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2506.02539"}], "name": "VerificAgent: Domain-Specific Memory Verification for Scalable Oversight of Aligned Computer-Use Agents", "notes": "Reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2025-06-03", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces VerificAgent, which grows domain-specific memories from task trajectories and subjects them to human fact-checking before deployment. Its OSWorld evaluation treats persistent memory as an auditable constraint on later agent behavior.", "topics": ["memory", "human-in-the-loop", "safety-security"], "updated_at": "2025-08-07", "url": "https://arxiv.org/abs/2506.02539"}
{"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": ["Hussein Mozannar", "Gagan Bansal", "Cheng Tan", "Adam Fourney", "Victor Dibia", "Jingya Chen", "Jack Gerrits", "Tyler Payne", "Matheus Kunzler Maldaner", "Madeleine Grunde-McLaughlin", "Eric Zhu", "Griffin Bassman", "Jacob Alber", "Peter Chang", "Ricky Loynd", "Friederike Niedtner", "Ece Kamar", "Maya Murad", "Rafah Hosn", "Saleema Amershi"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "c9416ede89036d640412496b5f0def218b69271c5f2449689d4e4afacb43c556", "submitted_at": "2025-07-30T03:49:14Z", "updated_at": "2025-07-30T03:49:14Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2507.22358v1"}}, "environments": ["web"], "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:2507.22358", "identifiers": {"arxiv": "2507.22358"}, "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/2507.22358"}], "name": "Magentic-UI: Towards Human-in-the-loop 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": "2025-07-30", "relations": [], "review_status": "curated", "scope": "core", "summary": "Magentic-UI combines browser interaction, code execution, file operations, and MCP tools in an interface for human-agent collaboration. It studies collaborative planning and task execution, action guards, and persistent memory through benchmark tasks, simulated users, user studies, and safety assessments.", "topics": ["human-in-the-loop", "tool-integration", "web-navigation", "safety-security"], "updated_at": "2025-07-30", "url": "https://arxiv.org/abs/2507.22358"}
{"aliases": [], "authors": ["Chao Hao", "Shuai Wang", "Kaiwen Zhou"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "499a462f9f14b8afdd367bb7cf7a7ffb1b2e166c212fb5c36686d3145f08ab5f", "submitted_at": "2025-08-06T02:38:02Z", "updated_at": "2025-08-06T02:38:02Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2508.04025v1"}}, "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:2508.04025", "identifiers": {"arxiv": "2508.04025"}, "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/2508.04025"}], "name": "Uncertainty-Aware GUI Agent: Adaptive Perception through Component Recommendation and Human-in-the-Loop Refinement", "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": "RecAgent narrows attention to recommended UI components and requests user feedback when task intent or decisions remain uncertain. It introduces ComplexAction to evaluate specified single-step GUI actions in difficult interface scenarios.", "topics": ["mobile-navigation", "human-in-the-loop", "screen-understanding", "evaluation"], "updated_at": "2025-08-06", "url": "https://arxiv.org/abs/2508.04025"}
{"aliases": [], "authors": ["Zeyu Huang", "Juyuan Wang", "Longfeng Chen", "Boyi Xiao", "Leng Cai", "Yawen Zeng", "Jin Xu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d4efa329c3b106219c9b25f675be0c561743085ac999d2964458a5f22ff6e62f", "submitted_at": "2025-08-12T16:18:30Z", "updated_at": "2025-08-14T16:36:45Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2508.09057v2"}}, "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"], "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:2508.09057", "identifiers": {"arxiv": "2508.09057"}, "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/2508.09057"}], "name": "MVISU-Bench: Benchmarking Mobile Agents for Real-World Tasks by Multi-App, Vague, Interactive, Single-App and Unethical Instructions", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-08-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "MVISU-Bench evaluates bilingual mobile tasks involving cross-app work, vague requests, user interaction and unethical instructions alongside ordinary single-app tasks. Its Aider prompting module is designed to clarify intent and mitigate risky execution within agent workflows.", "topics": ["evaluation", "mobile-navigation", "human-in-the-loop", "safety-security"], "updated_at": "2025-08-14", "url": "https://arxiv.org/abs/2508.09057"}
{"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": ["Qihang Ai", "Pi Bu", "Yue Cao", "Yingyao Wang", "Jihao Gu", "Jingxuan Xing", "Zekun Zhu", "Wei Jiang", "Zhicheng Zheng", "Jun Song", "Yuning Jiang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "385ed1b6e60c27660f0cbf281b0f2bbde052691801f7faae99cc053dfbc0d127", "submitted_at": "2025-08-27T08:40:05Z", "updated_at": "2026-04-25T13:29:08Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2508.19679v2"}}, "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"], "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:2508.19679", "identifiers": {"arxiv": "2508.19679"}, "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/2508.19679"}], "name": "InquireMobile: Teaching VLM-based Mobile Agent to Request Human Assistance via Reinforcement Fine-Tuning", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-08-27", "relations": [], "review_status": "curated", "scope": "core", "summary": "InquireMobile trains mobile agents to seek human confirmation at critical decisions using staged reinforcement fine-tuning and pre-action reasoning. InquireBench evaluates safe interaction and proactive inquiry, distinguishing asking for help from completing the underlying task.", "topics": ["human-in-the-loop", "reinforcement-learning", "evaluation", "mobile-navigation"], "updated_at": "2026-04-25", "url": "https://arxiv.org/abs/2508.19679"}
{"aliases": [], "authors": ["Yi-Hao Peng", "Dingzeyu Li", "Jeffrey P. Bigham", "Amy Pavel"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "5dc6411807fd58daed4824e86c39377b24d9fd022ac19daceb836bfb382353cf", "submitted_at": "2025-08-29T09:39:00Z", "updated_at": "2025-08-29T09:39:00Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2508.21456v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T07:57:56Z", "role": "primary", "sha256": "3429cc2fc7737050cac964c976293fc4793e52c5913636c1f67d04243fbefb6a", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2508.20410%2C2508.20890%2C2508.21456%2C2509.00088%2C2509.00531%2C2509.00969%2C2509.01684%2C2509.03536%2C2509.04018%2C2509.04908%2C2509.05368%2C2509.05831%2C2509.06917%2C2509.07098%2C2509.07553%2C2509.08820%2C2509.10345%2C2509.10540%2C2509.11253%2C2509.12543%2C2509.12987%2C2509.13615%2C2509.14547%2C2509.14744%2C2509.15061%2C2509.15237%2C2509.17336%2C2509.19185%2C2509.19199%2C2509.21072%2C2509.21291%2C2509.21766%2C2509.21799%2C2509.22715%2C2509.22830%2C2509.23738%2C2509.24704%2C2509.25455%2C2509.25693%2C2510.00172%2C2510.00515%2C2510.02190%2C2510.04491%2C2510.05096%2C2510.05156%2C2510.05788%2C2510.07284%2C2510.07915%2C2510.08787%2C2510.09517&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:20:40Z", "role": "primary", "sha256": "713ecfafe72365a57604306ccaba68f94158fc9b1aa169b408945922cdd1418c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/abs/2508.21456v1"}], "id": "arxiv:2508.21456", "identifiers": {"arxiv": "2508.21456"}, "kind": "paper", "last_verified_at": "2026-10-17T11:20:40Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2508.21456"}], "name": "Morae: Proactively Pausing UI Agents for User Choices", "notes": "Reviewed against the exact-version official arXiv abstract page; its complete title, ordered author list, visible version, and normalized abstract SHA match the retained snapshot. The original API response body was not locally available. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2025-08-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces Morae, a UI agent that detects decision points and pauses for user choices using interface code and screenshots. A study with blind and low-vision participants evaluates this mixed-initiative approach on real web tasks.", "topics": ["human-in-the-loop", "web-navigation", "evaluation"], "updated_at": "2025-08-29", "url": "https://arxiv.org/abs/2508.21456"}
{"aliases": [], "authors": ["Zheng Wu", "Heyuan Huang", "Xingyu Lou", "Xiangmou Qu", "Pengzhou Cheng", "Zongru Wu", "Weiwen Liu", "Weinan Zhang", "Jun Wang", "Zhaoxiang Wang", "Zhuosheng Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "3bd6b323a376d7606b8c5ec0615b992f47a388cbd3604191241f6a0a9b2cce6e", "submitted_at": "2025-09-09T09:46:01Z", "updated_at": "2026-04-03T02:36:28Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2509.07553v3"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T07:57:56Z", "role": "primary", "sha256": "3429cc2fc7737050cac964c976293fc4793e52c5913636c1f67d04243fbefb6a", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2508.20410%2C2508.20890%2C2508.21456%2C2509.00088%2C2509.00531%2C2509.00969%2C2509.01684%2C2509.03536%2C2509.04018%2C2509.04908%2C2509.05368%2C2509.05831%2C2509.06917%2C2509.07098%2C2509.07553%2C2509.08820%2C2509.10345%2C2509.10540%2C2509.11253%2C2509.12543%2C2509.12987%2C2509.13615%2C2509.14547%2C2509.14744%2C2509.15061%2C2509.15237%2C2509.17336%2C2509.19185%2C2509.19199%2C2509.21072%2C2509.21291%2C2509.21766%2C2509.21799%2C2509.22715%2C2509.22830%2C2509.23738%2C2509.24704%2C2509.25455%2C2509.25693%2C2510.00172%2C2510.00515%2C2510.02190%2C2510.04491%2C2510.05096%2C2510.05156%2C2510.05788%2C2510.07284%2C2510.07915%2C2510.08787%2C2510.09517&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:20:50Z", "role": "primary", "sha256": "776192081826494b05077d71757f3006575ed50a8cc244004e2de7053104d7eb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/abs/2509.07553v3"}], "id": "arxiv:2509.07553", "identifiers": {"arxiv": "2509.07553"}, "kind": "paper", "last_verified_at": "2026-10-17T11:20:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2509.07553"}], "name": "VeriOS: Query-Driven Proactive Human-Agent-GUI Interaction for Trustworthy OS Agents", "notes": "Reviewed against the exact-version official arXiv abstract page; its complete title, ordered author list, visible version, and normalized abstract SHA match the retained snapshot. The original API response body was not locally available. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2025-09-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces VeriOS-Agent, which learns when to execute GUI actions autonomously and when to query a user under untrustworthy conditions. Its staged training combines supervised learning and group-relative policy optimization.", "topics": ["human-in-the-loop", "safety-security", "reinforcement-learning"], "updated_at": "2026-04-03", "url": "https://arxiv.org/abs/2509.07553"}
{"aliases": [], "authors": ["Jingyu Tang", "Chaoran Chen", "Jiawen Li", "Zhiping Zhang", "Bingcan Guo", "Ibrahim Khalilov", "Simret Araya Gebreegziabher", "Bingsheng Yao", "Dakuo Wang", "Yanfang Ye", "Tianshi Li", "Ziang Xiao", "Yaxing Yao", "Toby Jia-Jun Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "247ed053a4c8f62106d9f8e9af0801d94265ed712f4ae9f095fd8aaccbe33c59", "submitted_at": "2025-09-12T22:26:31Z", "updated_at": "2025-09-12T22:26:31Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2509.10723v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2509.10723", "identifiers": {"arxiv": "2509.10723"}, "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/2509.10723"}], "name": "Dark Patterns Meet GUI Agents: LLM Agent Susceptibility to Manipulative Interfaces and the Role of Human Oversight", "notes": "Reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2025-09-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "Studies how GUI agents, people, and human-agent teams respond to manipulative interface designs. The two-phase evaluation examines failure modes and the benefits and costs of human oversight.", "topics": ["human-in-the-loop", "safety-security", "evaluation"], "updated_at": "2025-09-12", "url": "https://arxiv.org/abs/2509.10723"}
{"aliases": [], "authors": ["Shuning Zhang", "Yutong Jiang", "Rongjun Ma", "Yuting Yang", "Mingyao Xu", "Zhixin Huang", "Xin Yi", "Hewu Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "f37ebc10535f96d17d28990e608a1f2eb6cc004f40d58b3288c4a81d781160ea", "submitted_at": "2025-09-15T13:58:52Z", "updated_at": "2025-09-15T13:58:52Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2509.11939v1"}}, "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:2509.11939", "identifiers": {"arxiv": "2509.11939"}, "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/2509.11939"}], "name": "PrivWeb: Unobtrusive and Content-aware Privacy Protection For Web Agents", "notes": "Reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2025-09-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces PrivWeb, which uses a local language model to anonymize interface information according to user preferences. Adaptive notifications and selective pauses give users control over sensitive information access during web-agent tasks.", "topics": ["safety-security", "human-in-the-loop", "web-navigation"], "updated_at": "2025-09-15", "url": "https://arxiv.org/abs/2509.11939"}
{"aliases": [], "authors": ["Jiazheng Sun", "Ruimeng Yang", "Xu Han", "Jiayang Niu", "Mingxuan Li", "Te Yang", "Yongyong Lu", "Xin Peng"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "571eee6476c9e75ffae65cf98ec589e98fd3b953313cc6d2191bb48bd27303fe", "submitted_at": "2025-09-25T04:21:31Z", "updated_at": "2025-12-01T08:01:05Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2509.20729v2"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2509.20729", "identifiers": {"arxiv": "2509.20729"}, "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/2509.20729"}], "name": "Robust, Observable, and Evolvable Agentic Systems Engineering: A Principled Framework Validated via the Fairy GUI Agent", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2025-09-25", "relations": [], "review_status": "curated", "scope": "core", "summary": "Presents an agent-engineering framework combining runtime goal clarification, an observable component architecture and evolving execution memory. The Fairy mobile GUI agent evaluates these mechanisms on ambiguous tasks and through controlled and user studies.", "topics": ["observability", "human-in-the-loop", "memory", "robustness"], "updated_at": "2025-12-01", "url": "https://arxiv.org/abs/2509.20729"}
{"aliases": [], "authors": ["Sagnik Anupam", "Davis Brown", "Shuo Li", "Eric Wong", "Hamed Hassani", "Osbert Bastani"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "7906e06c2efb8778a43ec774821f90947e70cec0b78d6d43ac8901f7f0562332", "submitted_at": "2025-10-02T15:22:21Z", "updated_at": "2025-10-07T15:12:39Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2510.02418v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:32:52Z", "role": "primary", "sha256": "0b730323b2ce85c39aa7d534665901a480ff06540c73f1f80832fcc0735b8789", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2310.08588%2C2402.09205%2C2406.05720%2C2407.21075%2C2410.07166%2C2503.03743%2C2508.04037%2C2508.10833%2C2508.11360%2C2508.13024%2C2508.13186%2C2508.14104%2C2509.01619%2C2509.01720%2C2509.02322%2C2509.04198%2C2509.07764%2C2509.21823%2C2509.22256%2C2509.23263%2C2509.23866%2C2509.25047%2C2509.26539%2C2510.00413%2C2510.00536%2C2510.01524%2C2510.02204%2C2510.02250%2C2510.02418%2C2510.03204%2C2510.03612%2C2510.04039%2C2510.04373%2C2510.04607%2C2510.06223%2C2510.06607%2C2510.09872%2C2510.10666%2C2510.11035%2C2510.11221%2C2510.11892%2C2510.13543%2C2510.15455%2C2510.15863%2C2510.16252%2C2510.18113%2C2510.18314%2C2510.19205%2C2510.22732%2C2511.06101&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:32:52Z", "role": "primary", "sha256": "0b730323b2ce85c39aa7d534665901a480ff06540c73f1f80832fcc0735b8789", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?id_list=2310.08588%2C2402.09205%2C2406.05720%2C2407.21075%2C2410.07166%2C2503.03743%2C2508.04037%2C2508.10833%2C2508.11360%2C2508.13024%2C2508.13186%2C2508.14104%2C2509.01619%2C2509.01720%2C2509.02322%2C2509.04198%2C2509.07764%2C2509.21823%2C2509.22256%2C2509.23263%2C2509.23866%2C2509.25047%2C2509.26539%2C2510.00413%2C2510.00536%2C2510.01524%2C2510.02204%2C2510.02250%2C2510.02418%2C2510.03204%2C2510.03612%2C2510.04039%2C2510.04373%2C2510.04607%2C2510.06223%2C2510.06607%2C2510.09872%2C2510.10666%2C2510.11035%2C2510.11221%2C2510.11892%2C2510.13543%2C2510.15455%2C2510.15863%2C2510.16252%2C2510.18113%2C2510.18314%2C2510.19205%2C2510.22732%2C2511.06101&max_results=50"}], "id": "arxiv:2510.02418", "identifiers": {"arxiv": "2510.02418"}, "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/2510.02418"}], "name": "BrowserArena: Evaluating LLM Agents on Real-World Web Navigation Tasks", "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": "2025-10-02", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces BrowserArena, a live-web platform for head-to-head agent comparisons on user-submitted tasks. Step-level human annotations support targeted analysis of navigation, pop-up handling and CAPTCHA-related failures.", "topics": ["evaluation", "web-navigation", "human-in-the-loop"], "updated_at": "2025-10-07", "url": "https://arxiv.org/abs/2510.02418"}
{"aliases": [], "authors": ["Haitao Jia", "Ming He", "Zimo Yin", "Likang Wu", "Jianping Fan", "Jitao Sang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "9460a0926936d3db23c36c6c0e069d664265a8a9513c08c1a04f5a49970d6567", "submitted_at": "2025-10-09T09:22:05Z", "updated_at": "2025-10-09T09:22:05Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2510.07988v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2510.07988", "identifiers": {"arxiv": "2510.07988"}, "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/2510.07988"}], "name": "ReInAgent: A Context-Aware GUI Agent Enabling Human-in-the-Loop Mobile Task Navigation", "notes": "Reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2025-10-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces ReInAgent, coordinating information management, planning, and reflection around a shared memory for mobile tasks. It proactively asks users for information and checks consistency when instructions are ambiguous or change during execution.", "topics": ["human-in-the-loop", "memory", "mobile-navigation"], "updated_at": "2025-10-09", "url": "https://arxiv.org/abs/2510.07988"}
{"aliases": [], "authors": ["Hyunjun Kim", "Sooyoung Ryu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "7e445d0ca79f727a6c1d51e83381c3fd1d8e1fddeff7de99fe70ca06bc1bff49", "submitted_at": "2025-12-01T01:18:21Z", "updated_at": "2025-12-01T01:18:21Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2512.01174v1"}}, "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:2512.01174", "identifiers": {"arxiv": "2512.01174"}, "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/2512.01174"}], "name": "DrawingBench: Evaluating Spatial Reasoning and UI Interaction Capabilities of Large Language Models through Mouse-Based Drawing 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": "2025-12-01", "relations": [], "review_status": "curated", "scope": "core", "summary": "DrawingBench evaluates sequences of low-level mouse-based GUI actions through 250 drawing prompts and rule-based criteria. Action inspection and multi-turn external feedback support analysis of spatial reasoning, tool-state management, and refinement.", "topics": ["evaluation", "grounding", "human-in-the-loop", "state-tracking"], "updated_at": "2025-12-01", "url": "https://arxiv.org/abs/2512.01174"}
{"aliases": [], "authors": ["Jinzhe Tan", "Karim Benyekhlef"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d362a839fd1ee917ebdde6fcd794640523e930cd711769154532200903fe7de5", "submitted_at": "2025-11-28T07:26:29Z", "updated_at": "2025-11-28T07:26:29Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2512.04105v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2512.04105", "identifiers": {"arxiv": "2512.04105"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2512.04105"}], "name": "LegalWebAgent: Empowering Access to Justice via 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": "2025-11-28", "relations": [], "review_status": "curated", "scope": "core", "summary": "LegalWebAgent separates user-needs clarification, webpage browsing, and actions such as form completion or scheduling. A 15-task evaluation studies legal-service workflows relevant to Québec civil-law users, using HTML and screenshot observations.", "topics": ["industry", "web-navigation", "human-in-the-loop", "evaluation"], "updated_at": "2025-11-28", "url": "https://arxiv.org/abs/2512.04105"}
{"aliases": [], "authors": ["Hanzhang Zhou", "Xu Zhang", "Panrong Tong", "Jianan Zhang", "Liangyu Chen", "Quyu Kong", "Chenglin Cai", "Chen Liu", "Yue Wang", "Jingren Zhou", "Steven Hoi"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a89eb3fd173680c127245933d252d4e245b7179d77bca4b463d50800c78f7238", "submitted_at": "2025-12-26T14:51:52Z", "updated_at": "2025-12-26T14:51:52Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2512.22047v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2512.22047", "identifiers": {"arxiv": "2512.22047"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2512.22047"}], "name": "MAI-UI Technical Report: Real-World Centric Foundation GUI Agents", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-12-26", "relations": [], "review_status": "curated", "scope": "core", "summary": "MAI-UI combines GUI navigation, native user interaction, and MCP tool calls in a family of policies trained through generated data and online reinforcement learning. The report also studies device-cloud routing and scaling of parallel environments and interaction budgets, with grounding and Android navigation evaluation.", "topics": ["human-in-the-loop", "tool-integration", "reinforcement-learning", "data-generation", "efficiency"], "updated_at": "2025-12-26", "url": "https://arxiv.org/abs/2512.22047"}
{"aliases": [], "authors": ["Jihong Wang", "Jiamu Zhou", "Weiming Zhang", "Teng Wang", "Weiwen Liu", "Zhuosheng Zhang", "Xingyu Lou", "Weinan Zhang", "Huarong Deng", "Jun Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a2afe8d0d8d83f91003b7885635e098e19f69024306c43c3edb23c110e553f63", "submitted_at": "2026-01-12T07:08:42Z", "updated_at": "2026-04-23T09:34:51Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2601.07262v3"}}, "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:2601.07262", "identifiers": {"arxiv": "2601.07262"}, "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/2601.07262"}], "name": "ColorBrowserAgent: Complex Long-Horizon Browser Agent with Adaptive Knowledge Evolution", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-01-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "ColorBrowserAgent converts human feedback into reusable website knowledge and uses that knowledge to guide progressive context summaries. It evaluates the resulting memory and adaptation mechanisms on long browser workflows.", "topics": ["memory", "human-in-the-loop", "long-horizon", "web-navigation"], "updated_at": "2026-04-23", "url": "https://arxiv.org/abs/2601.07262"}
{"aliases": [], "authors": ["Simret Araya Gebreegziabher", "Yukun Yang", "Charles Chiang", "Hojun Yoo", "Chaoran Chen", "Hyo Jin Do", "Zahra Ashktorab", "Werner Geyer", "Diego Gómez-Zará", "Toby Jia-Jun Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "1bdc29648446cd79e0c0b9f72814457b55258d501283ce2fa2dedcd8262f885b", "submitted_at": "2026-01-22T22:42:40Z", "updated_at": "2026-01-22T22:42:40Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2601.16356v1"}}, "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:2601.16356", "identifiers": {"arxiv": "2601.16356"}, "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/2601.16356"}], "name": "The Behavioral Fabric of LLM-Powered GUI Agents: Human Values and Interaction Outcomes", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-01-22", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces a controlled web-task testbed for studying how stated preferences and human values influence GUI-agent decisions. It analyzes how personas alter navigation and how prominent interface cues can override the intended preference guidance.", "topics": ["evaluation", "ui-design", "human-in-the-loop"], "updated_at": "2026-01-22", "url": "https://arxiv.org/abs/2601.16356"}
{"aliases": [], "authors": ["Nan Chen", "Jing Lu", "Zilong Wang", "Luna K. Qiu", "Siming Chen", "Yuqing Yang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "ae1716f673e47cd0e61b5f2225ca4871dfd8522e35037fc8d226f9c7cd0e3e75", "submitted_at": "2026-01-26T03:02:26Z", "updated_at": "2026-01-26T03:02:26Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2601.18092v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2601.18092", "identifiers": {"arxiv": "2601.18092"}, "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/2601.18092"}], "name": "From Struggle to Success: Context-Aware Guidance for Screen Reader Users in Computer Use", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-01-26", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "Introduces AskEase, an on-demand assistant that combines contextual information to provide screen-reader-friendly computer-use guidance. A study with screen-reader users evaluates task completion and perceived workload while preserving human control of the interaction.", "topics": ["human-in-the-loop", "evaluation", "workflow"], "updated_at": "2026-01-26", "url": "https://arxiv.org/abs/2601.18092"}
{"aliases": [], "authors": ["Jing Wu", "Wenjie Ai", "Daphne Barretto", "Yiye Chen", "Qingyu Chen", "Yuhang He", "Pranit Chawla", "Nicholas Gydé", "Yanan Jian", "Vibhav Vineet"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "0139f7a1402d8dbfdecf65cc53c032495b2084f24cc38ff43841642c9102f29c", "submitted_at": "2026-01-28T14:35:23Z", "updated_at": "2026-08-26T22:31:44Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2601.20650v3"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2601.20650", "identifiers": {"arxiv": "2601.20650"}, "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/2601.20650"}], "name": "OS-Marathon: Benchmarking Computer-Use Agents on Vast-Horizon, Repetitive 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": "2026-01-28", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces OS-Marathon to evaluate long repetitive computer-use workflows whose execution grows with the amount of input data. It studies failure accumulation under task decomposition and GraphDemo, which adapts agents using a demonstration of the recurring workflow.", "topics": ["evaluation", "long-horizon", "human-in-the-loop"], "updated_at": "2026-08-26", "url": "https://arxiv.org/abs/2601.20650"}
{"aliases": [], "authors": ["Ruijia Cheng", "Jenny T. Liang", "Eldon Schoop", "Jeffrey Nichols"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d556ebbfbf4fc5875395b449a92b8ff27a7ed59d41bc062ac60fcf175f9064f8", "submitted_at": "2026-02-07T00:19:08Z", "updated_at": "2026-02-07T00:19:08Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2602.07283v1"}}, "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:2602.07283", "identifiers": {"arxiv": "2602.07283"}, "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/2602.07283"}], "name": "Mapping the Design Space of User Experience for Computer Use Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-02-07", "relations": [], "review_status": "curated", "scope": "core", "summary": "Develops a taxonomy of computer-use-agent interaction design through system review and practitioner interviews. A Wizard-of-Oz study then examines how users respond to normal, erroneous, and risky web-based execution, focusing on control, explanation, prompts, and mental models.", "topics": ["human-in-the-loop", "ui-design", "evaluation"], "updated_at": "2026-02-07", "url": "https://arxiv.org/abs/2602.07283"}
{"aliases": [], "authors": ["Ananya Gubbi Mohanbabu", "Rosiana Natalie", "Brandon Kim", "Anhong Guo", "Amy Pavel"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "c3606d090749737d4223a9f826400c2b2aebde245c623391f2a8c6702b2468f8", "submitted_at": "2026-02-10T01:00:55Z", "updated_at": "2026-02-10T01:00:55Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2602.09310v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2602.09310", "identifiers": {"arxiv": "2602.09310"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2602.09310"}], "name": "A11y-CUA Dataset: Characterizing the Accessibility Gap in Computer Use Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-02-10", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces A11y-CUA, a dataset comparing how blind or low-vision users and sighted users perform everyday computer tasks. It contrasts those traces with agent behavior under default, keyboard-only, and magnifier conditions to identify barriers to shared control.", "topics": ["human-in-the-loop", "evaluation", "data-generation"], "updated_at": "2026-02-10", "url": "https://arxiv.org/abs/2602.09310"}
{"aliases": [], "authors": ["Jiazheng Sun", "Mingxuan Li", "Yingying Zhang", "Jiayang Niu", "Yachen Wu", "Ruihan Jin", "Shuyu Lei", "Pengrongrui Tan", "Zongyu Zhang", "Ruoyi Wang", "Jiachen Yang", "Boyu Yang", "Jiacheng Liu", "Xin Peng"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2bbba09ca845f4fb772516d26d59fc3d99f8e87ba3657bb29c3c062b50925ec5", "submitted_at": "2026-02-12T09:25:15Z", "updated_at": "2026-02-12T09:25:15Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2602.11750v1"}}, "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:2602.11750", "identifiers": {"arxiv": "2602.11750"}, "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/2602.11750"}], "name": "AmbiBench: Benchmarking Mobile GUI Agents Beyond One-Shot Instructions in the Wild", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-02-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "AmbiBench evaluates mobile GUI agents under four levels of instruction clarity, including tasks requiring active clarification. Its MUSE evaluator separately audits outcomes, execution quality and interaction quality to study alignment with user intent.", "topics": ["evaluation", "human-in-the-loop", "mobile-navigation"], "updated_at": "2026-02-12", "url": "https://arxiv.org/abs/2602.11750"}
{"aliases": [], "authors": ["Faria Huq", "Zora Zhiruo Wang", "Zhanqiu Guo", "Venu Arvind Arangarajan", "Tianyue Ou", "Frank Xu", "Shuyan Zhou", "Graham Neubig", "Jeffrey P. Bigham"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "fe83d38ece7cd8f39d932611abaf7a170e7cb9b7c9744bd1176eb42e73be85b5", "submitted_at": "2026-02-19T18:11:28Z", "updated_at": "2026-07-07T23:37:46Z", "version": 4, "versioned_url": "https://arxiv.org/abs/2602.17588v4"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2602.17588", "identifiers": {"arxiv": "2602.17588"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2602.17588"}], "name": "Modeling Distinct Human Interaction in Web Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-02-19", "relations": [], "review_status": "curated", "scope": "core", "summary": "CowCorpus records interleaved human and agent actions during web navigation and identifies distinct patterns of user intervention. Models trained to anticipate these interventions are integrated into live web agents and assessed in a user study.", "topics": ["human-in-the-loop", "web-navigation", "data-generation", "action-prediction"], "updated_at": "2026-07-07", "url": "https://arxiv.org/abs/2602.17588"}
{"aliases": [], "authors": ["Yuxiang Chai", "Shunye Tang", "Han Xiao", "Rui Liu", "Hongsheng Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "7334e01b1fbf5cdd02c275c73d2a1b5c2931718d6019fcf6a0ea72b449704d94", "submitted_at": "2026-03-09T06:41:32Z", "updated_at": "2026-03-09T06:41:32Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2603.08013v1"}}, "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:2603.08013", "identifiers": {"arxiv": "2603.08013"}, "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/2603.08013"}], "name": "PIRA-Bench: A Transition from Reactive GUI Agents to GUI-based Proactive Intent Recommendation Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-03-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "PIRA-Bench evaluates proactive intent recommendations from noisy GUI screenshot sequences containing interleaved tasks and user context. Its PIRF baseline combines memory and state tracking to maintain multiple task threads and identify opportunities to assist.", "topics": ["evaluation", "screen-understanding", "memory", "state-tracking", "human-in-the-loop"], "updated_at": "2026-03-09", "url": "https://arxiv.org/abs/2603.08013"}
{"aliases": [], "authors": ["Alexandre De Masi"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "5944045ee4562641ee02612969d1c2d038e019fa02cc8d48c671c1d5397557ad", "submitted_at": "2026-03-11T11:24:31Z", "updated_at": "2026-03-11T11:24:31Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2603.10664v1"}}, "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:2603.10664", "identifiers": {"arxiv": "2603.10664"}, "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/2603.10664"}], "name": "Terminal Is All You Need: Design Properties for Human-AI Agent Collaboration", "notes": "Reviewed against the complete retained official abstract. Experimental findings and release availability have not been independently reproduced or verified. Adjacent conceptual HCI work about agent-facing interfaces; the abstract does not report GUI task experiments or a new evaluated CUA.", "published_at": "2026-03-11", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "This position paper derives interface design principles from terminal-based human-agent collaboration: representational compatibility, transparent actions and low participation barriers. It argues that graphical and other agent-facing modalities should provide these properties explicitly.", "topics": ["human-in-the-loop", "ui-design", "observability", "tool-integration"], "updated_at": "2026-03-11", "url": "https://arxiv.org/abs/2603.10664"}
{"aliases": [], "authors": ["Saelyne Yang", "Jaesang Yu", "Yi-Hao Peng", "Kevin Qinghong Lin", "Jae Won Cho", "Yale Song", "Juho Kim"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "9b3dd3efca196fa3dd168431102dbb5f0f763e7d9e54b3d709ad2362a9c6f154", "submitted_at": "2026-03-26T19:37:53Z", "updated_at": "2026-03-26T19:37:53Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2603.25864v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:32:00Z", "role": "primary", "sha256": "4d09ebfb093e43a4c44adef86c8fa8374f90d907fc33bd8e12a1315e62d5a583", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2603.12823%2C2603.14707%2C2603.15039%2C2603.17357%2C2603.18429%2C2603.19191%2C2603.24440%2C2603.25226%2C2603.25864%2C2603.26211%2C2603.26648%2C2603.29318%2C2604.01535%2C2604.02947%2C2604.04399%2C2604.05477%2C2604.06126%2C2604.06182%2C2604.06367%2C2604.07429%2C2604.08455%2C2604.08516%2C2604.08523%2C2604.09155%2C2604.09574%2C2604.09587%2C2604.09815%2C2604.09937%2C2604.11259%2C2604.11784%2C2604.12284%2C2604.13019%2C2604.13822%2C2604.14113%2C2604.14262%2C2604.14956%2C2604.15093%2C2604.15309%2C2604.17817%2C2604.17821%2C2604.18860%2C2604.24964%2C2604.25380%2C2604.25562%2C2604.27151%2C2604.27253%2C2604.27776%2C2605.00642%2C2605.06393%2C2605.06664&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:32:00Z", "role": "primary", "sha256": "4d09ebfb093e43a4c44adef86c8fa8374f90d907fc33bd8e12a1315e62d5a583", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "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:2603.25864", "identifiers": {"arxiv": "2603.25864"}, "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/2603.25864"}], "name": "GUIDE: A Benchmark for Understanding and Assisting Users in Open-Ended 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": "2026-03-26", "relations": [], "review_status": "curated", "scope": "core", "summary": "GUIDE evaluates whether models can recognize a user’s behavior, infer intent and decide when to offer help during open-ended software tasks. It uses narrated screen recordings of novice users and studies the effect of additional user context on these assistance decisions.", "topics": ["evaluation", "human-in-the-loop", "video", "screen-understanding"], "updated_at": "2026-03-26", "url": "https://arxiv.org/abs/2603.25864"}
{"aliases": [], "authors": ["Chaoran Chen", "Zhiping Zhang", "Zeya Chen", "Eryue Xu", "Yinuo Yang", "Ibrahim Khalilov", "Simret A Gebreegziabher", "Yanfang Ye", "Ziang Xiao", "Yaxing Yao", "Tianshi Li", "Toby Jia-Jun Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "1411ee71ad8f65e4941af4ef538f52da6e9bf9c57fb537b6ecf37c8478609693", "submitted_at": "2026-04-06T17:57:16Z", "updated_at": "2026-09-12T06:05:52Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2604.04918v2"}}, "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:2604.04918", "identifiers": {"arxiv": "2604.04918"}, "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/2604.04918"}], "name": "Comparing Human Oversight Strategies 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": "Compares four human oversight strategies during live web-agent sessions, examining exposure to problematic actions and opportunities to intervene. The analysis identifies timing and attention problems in control handoffs and upfront plan approval.", "topics": ["human-in-the-loop", "safety-security", "evaluation"], "updated_at": "2026-09-12", "url": "https://arxiv.org/abs/2604.04918"}
{"aliases": [], "authors": ["Pan Hao", "Rishi Selvakumaran", "Jacob Sun", "Qianwen Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "bd3b5e5b9a607023c82150673255eaf2f90c22db86d3c049b2a2fd536aaef17f", "submitted_at": "2026-04-16T06:22:48Z", "updated_at": "2026-04-16T06:22:48Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2604.14668v1"}}, "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:2604.14668", "identifiers": {"arxiv": "2604.14668"}, "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/2604.14668"}], "name": "Beyond Chat and Clicks: GUI Agents for In-Situ Assistance via Live Interface Transformation", "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": "2026-04-16", "relations": [], "review_status": "curated", "scope": "core", "summary": "DOMSteer translates help requests and live page context into reversible DOM changes such as control highlighting, contextual tooltips and layout reorganization. The work evaluates in-place GUI assistance with interface tests, use cases and a comparative user study.", "topics": ["human-in-the-loop", "ui-design", "grounding", "evaluation"], "updated_at": "2026-04-16", "url": "https://arxiv.org/abs/2604.14668"}
{"aliases": [], "authors": ["Haomin Zhuang", "Hanwen Xing", "Xiangliang Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "7b1788e6fbb4c14a6d993ce329bf7bf05e3a1e6ffbfc7c14a9fa9d36fbf7ef16", "submitted_at": "2026-04-15T18:22:50Z", "updated_at": "2026-04-15T18:22:50Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2604.16520v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T12:31:19Z", "role": "primary", "sha256": "05ec5d668ac75d3d7c5b71ea7d6e76a3fbf5ebeb742e0843db2329e5e41aa208", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2604.03598%2C2604.04759%2C2604.05547%2C2604.06170%2C2604.07340%2C2604.10547%2C2604.10938%2C2604.11978%2C2604.12282%2C2604.13018%2C2604.13346%2C2604.13954%2C2604.15579%2C2604.15671%2C2604.15715%2C2604.16520%2C2604.18240%2C2604.18934%2C2604.19572%2C2604.19667%2C2604.19742%2C2604.20841%2C2604.20987%2C2604.22436%2C2604.23781%2C2604.24479%2C2604.26148%2C2604.27621%2C2604.27996%2C2604.28093%2C2604.28139%2C2604.28181%2C2605.00425%2C2605.01462%2C2605.02240%2C2605.03042%2C2605.03242%2C2605.04811%2C2605.05191%2C2605.05811%2C2605.06110%2C2605.07073%2C2605.07358%2C2605.08374%2C2605.08985%2C2605.10912%2C2605.10966%2C2605.11363%2C2605.11732%2C2605.11817&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:31:19Z", "role": "primary", "sha256": "05ec5d668ac75d3d7c5b71ea7d6e76a3fbf5ebeb742e0843db2329e5e41aa208", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2604.03598%2C2604.04759%2C2604.05547%2C2604.06170%2C2604.07340%2C2604.10547%2C2604.10938%2C2604.11978%2C2604.12282%2C2604.13018%2C2604.13346%2C2604.13954%2C2604.15579%2C2604.15671%2C2604.15715%2C2604.16520%2C2604.18240%2C2604.18934%2C2604.19572%2C2604.19667%2C2604.19742%2C2604.20841%2C2604.20987%2C2604.22436%2C2604.23781%2C2604.24479%2C2604.26148%2C2604.27621%2C2604.27996%2C2604.28093%2C2604.28139%2C2604.28181%2C2605.00425%2C2605.01462%2C2605.02240%2C2605.03042%2C2605.03242%2C2605.04811%2C2605.05191%2C2605.05811%2C2605.06110%2C2605.07073%2C2605.07358%2C2605.08374%2C2605.08985%2C2605.10912%2C2605.10966%2C2605.11363%2C2605.11732%2C2605.11817&max_results=50"}], "id": "arxiv:2604.16520", "identifiers": {"arxiv": "2604.16520"}, "kind": "paper", "last_verified_at": "2026-10-17T12:31:19Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2604.16520"}], "name": "AgentClick: A Skill-Based Human-in-the-Loop Review Layer for Terminal AI Agents", "notes": "Reviewed against the complete retained official abstract. Experimental findings and release availability have not been independently reproduced or verified. Adjacent supervision interface: the browser UI is for human review of terminal agents; the abstract does not claim that those agents operate graphical interfaces.", "published_at": "2026-04-15", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "AgentClick gives terminal-based agents a browser review interface for plans, drafts, execution traces and editable memory. A local server and skill plugin let users inspect proposed actions and provide structured feedback.", "topics": ["human-in-the-loop", "observability", "workflow", "memory", "tool-integration"], "updated_at": "2026-04-15", "url": "https://arxiv.org/abs/2604.16520"}
{"aliases": [], "authors": ["Yanqiu Zhao", "Dongying Zheng", "Kaibo Huang", "Yukun Wei", "Zhongliang Yang", "Linna Zhou"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6eb9b67f7e51edecfbe57af47b6bad445c05129c11c377d32ea84369b430f895", "submitted_at": "2026-05-27T15:51:22Z", "updated_at": "2026-05-29T04:03:46Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2605.28646v2"}}, "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:2605.28646", "identifiers": {"arxiv": "2605.28646"}, "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/2605.28646"}], "name": "MaskClaw: Edge-Side Personalized Privacy Arbitration for GUI Agents with Behavior-Driven Skill Evolution", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-05-27", "relations": [], "review_status": "curated", "scope": "core", "summary": "MaskClaw makes local Allow, Mask or Ask decisions before GUI screenshots leave a trusted environment, using visual evidence and personalized policy memory. Its benchmark and designed evolution scenarios test how corrections and edits can become privacy skills checked by a sandbox gate.", "topics": ["safety-security", "memory", "human-in-the-loop", "evaluation"], "updated_at": "2026-05-29", "url": "https://arxiv.org/abs/2605.28646"}
{"aliases": [], "authors": ["Wenkai Wang", "Tao Xiong", "Jingchen Ni", "Yunpeng Bao", "Xiyun Li", "Tianqi Liu", "Hongcan Guo", "Zilong Huang", "Shengyu Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "bf0fcafeb2d40e489ec80784a3541114b02c6a9077a330265d5c49775eca4582", "submitted_at": "2026-06-02T03:42:34Z", "updated_at": "2026-06-02T03:42:34Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.03103v1"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T12:31:23Z", "role": "primary", "sha256": "9a71e95803ce58c462c84e2b45a891555e31d1fb409e643bb3ca4675a44c96e6", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2605.12233%2C2605.12493%2C2605.12978%2C2605.13119%2C2605.14133%2C2605.16909%2C2605.17260%2C2605.17324%2C2605.17448%2C2605.17656%2C2605.17721%2C2605.19447%2C2605.19587%2C2605.20894%2C2605.22642%2C2605.23204%2C2605.25233%2C2605.25901%2C2605.26086%2C2605.26144%2C2605.26252%2C2605.26646%2C2605.27209%2C2605.27705%2C2605.27898%2C2605.28158%2C2605.28255%2C2605.29341%2C2605.29463%2C2605.29801%2C2605.30010%2C2605.30052%2C2605.30126%2C2605.30434%2C2605.30907%2C2605.30914%2C2605.31264%2C2605.31557%2C2606.00530%2C2606.00543%2C2606.00856%2C2606.01057%2C2606.01961%2C2606.02540%2C2606.03103%2C2606.03889%2C2606.03895%2C2606.04315%2C2606.04455%2C2606.05525&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:31:23Z", "role": "primary", "sha256": "9a71e95803ce58c462c84e2b45a891555e31d1fb409e643bb3ca4675a44c96e6", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2605.12233%2C2605.12493%2C2605.12978%2C2605.13119%2C2605.14133%2C2605.16909%2C2605.17260%2C2605.17324%2C2605.17448%2C2605.17656%2C2605.17721%2C2605.19447%2C2605.19587%2C2605.20894%2C2605.22642%2C2605.23204%2C2605.25233%2C2605.25901%2C2605.26086%2C2605.26144%2C2605.26252%2C2605.26646%2C2605.27209%2C2605.27705%2C2605.27898%2C2605.28158%2C2605.28255%2C2605.29341%2C2605.29463%2C2605.29801%2C2605.30010%2C2605.30052%2C2605.30126%2C2605.30434%2C2605.30907%2C2605.30914%2C2605.31264%2C2605.31557%2C2606.00530%2C2606.00543%2C2606.00856%2C2606.01057%2C2606.01961%2C2606.02540%2C2606.03103%2C2606.03889%2C2606.03895%2C2606.04315%2C2606.04455%2C2606.05525&max_results=50"}], "id": "arxiv:2606.03103", "identifiers": {"arxiv": "2606.03103"}, "kind": "paper", "last_verified_at": "2026-10-17T12:31:23Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2606.03103"}], "name": "DeskCraft: Benchmarking Desktop Agents on Professional Workflows and Human-in-the-Loop Collaboration", "notes": "Reviewed against the complete retained official abstract. Experimental findings and release availability have not been independently reproduced or verified.", "published_at": "2026-06-02", "relations": [], "review_status": "curated", "scope": "core", "summary": "DeskCraft evaluates professional desktop workflows and human-agent collaboration through mid-execution clarification, interruption and post-completion feedback. Its tasks cover creative applications and expose difficulties in long-horizon execution and proactive clarification.", "topics": ["evaluation", "human-in-the-loop", "long-horizon", "workflow"], "updated_at": "2026-06-02", "url": "https://arxiv.org/abs/2606.03103"}
{"aliases": [], "authors": ["Bo Zhang", "Borui Zhang", "Chenghao Jiang", "Minglei Shi", "Xiaofeng Wang", "Zheng Zhu", "Jie Zhou", "Jiwen Lu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "33c343e9bc02976fc3e32057cdcdebc3abeae5ef049b877443e81917a8abc550", "submitted_at": "2026-05-28T17:59:31Z", "updated_at": "2026-05-28T17:59:31Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.07594v1"}}, "environments": ["desktop", "macos"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T12:31:28Z", "role": "primary", "sha256": "7eaa9a5430d3a45049e612a117da345c1daafc54277a35a7cf56b42fbb1e8618", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2606.05646%2C2606.05670%2C2606.06556%2C2606.06891%2C2606.07367%2C2606.07594%2C2606.07682%2C2606.07723%2C2606.07867%2C2606.10394%2C2606.10525%2C2606.11290%2C2606.12203%2C2606.12736%2C2606.13040%2C2606.13368%2C2606.13578%2C2606.13669%2C2606.13802%2C2606.13898%2C2606.14397%2C2606.15057%2C2606.15300%2C2606.15441%2C2606.17162%2C2606.18425%2C2606.19336%2C2606.19409%2C2606.19704%2C2606.20950%2C2606.21033%2C2606.21565%2C2606.23521%2C2606.23654%2C2606.24775%2C2606.25084%2C2606.26080%2C2606.27406%2C2606.27634%2C2606.28430%2C2606.28480%2C2606.29955%2C2606.30616%2C2607.00692%2C2607.01084%2C2607.02032%2C2607.02448%2C2607.02490%2C2607.02577%2C2607.02911&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:31:28Z", "role": "primary", "sha256": "7eaa9a5430d3a45049e612a117da345c1daafc54277a35a7cf56b42fbb1e8618", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2606.05646%2C2606.05670%2C2606.06556%2C2606.06891%2C2606.07367%2C2606.07594%2C2606.07682%2C2606.07723%2C2606.07867%2C2606.10394%2C2606.10525%2C2606.11290%2C2606.12203%2C2606.12736%2C2606.13040%2C2606.13368%2C2606.13578%2C2606.13669%2C2606.13802%2C2606.13898%2C2606.14397%2C2606.15057%2C2606.15300%2C2606.15441%2C2606.17162%2C2606.18425%2C2606.19336%2C2606.19409%2C2606.19704%2C2606.20950%2C2606.21033%2C2606.21565%2C2606.23521%2C2606.23654%2C2606.24775%2C2606.25084%2C2606.26080%2C2606.27406%2C2606.27634%2C2606.28430%2C2606.28480%2C2606.29955%2C2606.30616%2C2607.00692%2C2607.01084%2C2607.02032%2C2607.02448%2C2607.02490%2C2607.02577%2C2607.02911&max_results=50"}], "id": "arxiv:2606.07594", "identifiers": {"arxiv": "2606.07594"}, "kind": "paper", "last_verified_at": "2026-10-17T12:31:28Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2606.07594"}], "name": "Syll: Open-Source Personal Automation with Cross-Surface Execution", "notes": "Reviewed against the complete retained official abstract. Experimental findings and release availability have not been independently reproduced or verified.", "published_at": "2026-05-28", "relations": [], "review_status": "curated", "scope": "core", "summary": "Syll combines tool calls, shell execution and visual GUI control in a personal-automation runtime. It turns user demonstrations into reusable skills and exposes execution logs, keyframes and approval checkpoints for inspection.", "topics": ["automation", "human-in-the-loop", "workflow", "memory", "recording", "observability"], "updated_at": "2026-05-28", "url": "https://arxiv.org/abs/2606.07594"}
{"aliases": [], "authors": ["Yujin Zhang", "Daye Nam"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "c52695e860206d02735839853416965ec3d42fac134ebfa0e8f2f5f5ea043914", "submitted_at": "2026-06-17T04:13:20Z", "updated_at": "2026-06-17T04:13:20Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.18671v1"}}, "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:2606.18671", "identifiers": {"arxiv": "2606.18671"}, "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/2606.18671"}], "name": "HANSEL: Extracting Breadcrumbs from Web Agent Trajectories for Interactive Verification", "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": "HANSEL extracts evidence pages and snippets from browser-agent traces and preserves relevant page state in interactive verification views. It flags untraceable answers and evaluates evidence selection and human verification effort separately.", "topics": ["human-in-the-loop", "observability", "recording", "evaluation"], "updated_at": "2026-06-17", "url": "https://arxiv.org/abs/2606.18671"}
{"aliases": [], "authors": ["Aradhana Nayak", "Mussadiq Nazeer", "Wang Peng", "Feng Liu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "3c5a2d0e461b011e2d5b9dbd73012d007261d8df37bd1d687fa3bd86dc6f2246", "submitted_at": "2026-06-24T11:18:01Z", "updated_at": "2026-06-24T11:18:01Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.25705v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2606.25705", "identifiers": {"arxiv": "2606.25705"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2606.25705"}], "name": "GUI agent: Guided Exploration of User-Sensitive Screens", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-06-24", "relations": [], "review_status": "curated", "scope": "core", "summary": "This short paper develops an explorer agent that expands from a demonstrated GUI task to discover requests likely to reach user-sensitive screens. The proposed exploration supports identifying situations where task execution should be handed back to the user.", "topics": ["safety-security", "human-in-the-loop", "data-generation"], "updated_at": "2026-06-24", "url": "https://arxiv.org/abs/2606.25705"}
{"aliases": [], "authors": ["Meng Chen", "Anya Ji", "Tsung-Han Wu", "Tobias Maringgele", "David M. Chan", "Alane Suhr", "Amy Pavel"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a0a3e641e0a71051f5fbd20f57af3db06f2b65c61025e78adcdd122f773d034b", "submitted_at": "2026-06-30T17:18:47Z", "updated_at": "2026-09-01T04:17:49Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2606.31980v2"}}, "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:2606.31980", "identifiers": {"arxiv": "2606.31980"}, "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/2606.31980"}], "name": "DigitalCoach: Communication and Grounding Gaps in Human and Agentic Computer Use Coaching", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-06-30", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "DigitalCoach records expert–novice software coaching conversations aligned with screen and input events. Automated and interactive evaluations examine how model coaches differ from humans in explanation, error diagnosis and visual grounding when helping a learner operate software.", "topics": ["human-in-the-loop", "video", "evaluation", "grounding"], "updated_at": "2026-09-01", "url": "https://arxiv.org/abs/2606.31980"}
{"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": ["Ruei-Che Chang", "Wenqian Xu", "Dingzeyu Li", "Bryan Wang", "Anhong Guo"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2069acf693e0dced49d0e10e21bf9fe28b99f50f1b932837f16137e0d3170a80", "submitted_at": "2026-07-20T04:03:38Z", "updated_at": "2026-07-20T04:03:38Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2607.17527v1"}}, "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:2607.17527", "identifiers": {"arxiv": "2607.17527"}, "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/2607.17527"}], "name": "Sidekick: Designing Communication for Effective Multitasking with Computer Use Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-07-20", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces Sidekick, a communication interface that uses ambient status cues, multimodal action summaries, and visual and verbal explanations of reasoning to help people monitor computer-use agents. A user study examines progress awareness, error traceability, and multitasking across background and foreground interaction.", "topics": ["human-in-the-loop", "observability", "long-horizon"], "updated_at": "2026-07-20", "url": "https://arxiv.org/abs/2607.17527"}
{"aliases": [], "authors": ["Santosh Patapati"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d5ccb0d41798a1a3236e6f9b4361814cccdd40facf9b43bf0bf52932337484dc", "submitted_at": "2026-07-01T16:24:07Z", "updated_at": "2026-07-01T16:24:07Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2608.09944v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-18T10:00:45.064380+00:00", "role": "primary", "sha256": "f26fba818a44c6a7f541ab3b0341ca8f4798d837a425a2ac9fb0fc47a553b7bc", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+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=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:00:45.064380+00:00", "role": "primary", "sha256": "f26fba818a44c6a7f541ab3b0341ca8f4798d837a425a2ac9fb0fc47a553b7bc", "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=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:05:26.923383+00:00", "role": "primary", "sha256": "777197864eeaa34d180bf6b91f32408b7b3d8cc5c9faecc95eca276f5bd426de", "supports": ["summary", "scope", "environments", "notes"], "url": "https://arxiv.org/html/2608.09944v1"}], "id": "arxiv:2608.09944", "identifiers": {"arxiv": "2608.09944"}, "kind": "paper", "last_verified_at": "2026-10-18T10:00:45.064380+00:00", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2608.09944"}], "name": "Navigation Alone Is Not Enough: Evaluating Explanatory Assistive UI Agents", "notes": "Packaged tasks use recorded linear transition graphs. Reported model experiments use a text-input validation subset, so their results are not evidence of screenshot-based navigation performance. No results were independently reproduced. Basis: complete retained abstract and source-reviewed sections of the exact-version paper (3.3 Interface states and task annotations; 3.4 Evaluation; 4.1 Models and experimental setup; 5 Discussion); not an entire-paper reading or reproduction claim.", "published_at": "2026-07-01", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces NeXUI, an assistive UI-agent benchmark that evaluates navigation together with interface-grounded explanations and user oversight in recorded interface states.", "topics": ["evaluation", "human-in-the-loop", "web-navigation"], "updated_at": "2026-07-01", "url": "https://arxiv.org/abs/2608.09944"}
{"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": ["Joy Chen", "Alejandro Castillejo Munoz", "Pierluca D'Oro", "Yuxuan Sun", "Chloe Evans", "Joseph Tighe"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "50c26a7ab741ecb3cb855305ee4f7d5b5729296248d650c469600aeb73101781", "submitted_at": "2026-08-25T23:21:57Z", "updated_at": "2026-09-15T04:43:00Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2608.26204v2"}}, "environments": ["desktop", "cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2608.26204", "identifiers": {"arxiv": "2608.26204"}, "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/2608.26204"}], "name": "ADeptS-Bench: Measuring the Trustworthiness of Computer Use Agents Across Devices", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-08-25", "relations": [], "review_status": "curated", "scope": "core", "summary": "Evaluates interface-embedded threats and whether computer-use agents seek clarification when mobile or desktop instructions are ambiguous.", "topics": ["safety-security", "evaluation", "human-in-the-loop"], "updated_at": "2026-09-15", "url": "https://arxiv.org/abs/2608.26204"}
{"aliases": [], "authors": ["Satwik Ram Kodandaram", "Monalika Padma Reddy", "Xiaojun Bi", "Jiawei Zhou", "I. V. Ramakrishnan", "Vikas Ashok"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "deb130d93bbbd6dfdc04c72d232a9dcfa1e92b0586831e5864070e7fe0de14dd", "submitted_at": "2026-09-01T00:44:19Z", "updated_at": "2026-09-01T00:44:19Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.00524v1"}}, "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"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%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:2609.00524", "identifiers": {"arxiv": "2609.00524"}, "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/2609.00524"}], "name": "Are We There Yet? Assessing Computer-Use Agents for Blind Users' Accessible Interaction with Desktop 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": "2026-09-01", "relations": [], "review_status": "curated", "scope": "core", "summary": "A diary study deploys the screen-reader-accessible OLLA computer-use prototype with blind users across desktop applications. Recorded commands and interaction traces support model comparisons and analysis of grounding, planning, constraint tracking and termination failures.", "topics": ["evaluation", "human-in-the-loop", "grounding"], "updated_at": "2026-09-01", "url": "https://arxiv.org/abs/2609.00524"}
{"aliases": [], "authors": ["Yixiong Xiao", "Lang An", "Hucheng Yang", "Pinxue Ma", "Yongquan Chen", "Jingjia Cao", "Yusai Zhao", "Ting Wang", "Ting Liu", "Siqi Bao", "Jingbo Zhou", "Hua Wu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "52849171f507e126093411ba81dc580d460b84f831358b08bd57cb54ca41b665", "submitted_at": "2026-09-02T06:03:18Z", "updated_at": "2026-09-10T15:52:42Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2609.02149v3"}}, "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:2609.02149", "identifiers": {"arxiv": "2609.02149"}, "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/2609.02149"}], "name": "OmegaUse-SOP: SOP Engineering for Professional Computer Use from Human Demonstrations", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-09-02", "relations": [], "review_status": "curated", "scope": "core", "summary": "OmegaUse-SOP converts expert GUI demonstrations into configurable procedures and refines them with domain rules and human input. It records, abstracts, grounds and verifies live execution, with a case study of photovoltaic simulation workflows in PVsyst 7.2.", "topics": ["human-in-the-loop", "recording", "workflow", "automation"], "updated_at": "2026-09-10", "url": "https://arxiv.org/abs/2609.02149"}
{"aliases": [], "authors": ["Weide Zhan", "Qumu Shaqu", "Yuanqing Liu", "Peng Zhang", "Jiahao Liu", "Kam Him Lam", "Ning Gu", "Zhan Hu", "Tun Lu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "61587798aefd8e9592ddaef756a13e02f0e4a987e4068ef582d0aaf66e8eb680", "submitted_at": "2026-09-04T08:09:47Z", "updated_at": "2026-09-04T08:09:47Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.04850v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:13Z", "role": "primary", "sha256": "ca895b2f0dcd74816fb5e775a7161a0932ff7120a72477791f5e4c0bb1ead88e", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2608.17319%2C2608.18307%2C2608.20319%2C2608.20707%2C2608.20797%2C2608.21794%2C2608.21830%2C2608.21832%2C2608.21898%2C2608.22077%2C2608.22533%2C2608.22577%2C2608.22847%2C2608.24015%2C2608.24017%2C2608.24099%2C2608.24174%2C2608.24680%2C2608.24848%2C2608.25777%2C2608.26200%2C2608.26204%2C2608.27477%2C2608.27508%2C2608.27808%2C2608.27866%2C2608.29305%2C2608.30038%2C2608.30207%2C2609.00028%2C2609.00048%2C2609.00524%2C2609.02057%2C2609.02149%2C2609.02154%2C2609.02309%2C2609.02885%2C2609.03438%2C2609.04850%2C2609.04869%2C2609.05374%2C2609.05511%2C2609.05513%2C2609.05535%2C2609.06530%2C2609.06792%2C2609.07603%2C2609.07712%2C2609.09212%2C2609.10031&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:13Z", "role": "primary", "sha256": "ca895b2f0dcd74816fb5e775a7161a0932ff7120a72477791f5e4c0bb1ead88e", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2608.17319%2C2608.18307%2C2608.20319%2C2608.20707%2C2608.20797%2C2608.21794%2C2608.21830%2C2608.21832%2C2608.21898%2C2608.22077%2C2608.22533%2C2608.22577%2C2608.22847%2C2608.24015%2C2608.24017%2C2608.24099%2C2608.24174%2C2608.24680%2C2608.24848%2C2608.25777%2C2608.26200%2C2608.26204%2C2608.27477%2C2608.27508%2C2608.27808%2C2608.27866%2C2608.29305%2C2608.30038%2C2608.30207%2C2609.00028%2C2609.00048%2C2609.00524%2C2609.02057%2C2609.02149%2C2609.02154%2C2609.02309%2C2609.02885%2C2609.03438%2C2609.04850%2C2609.04869%2C2609.05374%2C2609.05511%2C2609.05513%2C2609.05535%2C2609.06530%2C2609.06792%2C2609.07603%2C2609.07712%2C2609.09212%2C2609.10031&max_results=50"}], "id": "arxiv:2609.04850", "identifiers": {"arxiv": "2609.04850"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:13Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.04850"}], "name": "ElderBench: Benchmarking Autonomous Mobile Agents for Older Adults", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-09-04", "relations": [], "review_status": "curated", "scope": "core", "summary": "ElderBench evaluates mobile GUI agents on smartphone requests elicited from older adults, including indirect and underspecified language. Online and offline tests, instruction normalization and linguistic analyses examine how these request patterns contribute to failures.", "topics": ["evaluation", "human-in-the-loop", "mobile-navigation", "robustness"], "updated_at": "2026-09-04", "url": "https://arxiv.org/abs/2609.04850"}
{"aliases": [], "authors": ["Kratika Bhagtani", "Kusha Sridhar", "Maziyar Baran Pouyan", "Yuying Zhao", "Eugene Siow"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "ab0dbea71d0b8bfe035d3b0a71d6ed99fbd261202976cdc322a0769074827d76", "submitted_at": "2026-09-15T22:23:17Z", "updated_at": "2026-09-23T01:31:01Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2609.17885v2"}}, "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"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%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:2609.17885", "identifiers": {"arxiv": "2609.17885"}, "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/2609.17885"}], "name": "ERPBench: A State-Grounded Evaluation Paradigm for Computer-Use Agents in Enterprise Software", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-09-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "ERPBench evaluates screenshot-only agents in enterprise software using database values to check the business records produced by their actions. It pairs these state-based checks with a harness for human approval and analyzes cases where visible completion masks incorrect persisted data.", "topics": ["evaluation", "workflow", "human-in-the-loop", "robustness"], "updated_at": "2026-09-23", "url": "https://arxiv.org/abs/2609.17885"}
{"aliases": [], "authors": ["Wenkui Yang", "Yiming Hao", "Kai Zhou", "Chengyu Pu", "Yuzhe Zhang", "Jie Cao", "Jin Huang", "Shuai Ma", "Feng Tian"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "602f4eb7aacc05748b83b9c32b28439098005c77eafb7624804f820995521a79", "submitted_at": "2026-10-14T12:35:59Z", "updated_at": "2026-10-14T12:35:59Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.20100v1"}}, "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:2610.20100", "identifiers": {"arxiv": "2610.20100"}, "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/2610.20100"}], "name": "Task-State Governance: A Virtual Machine for Verifiable and Recoverable GUI Agent Execution", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-10-14", "relations": [], "review_status": "curated", "scope": "core", "summary": "Presents TaskVM, an interface for specifying desired and preserved application state, checking agent actions, and exposing whether changes can be reversed or compensated.", "topics": ["human-in-the-loop", "safety-security", "workflow"], "updated_at": "2026-10-14", "url": "https://arxiv.org/abs/2610.20100"}
{"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": ["Haoming Li", "Defu Cao", "Xu Hu", "Yingtao Luo", "Yiyang Wang", "Xianghe Xu", "Jiashu He", "Wanting Jiang", "Yifan Zhang", "Yunhan Zhao", "Siyang Wu", "Chen Yang", "Haixin Wang", "Zheyuan Deng", "Taoran Li", "Jingwen Zhang", "Zhejian Peng", "Shushu Liang", "Yue Zhao", "Bingzhe Li", "Feng Chen", "Yifan Wang", "Jintao Huang", "Yijun Wang", "Yuexing Hao", "Xiaomin Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "f613ef7ada260b494a411058bc733c7a5e0b074cdb01906c38132acf7275b9b1", "submitted_at": "2026-10-16T06:14:55Z", "updated_at": "2026-10-16T06:14:55Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.22990v1"}}, "environments": ["web"], "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": "2026-10-20T10:42:06.870916+00:00", "role": "primary", "sha256": "bf9066e8b3472100f120e21534f92655f4d0814f2521b9d23d778e5239c7a2a4", "supports": ["summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/html/2610.22990v1"}], "id": "arxiv:2610.22990", "identifiers": {"arxiv": "2610.22990"}, "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/2610.22990"}], "name": "PersonaCUABench: Benchmarking Computer-Use Agents Serving Diverse Personas", "notes": "Reviewed the complete retained abstract and selected exact-v1 construction, evaluation and limitations sections. Main interactions and satisfaction ratings use LLM-simulated personas; human annotators validate selected outputs. The harness observes page text, with screenshots only for auditing. Results are not evidence about real demographic populations or desktop agents. No independent reproduction.", "published_at": "2026-10-16", "relations": [], "review_status": "curated", "scope": "core", "summary": "PersonaCUABench varies simulated user requests and clarification dialogue over fixed web tasks, comparing task completion with requirement handling and persona-agent satisfaction ratings.", "topics": ["evaluation", "web-navigation", "human-in-the-loop"], "updated_at": "2026-10-16", "url": "https://arxiv.org/abs/2610.22990"}
{"aliases": [], "authors": ["Egor Zverev", "Juhee Kim", "Leon Trampert", "Boris Köpf", "Aashish Kolluri", "Shruti Tople", "Santiago Zanella-Béguelin"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e675a3082f571da547fb5b6cb89e4f68a034bfa57236886e1218cb2bb2cdf1c3", "submitted_at": "2026-10-16T22:12:45Z", "updated_at": "2026-10-16T22:12:45Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.24277v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-20T05:52:00Z", "role": "primary", "sha256": "af310824eea97aac8209023ff53bad7bea77778816586b10bb94110148347af9", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agent%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B202610060000+TO+202610200000%5D&start=0&max_results=200&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T08:27:26.236797+00:00", "role": "primary", "sha256": "80bed80786f939973d0d13b4f892300fee02a6110e326247131fc83d257796de", "supports": ["name", "kind", "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": "2026-10-20T09:17:33.132448+00:00", "role": "primary", "sha256": "daceefecb2e6b910a6c7f2ffe35e89c7070f6282f4f35ee89e885a2e77aa94c4", "supports": ["summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/html/2610.24277v1"}], "id": "arxiv:2610.24277", "identifiers": {"arxiv": "2610.24277"}, "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/2610.24277"}], "name": "SPIDES: Information-Flow Control for Browser Agents", "notes": "Reviewed the complete abstract and selected v1 threat-model, policy, evaluation and limitations sections. Guarantees assume correct labels/trusted origins and a trusted browser/harness mediating actions; evaluation uses idealized approvals. Origin-level reader sets can miss recipient distinctions within one origin. No unrestricted prompt-injection immunity or independent reproduction is claimed.", "published_at": "2026-10-16", "relations": [], "review_status": "curated", "scope": "core", "summary": "SPIDES propagates HTML integrity and confidentiality labels into agent context and mediates browser actions with information-flow policies and approval gates, evaluated in the labeled MicroNet ecosystem.", "topics": ["safety-security", "web-navigation", "human-in-the-loop", "evaluation"], "updated_at": "2026-10-16", "url": "https://arxiv.org/abs/2610.24277"}
{"aliases": [], "authors": [], "details": {}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T11:09:28Z", "role": "primary", "sha256": "c4da2b089f97296ac6c7b3ef7273db57b650d12cf3595f6f6ea74c901e5277b8", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/alipay/mobile-agent/HEAD/README.md"}], "id": "github:alipay/mobile-agent", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-16T11:09:28Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/alipay/mobile-agent"}], "name": "MobileAgent (Alipay)", "notes": "Distinct from the X-PLUG/MobileAgent project despite the similar name.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/alipay/mobile-agent/HEAD/README.md", "target_id": "arxiv:2401.04124", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Mobile-control research implementation combining operating-procedure context with human interaction for personalized task execution.", "topics": ["mobile-navigation", "workflow", "human-in-the-loop"], "updated_at": null, "url": "https://github.com/alipay/mobile-agent"}
{"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": ["Magentic UI", "Magentic-UI", "MagenticLite"], "authors": [], "details": {}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:02Z", "role": "primary", "sha256": "240e0c92b86c02d3790c0784d8c72193ffb4fd5f9c2b0ec618ef8121fe216b8b", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/microsoft/magentic-ui/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T06:42:04Z", "role": "primary", "sha256": "98a96c38494df4963951e32b81ec4effebe4c8a812ab8d9e7d185d59e86fe2fc", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/microsoft/magentic-ui/HEAD/LICENSE"}], "id": "github:microsoft/magentic-ui", "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/microsoft/magentic-ui/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/microsoft/magentic-ui"}], "name": "MagenticLite / Magentic UI", "notes": "", "published_at": null, "relations": [], "review_status": "curated", "scope": "core", "summary": "Microsoft agent application combining orchestration with browser interaction; successor to Magentic UI.", "topics": ["human-in-the-loop"], "updated_at": null, "url": "https://github.com/microsoft/magentic-ui"}
{"aliases": [], "authors": [], "details": {"availability": "Published source; official release metadata inspected", "release_tag": "v0.3.0"}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T03:59:54Z", "role": "primary", "sha256": "4d8a0b0d6286c4bab116c155f087948062166e7e169d6bbaf504fedb4f32dc67", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://raw.githubusercontent.com/nanobrowser/nanobrowser/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T03:59:53Z", "role": "primary", "sha256": "6b48386f32a971a724906de8b6d5ea53036f3813411ddc12931be3010c034400", "supports": ["identity"], "url": "https://api.github.com/repos/nanobrowser/nanobrowser"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T03:59:54Z", "role": "primary", "sha256": "afd0b71b01d3d89033d8f32abaf9871f38e8fb0c50442ed3a8cbb3079d4a4fe0", "supports": ["release_status"], "url": "https://api.github.com/repos/nanobrowser/nanobrowser/git/trees/master?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:01:00Z", "role": "primary", "sha256": "c7b28db1b1c8e78d100da066b08db713bc144baff99941b6c94489ef61baee8f", "supports": ["release_status"], "url": "https://api.github.com/repos/nanobrowser/nanobrowser/releases/latest"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T04:01:00Z", "role": "primary", "sha256": "c71d239df91726fc519c6eb72d318ec65820627232b2f796219e87dcf35d0ab4", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/nanobrowser/nanobrowser/HEAD/LICENSE"}, {"method": "license_header", "retrieved_at": "2026-10-17T04:01:00Z", "role": "primary", "sha256": "c71d239df91726fc519c6eb72d318ec65820627232b2f796219e87dcf35d0ab4", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/nanobrowser/nanobrowser/HEAD/LICENSE"}], "id": "github:nanobrowser/nanobrowser", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-17T03:59:54Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "Apache-2.0", "retrieved_at": "2026-10-17T04:01:00Z", "url": "https://raw.githubusercontent.com/nanobrowser/nanobrowser/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/nanobrowser/nanobrowser"}], "name": "Nanobrowser", "notes": "The official repository publishes extension source and a v0.3.0 ZIP release. It names Chrome and Edge as supported browsers. Apache-2.0 is declared for the code. Ask/Auto behavior and provider choices are documented, but no security guarantee, model performance or service-price claim is inferred.", "published_at": null, "relations": [], "review_status": "curated", "scope": "core", "summary": "A browser-extension agent that reads pages, interacts with elements and moves between tabs from a side-panel task interface.", "topics": ["web-navigation", "automation", "human-in-the-loop"], "updated_at": null, "url": "https://github.com/nanobrowser/nanobrowser"}
{"aliases": [], "authors": [], "details": {"license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/solplum/yunkai/2aa6cd83165bfc201b0128c250e80892895d63af/LICENSE"], "reason": "Complete root MIT license text inspected at the pinned commit. Repository software only; data, weights and third-party assets retain separate terms."}}, "repository_revision": "2aa6cd83165bfc201b0128c250e80892895d63af"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-20T11:57:12.431753+00:00", "role": "primary", "sha256": "eea31b35aa0b55c97f8ccda3438382ad49d2453671b49bb817efb5bc3384e900", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://raw.githubusercontent.com/solplum/yunkai/2aa6cd83165bfc201b0128c250e80892895d63af/README.md"}, {"method": "github_repository_identity", "retrieved_at": "2026-10-20T11:06:00.276978+00:00", "role": "primary", "sha256": "01f97a863952373bd10a1f849d79706344be63f5774fe7e0cfe9ba774e41d213", "supports": ["identity"], "url": "https://api.github.com/search/repositories?q=topic%3Amobile-agent+is%3Apublic+fork%3Afalse+created%3A2008-01-01..2026-10-20&per_page=100&page=1&sort=updated&order=asc"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T11:54:02.419278+00:00", "role": "primary", "sha256": "e2797712132e69cc50fa7ad2ae815a48339f77eebec1e85a5dc0cd75702574f1", "supports": ["identity", "details.repository_revision"], "url": "https://api.github.com/repos/solplum/yunkai/commits/main"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T11:56:30.226860+00:00", "role": "primary", "sha256": "08f48cb5c15b8d505e76a91d996c656d87765946cc7137bce088e73e6e64cf06", "supports": ["identity", "details.repository_revision", "notes"], "url": "https://api.github.com/repos/solplum/yunkai/git/trees/4ca94f2f7d8f13099ae2c351d9578ed5ef0cb509?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T11:57:22.238029+00:00", "role": "primary", "sha256": "c39441f92c303832d42f1f153c4442533ac925007e12d70f3199c568a0baac88", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/solplum/yunkai/2aa6cd83165bfc201b0128c250e80892895d63af/LICENSE"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T11:59:51.424713+00:00", "role": "primary", "sha256": "964aad706676fc7b74e532f8bc441fa766f3ce0ca3e4d0e115c7a90c70656883", "supports": ["summary", "environments", "topics", "scope", "notes"], "url": "https://raw.githubusercontent.com/solplum/yunkai/2aa6cd83165bfc201b0128c250e80892895d63af/yunkai-android/app/src/main/java/com/zhuolin/yunkai/service/screen/ScreenSenseService.kt"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T11:59:54.745065+00:00", "role": "primary", "sha256": "628cc3015999176052a09e3f3bbb8635d87163a9430d98cb6a6a7c09117dfd37", "supports": ["notes"], "url": "https://raw.githubusercontent.com/solplum/yunkai/2aa6cd83165bfc201b0128c250e80892895d63af/yunkai-harmony/entry/src/main/ets/service/tools/M2Tools.ets"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T12:04:53.632582+00:00", "role": "primary", "sha256": "ba8c9b0f4b5cea5ac36a4b49f3cf27e77e5c6beb49022e43c667e359be2c6fca", "supports": ["environments", "notes"], "url": "https://raw.githubusercontent.com/solplum/yunkai/2aa6cd83165bfc201b0128c250e80892895d63af/yunkai-harmony/entry/src/main/ets/service/screen/ScreenSenseExtension.ets"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T12:04:56.914360+00:00", "role": "primary", "sha256": "423a77e1fd57f3e105898f1adfc94c493de5c0de97fae4379f9cf540c89d4640", "supports": ["notes"], "url": "https://raw.githubusercontent.com/solplum/yunkai/2aa6cd83165bfc201b0128c250e80892895d63af/yunkai-harmony/tests/fullflow/drive_harmony.sh"}], "id": "github:solplum/yunkai", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-20T11:57:12.431753+00:00", "licenses": {"code": {"basis": "Complete root MIT license text inspected at the pinned commit. Repository software only; data, weights and third-party assets retain separate terms.", "declared": "MIT", "retrieved_at": "2026-10-20T11:57:22.238029+00:00", "url": "https://raw.githubusercontent.com/solplum/yunkai/2aa6cd83165bfc201b0128c250e80892895d63af/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/solplum/yunkai"}], "name": "Yunkai", "notes": "The Android AccessibilityService contains concrete screen-reading and action paths. The inspected HarmonyOS accessibility extension is a lifecycle/event scaffold with its screen-content probe disabled; HDC appears in an external UI-test driver, so Harmony GUI-control parity is not established by this review. Setup requires a configured OpenAI-compatible model endpoint and API key: local storage does not establish offline inference or a no-transmission guarantee. Documented approval and privacy mechanisms are not independently tested security guarantees. No app, device test or model inference was run.", "published_at": null, "relations": [], "review_status": "curated", "scope": "core", "summary": "Android phone assistant combining model-driven tool use with accessibility-tree reading, screenshots, and tap, swipe and text-entry primitives. Its documented interaction flow presents operation plans for user confirmation.", "topics": ["mobile-navigation", "screen-understanding", "human-in-the-loop"], "updated_at": null, "url": "https://github.com/solplum/yunkai"}
{"aliases": [], "authors": [], "details": {"availability": "Public task definitions, evaluator and container orchestration source.", "benchmark_protocol": {"actions": ["Android GUI operations", "MCP calls when enabled", "agent-user interaction when enabled"], "counts": [{"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/Tongyi-MAI/MobileWorld/HEAD/README.md", "population": "README benchmark overview", "qualifier": "exact", "unit": "tasks", "value": 201}, {"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/Tongyi-MAI/MobileWorld/HEAD/README.md", "population": "README benchmark overview", "qualifier": "exact", "unit": "mobile applications", "value": 20}], "evaluation": "The publisher documents text-answer matching, backend database verification, local-storage inspection and application callbacks.", "evidence_urls": ["https://raw.githubusercontent.com/Tongyi-MAI/MobileWorld/HEAD/README.md"], "focus": ["Android task completion", "agent-user interaction", "hybrid GUI and MCP use"], "notes": "Publisher-reported inventory. GUI-only, agent-user and MCP-augmented configurations have different dependencies and task eligibility; preserve enabled modes and step budget with results.", "observations": ["Android application interface", "user-agent interaction when enabled"], "setting": "interactive"}, "leaderboard_url": "https://tongyi-mai.github.io/MobileWorld/#leaderboard", "license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/Tongyi-MAI/MobileWorld/HEAD/LICENSE"], "reason": "Repository code license; self-hosted applications, model APIs and MCP services keep their own terms."}}, "repository_revision": "e7eee8981135ecbe39005a231edf63574defd79c"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:12:16Z", "role": "primary", "sha256": "fbc306b3414dfdb230171358efd6eb8d67573fbb94e11487a4a7e971e2abf0d9", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations", "details"], "url": "https://raw.githubusercontent.com/Tongyi-MAI/MobileWorld/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:12:15Z", "role": "primary", "sha256": "266e1fe5b5b1a4c1540ab0fd5eba206395084ee283f0079f424930a156cf8e18", "supports": ["identity"], "url": "https://api.github.com/repos/Tongyi-MAI/MobileWorld"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:14:08Z", "role": "primary", "sha256": "14500ed2dfa14bb5c34211090f0c8827e4a16c3233e35a2d695c0d7261f9358b", "supports": ["release_status"], "url": "https://api.github.com/repos/Tongyi-MAI/MobileWorld/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:14:09Z", "role": "primary", "sha256": "b09fcaf4fbc265223c2479c410beef2b98c5e687c78ecedf3452aa12ffc465f1", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/Tongyi-MAI/MobileWorld/HEAD/LICENSE"}], "id": "github:tongyi-mai/mobileworld", "identifiers": {}, "kind": "benchmark", "last_verified_at": "2026-10-17T09:12:16Z", "licenses": {"code": {"basis": "Repository code license; self-hosted applications, model APIs and MCP services keep their own terms.", "declared": "Apache-2.0", "retrieved_at": "2026-10-17T09:14:09Z", "url": "https://raw.githubusercontent.com/Tongyi-MAI/MobileWorld/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/Tongyi-MAI/MobileWorld"}], "name": "MobileWorld (Tongyi-MAI)", "notes": "This is Tongyi-MAI/MobileWorld, distinct from jacklishufan/MobileWorld (MobileWorldBench). Public source includes task definitions, verification logic and environment orchestration. The README reports 201 tasks across 20 applications; this is not an independent count of the evolving task tree. Linux/KVM or WSL2/KVM is required by the documented host setup, with macOS support in progress. MCP and simulated-user settings can require separate external services.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/Tongyi-MAI/MobileWorld/HEAD/README.md", "target_id": "arxiv:2512.19432", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "A containerized Android evaluation environment with GUI-only, agent-user interaction and MCP-augmented tasks, plus checks on application and backend state.", "topics": ["evaluation", "mobile-navigation", "human-in-the-loop", "tool-integration"], "updated_at": null, "url": "https://github.com/Tongyi-MAI/MobileWorld"}
{"aliases": [], "authors": [], "details": {}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T13:14:29Z", "role": "primary", "sha256": "15d40c50df8493575a94aa2aa4913c654fea7b7f821d99b7bf443a64f1cc50d3", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/Wuzheng02/OS-Kairos/HEAD/README.md"}], "id": "github:wuzheng02/os-kairos", "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/Wuzheng02/OS-Kairos"}], "name": "OS-Kairos", "notes": "The documented evaluation modes distinguish model-simulated human feedback from actual human interaction; preserve that distinction when interpreting results.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/Wuzheng02/OS-Kairos/HEAD/README.md", "target_id": "arxiv:2503.16465", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Android GUI-agent research implementation that uses confidence-guided interaction to choose between autonomous actions and human assistance.", "topics": ["human-in-the-loop", "safety-security", "mobile-navigation"], "updated_at": null, "url": "https://github.com/Wuzheng02/OS-Kairos"}
{"aliases": [], "authors": [], "details": {}, "environments": ["unspecified"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:59:55Z", "role": "primary", "sha256": "282883d02391a86c5a4d0cdafdc1d33f10a5073f31eba8c852095601d0aeb939", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "notes"], "url": "https://raw.githubusercontent.com/Wuzheng02/VeriOS/HEAD/README.md"}], "id": "github:wuzheng02/verios", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-18T10:59:55Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/Wuzheng02/VeriOS"}], "name": "VeriOS", "notes": "The released scripts evaluate recorded task inputs; the retained README does not establish a live Android execution environment. No runtime behavior or reported results were independently reproduced.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/Wuzheng02/VeriOS/HEAD/README.md", "target_id": "arxiv:2509.07553", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Research code for proactive human–GUI-agent verification, including offline interaction-loop and dual-agent evaluation scripts plus LLaMA-Factory fine-tuning configurations.", "topics": ["safety-security", "human-in-the-loop", "mobile-navigation"], "updated_at": null, "url": "https://github.com/Wuzheng02/VeriOS"}
{"aliases": [], "authors": [], "details": {}, "environments": ["unspecified"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T10:43:11Z", "role": "primary", "sha256": "2cea1be73ea75acd46990ab1fb567fbd83ad58a2bc69ca858b207660df9fe8c1", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/X-LANCE/META-GUI-baseline/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T10:48:42Z", "role": "primary", "sha256": "01fb614f0536ea76c1c25809cc8cde66cde21a96ac6ca0ea81b2c093e2157d2a", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/X-LANCE/META-GUI-baseline/HEAD/LICENSE"}], "id": "github:x-lance/meta-gui-baseline", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-16T10:43:11Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "MIT", "retrieved_at": "2026-10-16T10:48:42Z", "url": "https://raw.githubusercontent.com/X-LANCE/META-GUI-baseline/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/X-LANCE/META-GUI-baseline"}], "name": "META-GUI baseline", "notes": "Its README describes mobile GUI agents without naming an operating system.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/X-LANCE/META-GUI-baseline/HEAD/README.md", "target_id": "arxiv:2205.11029", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Training and evaluation pipeline for conversational mobile GUI agents, with action prediction, response generation, and dataset processing.", "topics": ["mobile-navigation", "human-in-the-loop"], "updated_at": null, "url": "https://github.com/X-LANCE/META-GUI-baseline"}
{"aliases": [], "authors": ["Toby Jia-Jun Li", "Amos Azaria", "Brad A. Myers"], "details": {"publication": {"authors": ["Toby Jia-Jun Li", "Amos Azaria", "Brad A. Myers"], "date_basis": "Supplied year precision: 2017. Publications: SUGILITE, CHI 2017 Only publication year is asserted; event dates are kept separate.", "date_parts": [2017], "doi": "10.1145/3025453.3025483", "event_dates": null, "evidence_urls": ["https://www.cs.cmu.edu/~NatProg/sugilite.html", "https://toby.li/files/TobyLi-CHI2017-Sugilite.pdf"], "identity_basis": "No same-work record located in the full active catalog. Similarity screening was not used as automatic identity approval. The raw archive and exclusions are additionally checked during normalization.", "issue": null, "notes": "Related to later APPINITE and PUMICE work but distinguished here by its 2017 three-author publication and base demonstration-automation contribution. Publication month/day are not independently identified by the retained sources.", "pages": "6038-6049", "pdf_url": "https://toby.li/files/TobyLi-CHI2017-Sugilite.pdf", "publisher": null, "series": null, "title": "SUGILITE: Creating Multimodal Smartphone Automation by Demonstration", "type": "conference-paper", "url": "https://doi.org/10.1145/3025453.3025483", "venue": "Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems", "volume": null}}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:53:22Z", "role": "primary", "sha256": "32ee05ff1a91ed4a22d4bc156a462872aa4e28f5c79abfbce5dbf3655f412033", "supports": ["details.publication.date_basis", "details.publication.date_parts", "details.publication.notes", "details.publication.type", "details.publication.venue"], "url": "https://www.cs.cmu.edu/~NatProg/sugilite.html"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:54:39Z", "role": "primary", "sha256": "cc23c19284f0a558c4b99b8b051abca8ed61e274e9e1f1fed1ae91487f4368ad", "supports": ["details.publication.authors", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.title", "details.publication.url"], "url": "https://toby.li/files/TobyLi-CHI2017-Sugilite.pdf"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:54:39Z", "role": "primary", "sha256": "cc23c19284f0a558c4b99b8b051abca8ed61e274e9e1f1fed1ae91487f4368ad", "supports": ["details.publication.authors", "details.publication.doi", "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://toby.li/files/TobyLi-CHI2017-Sugilite.pdf"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:53:22Z", "role": "primary", "sha256": "32ee05ff1a91ed4a22d4bc156a462872aa4e28f5c79abfbce5dbf3655f412033", "supports": ["details.publication.date_basis", "details.publication.date_parts", "details.publication.notes", "details.publication.type", "details.publication.venue", "kind"], "url": "https://www.cs.cmu.edu/~NatProg/sugilite.html"}], "id": "url:392a602c660ffa91be46", "identifiers": {"doi": "10.1145/3025453.3025483"}, "kind": "paper", "last_verified_at": "2026-10-17T09:54:39Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://doi.org/10.1145/3025453.3025483"}, {"rel": "publication_pdf", "url": "https://toby.li/files/TobyLi-CHI2017-Sugilite.pdf"}], "name": "SUGILITE: Creating Multimodal Smartphone Automation by Demonstration", "notes": "The annotation was reviewed against retained primary publication sources. Experiments have not been independently reproduced. Related to later APPINITE and PUMICE work but distinguished here by its 2017 three-author publication and base demonstration-automation contribution.", "published_at": null, "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces SUGILITE, which combines spoken instructions, demonstrated Android actions, and UI hierarchies to generate reusable automation scripts. Its interface includes context checking and repairs for unfamiliar situations and changing app interfaces.", "topics": ["mobile-navigation", "recording", "human-in-the-loop", "cross-app"], "updated_at": null, "url": "https://doi.org/10.1145/3025453.3025483"}
{"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": ["Toby Jia-Jun Li", "Oriana Riva"], "details": {"publication": {"authors": ["Toby Jia-Jun Li", "Oriana Riva"], "date_basis": "Supplied month precision: 2018-06. Publication line: MobiSys 2018, June 2018", "date_parts": [2018, 6], "doi": "10.1145/3210240.3210339", "event_dates": null, "evidence_urls": ["https://www.microsoft.com/en-us/research/publication/kite-building-conversational-bots-mobile-apps/", "https://www.microsoft.com/en-us/research/wp-content/uploads/2018/05/kite_mobisys18.pdf"], "identity_basis": "No same-work record located in the full active catalog. Similarity screening was not used as automatic identity approval. The raw archive and exclusions are additionally checked during normalization.", "issue": null, "notes": "Final proceedings page range and volume are not established in the retained first-party citation. The paper states a 14-page extent, not a page range. Exact publication day is unknown.", "pages": null, "pdf_url": "https://www.microsoft.com/en-us/research/wp-content/uploads/2018/05/kite_mobisys18.pdf", "publisher": null, "series": null, "title": "Kite: Building Conversational Bots from Mobile Apps", "type": "conference-paper", "url": "https://doi.org/10.1145/3210240.3210339", "venue": "The 16th Annual International Conference on Mobile Systems, Applications, and Services", "volume": null}}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:59:26Z", "role": "primary", "sha256": "e2f5403d77f563ec6a88458e32d057559a8f8ba4d2a80304e70c2c793fb60bb5", "supports": ["details.publication.date_basis", "details.publication.date_parts", "details.publication.notes"], "url": "https://www.microsoft.com/en-us/research/publication/kite-building-conversational-bots-mobile-apps/"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:59:27Z", "role": "primary", "sha256": "bfb67283f49dad7974d97e3b3da9d9ff24744d18e83c4d25cbe4f12ecf807071", "supports": ["details.publication.authors", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pdf_url", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue"], "url": "https://www.microsoft.com/en-us/research/wp-content/uploads/2018/05/kite_mobisys18.pdf"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:59:27Z", "role": "primary", "sha256": "bfb67283f49dad7974d97e3b3da9d9ff24744d18e83c4d25cbe4f12ecf807071", "supports": ["details.publication.authors", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pdf_url", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue", "environments", "kind", "name", "scope", "summary", "topics"], "url": "https://www.microsoft.com/en-us/research/wp-content/uploads/2018/05/kite_mobisys18.pdf"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:59:26Z", "role": "primary", "sha256": "e2f5403d77f563ec6a88458e32d057559a8f8ba4d2a80304e70c2c793fb60bb5", "supports": ["details.publication.date_basis", "details.publication.date_parts", "details.publication.notes"], "url": "https://www.microsoft.com/en-us/research/publication/kite-building-conversational-bots-mobile-apps/"}], "id": "url:901ca855640457dfe2f1", "identifiers": {"doi": "10.1145/3210240.3210339"}, "kind": "paper", "last_verified_at": "2026-10-17T09:59:27Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://doi.org/10.1145/3210240.3210339"}, {"rel": "publication_pdf", "url": "https://www.microsoft.com/en-us/research/wp-content/uploads/2018/05/kite_mobisys18.pdf"}], "name": "Kite: Building Conversational Bots from Mobile Apps", "notes": "The annotation was reviewed against retained primary publication sources. Experiments have not been independently reproduced. ", "published_at": null, "relations": [], "review_status": "curated", "scope": "core", "summary": "Derives task graphs from demonstrated Android app interactions and layers a conversational question-answer interface over those graphs. The evaluation examines task-model extraction and developer use of the resulting bot templates.", "topics": ["mobile-navigation", "workflow", "human-in-the-loop"], "updated_at": null, "url": "https://doi.org/10.1145/3210240.3210339"}
{"aliases": [], "authors": ["Toby Jia-Jun Li", "Igor Labutov", "Xiaohan Nancy Li", "Xiaoyi Zhang", "Wenze Shi", "Wanling Ding", "Tom M. Mitchell", "Brad A. Myers"], "details": {"publication": {"authors": ["Toby Jia-Jun Li", "Igor Labutov", "Xiaohan Nancy Li", "Xiaoyi Zhang", "Wenze Shi", "Wanling Ding", "Tom M. Mitchell", "Brad A. Myers"], "date_basis": "Supplied year precision: 2018. APPINITE publication citation: VL/HCC 2018", "date_parts": [2018], "doi": null, "event_dates": null, "evidence_urls": ["https://toby.li/files/TobyLi-VLHCC18-APPINITE.pdf", "https://www.cs.cmu.edu/~NatProg/sugilite.html", "https://toby.li/publications/"], "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 base SUGILITE framework and later PUMICE work are distinct publications; this paper focuses on resolving data-description ambiguity in demonstrations. Author-hosted manuscript and institutional publication lists supply the retained evidence. DOI, final page range, and exact publication day remain unknown.", "pages": null, "pdf_url": "https://toby.li/files/TobyLi-VLHCC18-APPINITE.pdf", "publisher": null, "series": null, "title": "APPINITE: A Multi-Modal Interface for Specifying Data Descriptions in Programming by Demonstration Using Natural Language Instructions", "type": "conference-paper", "url": "https://ieeexplore.ieee.org/document/8506506", "venue": "Proceedings of the 2018 IEEE Symposium on Visual Languages and Human-Centric Computing", "volume": null}}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:54:46Z", "role": "primary", "sha256": "6e05fd08b187adba32148e0835ef6b9e20cd6a3ba5e0de6e2778eb18c02788ee", "supports": ["details.publication.authors", "details.publication.identity_basis", "details.publication.notes", "details.publication.pdf_url", "details.publication.title"], "url": "https://toby.li/files/TobyLi-VLHCC18-APPINITE.pdf"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:53:22Z", "role": "primary", "sha256": "32ee05ff1a91ed4a22d4bc156a462872aa4e28f5c79abfbce5dbf3655f412033", "supports": ["details.publication.date_basis", "details.publication.date_parts", "details.publication.notes", "details.publication.type", "details.publication.venue"], "url": "https://www.cs.cmu.edu/~NatProg/sugilite.html"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:55:20Z", "role": "primary", "sha256": "adfcec73830b612b181d4846ace3fbaab1768e906715c1d90fe657e23fe4d985", "supports": ["details.publication.identity_basis", "details.publication.notes", "details.publication.url"], "url": "https://toby.li/publications/"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:54:46Z", "role": "primary", "sha256": "6e05fd08b187adba32148e0835ef6b9e20cd6a3ba5e0de6e2778eb18c02788ee", "supports": ["details.publication.authors", "details.publication.identity_basis", "details.publication.notes", "details.publication.pdf_url", "details.publication.title", "environments", "name", "scope", "summary", "topics"], "url": "https://toby.li/files/TobyLi-VLHCC18-APPINITE.pdf"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:53:22Z", "role": "primary", "sha256": "32ee05ff1a91ed4a22d4bc156a462872aa4e28f5c79abfbce5dbf3655f412033", "supports": ["details.publication.date_basis", "details.publication.date_parts", "details.publication.notes", "details.publication.type", "details.publication.venue", "kind"], "url": "https://www.cs.cmu.edu/~NatProg/sugilite.html"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:55:20Z", "role": "primary", "sha256": "adfcec73830b612b181d4846ace3fbaab1768e906715c1d90fe657e23fe4d985", "supports": ["details.publication.identity_basis", "details.publication.notes", "details.publication.url"], "url": "https://toby.li/publications/"}], "id": "url:ba2c664c1b3c6f1b172f", "identifiers": {}, "kind": "paper", "last_verified_at": "2026-10-17T09:54:46Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://ieeexplore.ieee.org/document/8506506"}, {"rel": "publication_pdf", "url": "https://toby.li/files/TobyLi-VLHCC18-APPINITE.pdf"}], "name": "APPINITE: A Multi-Modal Interface for Specifying Data Descriptions in Programming by Demonstration Using Natural Language Instructions", "notes": "The annotation was reviewed against retained primary publication sources. Experiments have not been independently reproduced. The base SUGILITE framework and later PUMICE work are distinct publications; this paper focuses on resolving data-description ambiguity in demonstrations.", "published_at": null, "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces APPINITE, a conversational interface for specifying the intended targets of demonstrated Android actions. It constructs semantic UI representations and uses an on-screen overlay to preview and refine executable data descriptions.", "topics": ["grounding", "human-in-the-loop", "mobile-navigation"], "updated_at": null, "url": "https://ieeexplore.ieee.org/document/8506506"}
{"aliases": [], "authors": ["Allen Cypher"], "details": {"publication": {"authors": ["Allen Cypher"], "date_basis": "Supplied year precision: 1991. Opening CHI 1991 citation and copyright", "date_parts": [1991], "doi": null, "event_dates": null, "evidence_urls": ["https://acypher.com/Eager/", "https://acypher.com/Publications/CHI91/EagerCHI.html"], "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 1991 conference paper is kept distinct from the 1991 video and 1993/1995 reprints listed on the author project page. Macintosh/HyperCard is tagged desktop; it is not represented as support for contemporary macOS. DOI not verified in a retained first-party source. Exact publication month/day is not inferred from conference dates.", "pages": "33-39", "pdf_url": null, "publisher": null, "series": null, "title": "Eager: Programming Repetitive Tasks by Example", "type": "conference-paper", "url": "https://acypher.com/Publications/CHI91/EagerCHI.html", "venue": "Proceedings of CHI '91", "volume": null}}, "environments": ["desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:53:27Z", "role": "primary", "sha256": "64cffd3fca934465e1ea28bdf62e087bc9abdf50b9b72d04c2e57df0bcf41adb", "supports": ["details.publication.identity_basis", "details.publication.notes", "details.publication.title"], "url": "https://acypher.com/Eager/"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:55:21Z", "role": "primary", "sha256": "876d4c8dfac48443336a5ee97b463642db013631a2b97cacd090e61afd95ad59", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.type", "details.publication.url", "details.publication.venue"], "url": "https://acypher.com/Publications/CHI91/EagerCHI.html"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:55:21Z", "role": "primary", "sha256": "876d4c8dfac48443336a5ee97b463642db013631a2b97cacd090e61afd95ad59", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.type", "details.publication.url", "details.publication.venue", "environments", "kind", "scope", "summary", "topics"], "url": "https://acypher.com/Publications/CHI91/EagerCHI.html"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:53:27Z", "role": "primary", "sha256": "64cffd3fca934465e1ea28bdf62e087bc9abdf50b9b72d04c2e57df0bcf41adb", "supports": ["details.publication.identity_basis", "details.publication.notes", "details.publication.title", "name"], "url": "https://acypher.com/Eager/"}], "id": "url:bf0c11018e49dff6aeec", "identifiers": {}, "kind": "paper", "last_verified_at": "2026-10-17T09:55:21Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://acypher.com/Publications/CHI91/EagerCHI.html"}], "name": "Eager: Programming Repetitive Tasks by Example", "notes": "The annotation was reviewed against retained primary publication sources. Experiments have not been independently reproduced. The 1991 conference paper is kept distinct from the 1991 video and 1993/1995 reprints listed on the author project page. Macintosh/HyperCard is tagged desktop; it is not represented as support for contemporary macOS.", "published_at": null, "relations": [], "review_status": "curated", "scope": "core", "summary": "Describes Eager, which detects repeated HyperCard interactions, generalizes them into loops, and previews predicted actions by highlighting interface objects. Users can then ask it to execute the inferred repetitive task.", "topics": ["automation", "human-in-the-loop", "interaction-history"], "updated_at": null, "url": "https://acypher.com/Publications/CHI91/EagerCHI.html"}