{"aliases": [], "authors": ["Longtao Zheng", "Rundong Wang", "Xinrun Wang", "Bo An"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "8fdb5fed2b2fa0ecbf5910dd6813d8de7910af286a1e4e00cdebb734a82e78b0", "submitted_at": "2023-06-13T15:49:41Z", "updated_at": "2024-01-19T06:59:26Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2306.07863v3"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=1802.08802%2C1808.09132%2C2005.03776%2C2101.09465%2C2103.16057%2C2104.08560%2C2105.13231%2C2107.03374%2C2107.13731%2C2108.03353%2C2202.02312%2C2202.08137%2C2204.01691%2C2205.11029%2C2207.01206%2C2209.08199%2C2209.08655%2C2209.14927%2C2209.15099%2C2210.02663%2C2210.03629%2C2210.03945%2C2211.07615%2C2301.10165%2C2301.13280%2C2302.04761%2C2303.07280%2C2303.08774%2C2303.17491%2C2303.17580%2C2304.04370%2C2304.07061%2C2305.04091%2C2305.05662%2C2305.08144%2C2305.10601%2C2305.11854%2C2305.12487%2C2305.14257%2C2305.14325%2C2305.16504%2C2305.16653%2C2305.19308%2C2306.00245%2C2306.06070%2C2306.07863%2C2307.10088%2C2307.12856%2C2307.13854%2C2308.01990&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=1802.08802%2C1808.09132%2C2005.03776%2C2101.09465%2C2103.16057%2C2104.08560%2C2105.13231%2C2107.03374%2C2107.13731%2C2108.03353%2C2202.02312%2C2202.08137%2C2204.01691%2C2205.11029%2C2207.01206%2C2209.08199%2C2209.08655%2C2209.14927%2C2209.15099%2C2210.02663%2C2210.03629%2C2210.03945%2C2211.07615%2C2301.10165%2C2301.13280%2C2302.04761%2C2303.07280%2C2303.08774%2C2303.17491%2C2303.17580%2C2304.04370%2C2304.07061%2C2305.04091%2C2305.05662%2C2305.08144%2C2305.10601%2C2305.11854%2C2305.12487%2C2305.14257%2C2305.14325%2C2305.16504%2C2305.16653%2C2305.19308%2C2306.00245%2C2306.06070%2C2306.07863%2C2307.10088%2C2307.12856%2C2307.13854%2C2308.01990&max_results=50"}], "id": "arxiv:2306.07863", "identifiers": {"arxiv": "2306.07863"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:20Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2306.07863"}], "name": "Synapse: Trajectory-as-Exemplar Prompting with Memory for Computer Control", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2023-06-13", "relations": [], "review_status": "curated", "scope": "core", "summary": "Synapse filters task-irrelevant page state and places complete abstracted state-action trajectories into the prompt. An embedding-based exemplar memory retrieves relevant demonstrations for new tasks, evaluated on MiniWoB++ and Mind2Web.", "topics": ["memory", "web-navigation", "planning", "efficiency"], "updated_at": "2024-01-19", "url": "https://arxiv.org/abs/2306.07863"}
{"aliases": [], "authors": ["Hao Wen", "Yuanchun Li", "Guohong Liu", "Shanhui Zhao", "Tao Yu", "Toby Jia-Jun Li", "Shiqi Jiang", "Yunhao Liu", "Yaqin Zhang", "Yunxin Liu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6a103fe10850a39d25bbb2b4a9ea0593d686938243bb912c674fd906e754c1fb", "submitted_at": "2023-08-29T13:02:30Z", "updated_at": "2024-03-09T09:38:51Z", "version": 4, "versioned_url": "https://arxiv.org/abs/2308.15272v4"}}, "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:2308.15272", "identifiers": {"arxiv": "2308.15272"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:24Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2308.15272"}], "name": "AutoDroid: LLM-powered Task Automation in Android", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2023-08-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "Combines language-model reasoning with app analysis, UI representations, and exploration-derived memory for Android task automation.", "topics": ["mobile-navigation", "memory"], "updated_at": "2024-03-09", "url": "https://arxiv.org/abs/2308.15272"}
{"aliases": [], "authors": ["Sunjae Lee", "Junyoung Choi", "Jungjae Lee", "Munim Hasan Wasi", "Hojun Choi", "Steven Y. Ko", "Sangeun Oh", "Insik Shin"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "b9b6f10b17513f565a13e035802d8bd681b5b0d468ba000df1427a8d59591e2c", "submitted_at": "2023-12-04T06:13:35Z", "updated_at": "2024-10-16T08:15:53Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2312.03003v3"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:24Z", "role": "primary", "sha256": "3aac6e9748769efff3c475ede0da192776daa03431d231b5f67e2aa50f8a3be9", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2308.03688%2C2308.15272%2C2309.08172%2C2309.11436%2C2310.04716%2C2310.04869%2C2310.05126%2C2310.06770%2C2310.08740%2C2310.10634%2C2310.11441%2C2310.16042%2C2310.18331%2C2311.01767%2C2311.07562%2C2311.09278%2C2311.12983%2C2311.18751%2C2312.03003%2C2312.06677%2C2312.08914%2C2312.11671%2C2312.13108%2C2312.13771%2C2312.15820%2C2312.16062%2C2401.01614%2C2401.04124%2C2401.07781%2C2401.10935%2C2401.13178%2C2401.13649%2C2401.13919%2C2401.14268%2C2401.16158%2C2402.01030%2C2402.04476%2C2402.04615%2C2402.05930%2C2402.06596%2C2402.07456%2C2402.07939%2C2402.07945%2C2402.08567%2C2402.11941%2C2402.14073%2C2402.15057%2C2402.16965%2C2402.17553%2C2403.02713&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:24Z", "role": "primary", "sha256": "3aac6e9748769efff3c475ede0da192776daa03431d231b5f67e2aa50f8a3be9", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?id_list=2308.03688%2C2308.15272%2C2309.08172%2C2309.11436%2C2310.04716%2C2310.04869%2C2310.05126%2C2310.06770%2C2310.08740%2C2310.10634%2C2310.11441%2C2310.16042%2C2310.18331%2C2311.01767%2C2311.07562%2C2311.09278%2C2311.12983%2C2311.18751%2C2312.03003%2C2312.06677%2C2312.08914%2C2312.11671%2C2312.13108%2C2312.13771%2C2312.15820%2C2312.16062%2C2401.01614%2C2401.04124%2C2401.07781%2C2401.10935%2C2401.13178%2C2401.13649%2C2401.13919%2C2401.14268%2C2401.16158%2C2402.01030%2C2402.04476%2C2402.04615%2C2402.05930%2C2402.06596%2C2402.07456%2C2402.07939%2C2402.07945%2C2402.08567%2C2402.11941%2C2402.14073%2C2402.15057%2C2402.16965%2C2402.17553%2C2403.02713&max_results=50"}], "id": "arxiv:2312.03003", "identifiers": {"arxiv": "2312.03003"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:24Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2312.03003"}], "name": "Explore, Select, Derive, and Recall: Augmenting LLM with Human-like Memory for Mobile Task Automation", "notes": "Reviewed against the retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2023-12-04", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces MobileGPT, which organizes mobile-app experience into reusable subtasks using an explore, select, derive, and recall cycle. Its memory supports learning task procedures and adapting them to new contexts.", "topics": ["mobile-navigation", "memory", "planning"], "updated_at": "2024-10-16", "url": "https://arxiv.org/abs/2312.03003"}
{"aliases": [], "authors": ["Chi Zhang", "Zhao Yang", "Jiaxuan Liu", "Yanda Li", "Yucheng Han", "Xin Chen", "Zebiao Huang", "Bin Fu", "Gang Yu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "61e54d83304755ce89c2ab69f7bb93a7fdc4d68cb5182daad9b5210f809d7ae1", "submitted_at": "2023-12-21T11:52:45Z", "updated_at": "2026-07-05T07:51:04Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2312.13771v3"}}, "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:2312.13771", "identifiers": {"arxiv": "2312.13771"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:24Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2312.13771"}], "name": "AppAgent: Multimodal Agents as Smartphone 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": "2023-12-21", "relations": [], "review_status": "curated", "scope": "core", "summary": "Builds app knowledge through exploration or human demonstrations and uses it to guide multimodal smartphone interactions.", "topics": ["mobile-navigation", "memory"], "updated_at": "2026-07-05", "url": "https://arxiv.org/abs/2312.13771"}
{"aliases": [], "authors": ["Lihang Pan", "Bowen Wang", "Chun Yu", "Yuxuan Chen", "Xiangyu Zhang", "Yuanchun Shi"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "0f592712517da90bfeb807ab445034c3974b406e0ed6b4933fbd19078984b9b8", "submitted_at": "2023-12-26T14:20:36Z", "updated_at": "2023-12-26T14:20:36Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2312.16062v1"}}, "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:2312.16062", "identifiers": {"arxiv": "2312.16062"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:24Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2312.16062"}], "name": "AutoTask: Executing Arbitrary Voice Commands by Exploring and Learning from 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": "2023-12-26", "relations": [], "review_status": "curated", "scope": "core", "summary": "Executes voice-requested mobile tasks through GUI exploration, backtracking, and accumulated interaction experience.", "topics": ["mobile-navigation", "memory"], "updated_at": "2023-12-26", "url": "https://arxiv.org/abs/2312.16062"}
{"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": ["Yang Deng", "Xuan Zhang", "Wenxuan Zhang", "Yifei Yuan", "See-Kiong Ng", "Tat-Seng Chua"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "5e4b233e423e9043b6d615c7b185a1eb0d47f646590e4fb9fd43ca156a1a1b16", "submitted_at": "2024-02-23T02:18:12Z", "updated_at": "2024-02-23T02:18:12Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2402.15057v1"}, "publication": {"authors": ["Yang Deng", "Xuan Zhang", "Wenxuan Zhang", "Yifei Yuan", "See Kiong Ng", "Tat-Seng Chua"], "date_basis": "Explicit publisher citation_publication_date: 2024/8. Preserve supplied precision; do not substitute conference event dates.", "date_parts": [2024, 8], "doi": "10.18653/v1/2024.acl-long.477", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2024.acl-long.477/"], "identity_basis": "Same work: normalized title, ordered author correspondence and the specific contribution in the complete publisher abstract were reviewed against the retained arXiv identity. The normalized publisher abstract matches the retained arXiv abstract SHA-256.", "issue": null, "notes": "Publisher proceedings edition; retained separately from the arXiv snapshot.", "pages": "8795-8812", "pdf_url": "https://aclanthology.org/2024.acl-long.477.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "On the Multi-turn Instruction Following for Conversational Web Agents", "type": "conference-paper", "url": "https://aclanthology.org/2024.acl-long.477/", "venue": "Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)", "volume": null}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:09:49Z", "role": "primary", "sha256": "ba56d6d0a422b340672036d5c872cd00d54a0f12cfe1326a3be65e8702786381", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.publisher", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue"], "url": "https://aclanthology.org/2024.acl-long.477/"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2402.15057", "identifiers": {"arxiv": "2402.15057"}, "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/2402.15057"}, {"rel": "publication", "url": "https://aclanthology.org/2024.acl-long.477/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2024.acl-long.477.pdf"}], "name": "On the Multi-turn Instruction Following for Conversational Web Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-02-23", "relations": [], "review_status": "curated", "scope": "core", "summary": "MT-Mind2Web evaluates web navigation through sequences of conversational user instructions. Self-MAP combines memory-augmented planning with reflection to preserve context and dependencies across those interaction turns.", "topics": ["evaluation", "memory", "self-reflection", "web-navigation"], "updated_at": "2024-02-23", "url": "https://arxiv.org/abs/2402.15057"}
{"aliases": [], "authors": ["Michael Lutz", "Arth Bohra", "Manvel Saroyan", "Artem Harutyunyan", "Giovanni Campagna"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "4c18dbfedcccf6434b3a50a9aa62552372df8b772ccda6a1aa699a8f052476bc", "submitted_at": "2024-04-08T23:10:47Z", "updated_at": "2024-04-08T23:10:47Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2404.05902v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2404.05902", "identifiers": {"arxiv": "2404.05902"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:58Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2404.05902"}], "name": "WILBUR: Adaptive In-Context Learning for Robust and Accurate Web Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-04-08", "relations": [], "review_status": "curated", "scope": "core", "summary": "WILBUR learns to rank prior task demonstrations for a web agent’s prompt and synthesizes instructions from past runs. A generated task curriculum and a backtracking mechanism support adaptation and recovery, evaluated on WebVoyager.", "topics": ["web-navigation", "memory", "robustness", "data-generation"], "updated_at": "2024-04-08", "url": "https://arxiv.org/abs/2404.05902"}
{"aliases": [], "authors": ["Shulin Tian", "Ziniu Zhang", "Liangyu Chen", "Ziwei Liu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "667b1f5053f3ae6ea712999932d5e9ee2f42e402cc72e713048d0790f5931976", "submitted_at": "2024-04-15T17:59:50Z", "updated_at": "2025-06-29T05:24:54Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2404.09992v2"}, "publication": {"authors": ["Shulin Tian", "Ziniu Zhang", "Liang-Yu Chen", "Ziwei Liu"], "date_basis": "Explicit publisher citation_publication_date: 2025/7. Preserve supplied precision; do not substitute conference event dates.", "date_parts": [2025, 7], "doi": "10.18653/v1/2025.findings-acl.703", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2025.findings-acl.703/"], "identity_basis": "Same work: normalized title, ordered author correspondence and the specific contribution in the complete publisher abstract were reviewed against the retained arXiv identity.", "issue": null, "notes": "Publisher proceedings edition; retained separately from the arXiv snapshot.", "pages": "13682-13697", "pdf_url": "https://aclanthology.org/2025.findings-acl.703.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "MMInA: Benchmarking Multihop Multimodal Internet Agents", "type": "conference-paper", "url": "https://aclanthology.org/2025.findings-acl.703/", "venue": "Findings of the Association for Computational Linguistics: ACL 2025", "volume": null}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:28Z", "role": "primary", "sha256": "be244bfae7f89ef9449b69d85aac44295eb7b5f7e5e1bbedd76df312696676a7", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2403.03186%2C2403.03788%2C2403.07718%2C2403.10170%2C2403.11905%2C2403.16971%2C2403.17918%2C2403.20271%2C2404.01549%2C2404.01744%2C2404.02475%2C2404.03648%2C2404.05427%2C2404.05719%2C2404.05955%2C2404.06474%2C2404.07972%2C2404.08755%2C2404.08860%2C2404.08948%2C2404.09992%2C2404.10887%2C2404.11459%2C2404.16006%2C2404.16048%2C2404.16054%2C2404.16660%2C2404.18074%2C2404.19296%2C2405.00516%2C2405.11120%2C2405.12842%2C2405.14573%2C2405.14751%2C2405.15793%2C2405.16247%2C2405.20309%2C2406.01014%2C2406.01623%2C2406.03679%2C2406.06947%2C2406.07822%2C2406.08184%2C2406.08340%2C2406.08451%2C2406.10227%2C2406.10819%2C2406.11317%2C2406.11896%2C2406.12373&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:09:51Z", "role": "primary", "sha256": "5e725b77b94b484284adaab9ec70674c9fc13a7fce34b764e59561f9180b915f", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.publisher", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue"], "url": "https://aclanthology.org/2025.findings-acl.703/"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:28Z", "role": "primary", "sha256": "be244bfae7f89ef9449b69d85aac44295eb7b5f7e5e1bbedd76df312696676a7", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2403.03186%2C2403.03788%2C2403.07718%2C2403.10170%2C2403.11905%2C2403.16971%2C2403.17918%2C2403.20271%2C2404.01549%2C2404.01744%2C2404.02475%2C2404.03648%2C2404.05427%2C2404.05719%2C2404.05955%2C2404.06474%2C2404.07972%2C2404.08755%2C2404.08860%2C2404.08948%2C2404.09992%2C2404.10887%2C2404.11459%2C2404.16006%2C2404.16048%2C2404.16054%2C2404.16660%2C2404.18074%2C2404.19296%2C2405.00516%2C2405.11120%2C2405.12842%2C2405.14573%2C2405.14751%2C2405.15793%2C2405.16247%2C2405.20309%2C2406.01014%2C2406.01623%2C2406.03679%2C2406.06947%2C2406.07822%2C2406.08184%2C2406.08340%2C2406.08451%2C2406.10227%2C2406.10819%2C2406.11317%2C2406.11896%2C2406.12373&max_results=50"}], "id": "arxiv:2404.09992", "identifiers": {"arxiv": "2404.09992"}, "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/2404.09992"}, {"rel": "publication", "url": "https://aclanthology.org/2025.findings-acl.703/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2025.findings-acl.703.pdf"}], "name": "MMInA: Benchmarking Multihop Multimodal Internet Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-04-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "Evaluates multimodal agents on compositional tasks across changing websites and studies replaying prior trajectories as memory support.", "topics": ["evaluation", "long-horizon", "memory"], "updated_at": "2025-06-29", "url": "https://arxiv.org/abs/2404.09992"}
{"aliases": [], "authors": ["Junyang Wang", "Haiyang Xu", "Haitao Jia", "Xi Zhang", "Ming Yan", "Weizhou Shen", "Ji Zhang", "Fei Huang", "Jitao Sang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a1338f3543a04bc2f9734e5fcabbfd5ebda6ac263461c2305f7c39383e938573", "submitted_at": "2024-06-03T05:50:00Z", "updated_at": "2024-06-03T05:50:00Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2406.01014v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:28Z", "role": "primary", "sha256": "be244bfae7f89ef9449b69d85aac44295eb7b5f7e5e1bbedd76df312696676a7", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2403.03186%2C2403.03788%2C2403.07718%2C2403.10170%2C2403.11905%2C2403.16971%2C2403.17918%2C2403.20271%2C2404.01549%2C2404.01744%2C2404.02475%2C2404.03648%2C2404.05427%2C2404.05719%2C2404.05955%2C2404.06474%2C2404.07972%2C2404.08755%2C2404.08860%2C2404.08948%2C2404.09992%2C2404.10887%2C2404.11459%2C2404.16006%2C2404.16048%2C2404.16054%2C2404.16660%2C2404.18074%2C2404.19296%2C2405.00516%2C2405.11120%2C2405.12842%2C2405.14573%2C2405.14751%2C2405.15793%2C2405.16247%2C2405.20309%2C2406.01014%2C2406.01623%2C2406.03679%2C2406.06947%2C2406.07822%2C2406.08184%2C2406.08340%2C2406.08451%2C2406.10227%2C2406.10819%2C2406.11317%2C2406.11896%2C2406.12373&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:28Z", "role": "primary", "sha256": "be244bfae7f89ef9449b69d85aac44295eb7b5f7e5e1bbedd76df312696676a7", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2403.03186%2C2403.03788%2C2403.07718%2C2403.10170%2C2403.11905%2C2403.16971%2C2403.17918%2C2403.20271%2C2404.01549%2C2404.01744%2C2404.02475%2C2404.03648%2C2404.05427%2C2404.05719%2C2404.05955%2C2404.06474%2C2404.07972%2C2404.08755%2C2404.08860%2C2404.08948%2C2404.09992%2C2404.10887%2C2404.11459%2C2404.16006%2C2404.16048%2C2404.16054%2C2404.16660%2C2404.18074%2C2404.19296%2C2405.00516%2C2405.11120%2C2405.12842%2C2405.14573%2C2405.14751%2C2405.15793%2C2405.16247%2C2405.20309%2C2406.01014%2C2406.01623%2C2406.03679%2C2406.06947%2C2406.07822%2C2406.08184%2C2406.08340%2C2406.08451%2C2406.10227%2C2406.10819%2C2406.11317%2C2406.11896%2C2406.12373&max_results=50"}], "id": "arxiv:2406.01014", "identifiers": {"arxiv": "2406.01014"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:28Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2406.01014"}], "name": "Mobile-Agent-v2: Mobile Device Operation Assistant with Effective Navigation via Multi-Agent Collaboration", "notes": "Summary and topic tags reviewed against the primary abstract. Reported experiments, task counts, and artifact availability are not independently reproduced.", "published_at": "2024-06-03", "relations": [], "review_status": "curated", "scope": "core", "summary": "Separates mobile planning, decision-making, and reflection, with a memory unit for task progress and relevant content.", "topics": ["planning", "memory", "mobile-navigation"], "updated_at": "2024-06-03", "url": "https://arxiv.org/abs/2406.01014"}
{"aliases": [], "authors": ["Yanda Li", "Chi Zhang", "Wenjia Jiang", "Wanqi Yang", "Bin Fu", "Pei Cheng", "Xin Chen", "Ling Chen", "Yunchao Wei"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d5b471c61c2ac9d4932bd71052ea35585699d70e9710f8366bc4fa8e4f2ebbae", "submitted_at": "2024-08-05T06:31:39Z", "updated_at": "2025-09-17T09:06:58Z", "version": 4, "versioned_url": "https://arxiv.org/abs/2408.11824v4"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:32Z", "role": "primary", "sha256": "b9a83994d8ac36ccdb606a5219511f8bb35ad3948ade4f9f3e28d386221aedfb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2406.12814%2C2406.13719%2C2406.14056%2C2406.14250%2C2406.14314%2C2406.18082%2C2406.19263%2C2407.00993%2C2407.01476%2C2407.01489%2C2407.01511%2C2407.01558%2C2407.03037%2C2407.03913%2C2407.04346%2C2407.05291%2C2407.07061%2C2407.08850%2C2407.09018%2C2407.09295%2C2407.10956%2C2407.13032%2C2407.13063%2C2407.15673%2C2407.15711%2C2407.16741%2C2407.17490%2C2407.18901%2C2407.19056%2C2407.20183%2C2408.00203%2C2408.02544%2C2408.06327%2C2408.07199%2C2408.11824%2C2408.15978%2C2409.02453%2C2409.04566%2C2409.07429%2C2409.08264%2C2409.09354%2C2409.10741%2C2409.11295%2C2409.13711%2C2409.14337%2C2409.14818%2C2409.15441%2C2409.15637%2C2409.17140%2C2409.20566&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:32Z", "role": "primary", "sha256": "b9a83994d8ac36ccdb606a5219511f8bb35ad3948ade4f9f3e28d386221aedfb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2406.12814%2C2406.13719%2C2406.14056%2C2406.14250%2C2406.14314%2C2406.18082%2C2406.19263%2C2407.00993%2C2407.01476%2C2407.01489%2C2407.01511%2C2407.01558%2C2407.03037%2C2407.03913%2C2407.04346%2C2407.05291%2C2407.07061%2C2407.08850%2C2407.09018%2C2407.09295%2C2407.10956%2C2407.13032%2C2407.13063%2C2407.15673%2C2407.15711%2C2407.16741%2C2407.17490%2C2407.18901%2C2407.19056%2C2407.20183%2C2408.00203%2C2408.02544%2C2408.06327%2C2408.07199%2C2408.11824%2C2408.15978%2C2409.02453%2C2409.04566%2C2409.07429%2C2409.08264%2C2409.09354%2C2409.10741%2C2409.11295%2C2409.13711%2C2409.14337%2C2409.14818%2C2409.15441%2C2409.15637%2C2409.17140%2C2409.20566&max_results=50"}], "id": "arxiv:2408.11824", "identifiers": {"arxiv": "2408.11824"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2408.11824"}], "name": "AppAgent v2: Advanced Agent for Flexible Mobile Interactions", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-08-05", "relations": [], "review_status": "curated", "scope": "core", "summary": "Separates exploration and deployment while using a structured application knowledge base and retrieval to guide mobile workflows.", "topics": ["mobile-navigation", "memory"], "updated_at": "2025-09-17", "url": "https://arxiv.org/abs/2408.11824"}
{"aliases": [], "authors": ["Zora Zhiruo Wang", "Jiayuan Mao", "Daniel Fried", "Graham Neubig"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d4c142be3a9c37bbb0d4f401a02de25bd25b4c3968f42ad9a61513abff2b5460", "submitted_at": "2024-09-11T17:21:00Z", "updated_at": "2024-09-11T17:21:00Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2409.07429v1"}}, "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:2409.07429", "identifiers": {"arxiv": "2409.07429"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2409.07429"}], "name": "Agent Workflow Memory", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-09-11", "relations": [], "review_status": "curated", "scope": "core", "summary": "Agent Workflow Memory extracts reusable task routines and selectively supplies them to guide later web actions. It studies both offline workflow induction from training examples and online induction from test experience on Mind2Web and WebArena.", "topics": ["memory", "workflow", "web-navigation"], "updated_at": "2024-09-11", "url": "https://arxiv.org/abs/2409.07429"}
{"aliases": [], "authors": ["Nikolai Rozanov", "Marek Rei"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "8a2e16e978f50c8d7161a11a11fba6aeda9617663de1746422151d38c4874cc9", "submitted_at": "2024-09-21T05:54:35Z", "updated_at": "2025-04-08T06:37:51Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2410.02810v3"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:35Z", "role": "primary", "sha256": "b3e67984f59a1b05ccd946eb54e2d37b0bb78ab0ca796fbb6d544793dafea3e0", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2410.00467%2C2410.00689%2C2410.02810%2C2410.02907%2C2410.05243%2C2410.06703%2C2410.08164%2C2410.09006%2C2410.09024%2C2410.11843%2C2410.11871%2C2410.11872%2C2410.13232%2C2410.13757%2C2410.13824%2C2410.13825%2C2410.13886%2C2410.14803%2C2410.15164%2C2410.16464%2C2410.17236%2C2410.17401%2C2410.17520%2C2410.17883%2C2410.18603%2C2410.18963%2C2410.18967%2C2410.19054%2C2410.19100%2C2410.19461%2C2410.19609%2C2410.20745%2C2410.22552%2C2410.23218%2C2410.24024%2C2411.00820%2C2411.02337%2C2411.02391%2C2411.04468%2C2411.04890%2C2411.06559%2C2411.07618%2C2411.10323%2C2411.13451%2C2411.13591%2C2411.15004%2C2411.17465%2C2411.18279%2C2412.00402%2C2412.01268&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:35Z", "role": "primary", "sha256": "b3e67984f59a1b05ccd946eb54e2d37b0bb78ab0ca796fbb6d544793dafea3e0", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2410.00467%2C2410.00689%2C2410.02810%2C2410.02907%2C2410.05243%2C2410.06703%2C2410.08164%2C2410.09006%2C2410.09024%2C2410.11843%2C2410.11871%2C2410.11872%2C2410.13232%2C2410.13757%2C2410.13824%2C2410.13825%2C2410.13886%2C2410.14803%2C2410.15164%2C2410.16464%2C2410.17236%2C2410.17401%2C2410.17520%2C2410.17883%2C2410.18603%2C2410.18963%2C2410.18967%2C2410.19054%2C2410.19100%2C2410.19461%2C2410.19609%2C2410.20745%2C2410.22552%2C2410.23218%2C2410.24024%2C2411.00820%2C2411.02337%2C2411.02391%2C2411.04468%2C2411.04890%2C2411.06559%2C2411.07618%2C2411.10323%2C2411.13451%2C2411.13591%2C2411.15004%2C2411.17465%2C2411.18279%2C2412.00402%2C2412.01268&max_results=50"}], "id": "arxiv:2410.02810", "identifiers": {"arxiv": "2410.02810"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:35Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2410.02810"}], "name": "StateAct: Enhancing LLM Base Agents via Self-prompting and State-tracking", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-09-21", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "Maintains explicit task goals and evolving state descriptions to improve base-agent decisions across interactive tasks, including web shopping.", "topics": ["planning", "memory"], "updated_at": "2025-04-08", "url": "https://arxiv.org/abs/2410.02810"}
{"aliases": [], "authors": ["Saaket Agashe", "Jiuzhou Han", "Shuyu Gan", "Jiachen Yang", "Ang Li", "Xin Eric Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "284888d16b049767f76448f23f5d35fda25b109fd67f450105950173ae81c97b", "submitted_at": "2024-10-10T17:43:51Z", "updated_at": "2024-10-10T17:43:51Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2410.08164v1"}, "publication": {"authors": ["Saaket Agashe", "Jiuzhou Han", "Shuyu Gan", "Jiachen Yang", "Ang Li", "Xin Wang"], "date_basis": "Explicit publisher citation_publication_date: 2025-05-01. Preserve supplied precision; do not substitute conference event dates.", "date_parts": [2025, 5, 1], "doi": null, "event_dates": null, "evidence_urls": ["https://proceedings.iclr.cc/paper_files/paper/2025/hash/394c7c30ea87b5c3521b4d9e9d419071-Abstract-Conference.html"], "identity_basis": "Same work: normalized title, ordered author correspondence and the specific contribution in the complete publisher abstract were reviewed against the retained arXiv identity.", "issue": null, "notes": "Publisher proceedings edition; retained separately from the arXiv snapshot. Publisher spelling, initials, name annotations or author lists differ from the retained preprint.", "pages": "22924-22946", "pdf_url": "https://proceedings.iclr.cc/paper_files/paper/2025/file/394c7c30ea87b5c3521b4d9e9d419071-Paper-Conference.pdf", "publisher": null, "series": null, "title": "Agent S: An Open Agentic Framework that Uses Computers Like a Human", "type": "conference-paper", "url": "https://proceedings.iclr.cc/paper_files/paper/2025/hash/394c7c30ea87b5c3521b4d9e9d419071-Abstract-Conference.html", "venue": "International Conference on Learning Representations", "volume": "2025"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:35Z", "role": "primary", "sha256": "b3e67984f59a1b05ccd946eb54e2d37b0bb78ab0ca796fbb6d544793dafea3e0", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2410.00467%2C2410.00689%2C2410.02810%2C2410.02907%2C2410.05243%2C2410.06703%2C2410.08164%2C2410.09006%2C2410.09024%2C2410.11843%2C2410.11871%2C2410.11872%2C2410.13232%2C2410.13757%2C2410.13824%2C2410.13825%2C2410.13886%2C2410.14803%2C2410.15164%2C2410.16464%2C2410.17236%2C2410.17401%2C2410.17520%2C2410.17883%2C2410.18603%2C2410.18963%2C2410.18967%2C2410.19054%2C2410.19100%2C2410.19461%2C2410.19609%2C2410.20745%2C2410.22552%2C2410.23218%2C2410.24024%2C2411.00820%2C2411.02337%2C2411.02391%2C2411.04468%2C2411.04890%2C2411.06559%2C2411.07618%2C2411.10323%2C2411.13451%2C2411.13591%2C2411.15004%2C2411.17465%2C2411.18279%2C2412.00402%2C2412.01268&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:10:01Z", "role": "primary", "sha256": "032cfd2f08acc7f85921748e0546bf6e16d038da412c207105ca9f6d433c70b3", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue", "details.publication.volume"], "url": "https://proceedings.iclr.cc/paper_files/paper/2025/hash/394c7c30ea87b5c3521b4d9e9d419071-Abstract-Conference.html"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:35Z", "role": "primary", "sha256": "b3e67984f59a1b05ccd946eb54e2d37b0bb78ab0ca796fbb6d544793dafea3e0", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2410.00467%2C2410.00689%2C2410.02810%2C2410.02907%2C2410.05243%2C2410.06703%2C2410.08164%2C2410.09006%2C2410.09024%2C2410.11843%2C2410.11871%2C2410.11872%2C2410.13232%2C2410.13757%2C2410.13824%2C2410.13825%2C2410.13886%2C2410.14803%2C2410.15164%2C2410.16464%2C2410.17236%2C2410.17401%2C2410.17520%2C2410.17883%2C2410.18603%2C2410.18963%2C2410.18967%2C2410.19054%2C2410.19100%2C2410.19461%2C2410.19609%2C2410.20745%2C2410.22552%2C2410.23218%2C2410.24024%2C2411.00820%2C2411.02337%2C2411.02391%2C2411.04468%2C2411.04890%2C2411.06559%2C2411.07618%2C2411.10323%2C2411.13451%2C2411.13591%2C2411.15004%2C2411.17465%2C2411.18279%2C2412.00402%2C2412.01268&max_results=50"}], "id": "arxiv:2410.08164", "identifiers": {"arxiv": "2410.08164"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:35Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2410.08164"}, {"rel": "publication", "url": "https://proceedings.iclr.cc/paper_files/paper/2025/hash/394c7c30ea87b5c3521b4d9e9d419071-Abstract-Conference.html"}, {"rel": "publication_pdf", "url": "https://proceedings.iclr.cc/paper_files/paper/2025/file/394c7c30ea87b5c3521b4d9e9d419071-Paper-Conference.pdf"}], "name": "Agent S: An Open Agentic Framework that Uses Computers Like a Human", "notes": "Editorial summary based on the primary paper abstract; experiments and released artifacts have not been independently reproduced.", "published_at": "2024-10-10", "relations": [], "review_status": "curated", "scope": "core", "summary": "Combines hierarchical planning, retrieved experience, and a computer interaction interface to execute multistep GUI tasks.", "topics": ["planning", "memory"], "updated_at": "2024-10-10", "url": "https://arxiv.org/abs/2410.08164"}
{"aliases": [], "authors": ["Zichen Zhu", "Hao Tang", "Yansi Li", "Dingye Liu", "Hongshen Xu", "Kunyao Lan", "Danyang Zhang", "Yixuan Jiang", "Hao Zhou", "Chenrun Wang", "Situo Zhang", "Liangtai Sun", "Yixiao Wang", "Yuheng Sun", "Lu Chen", "Kai Yu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "c0ef02f9c31bf2f095330cf018e7525aecc7644eb1552fb6b511ffced3889122", "submitted_at": "2024-10-17T16:53:50Z", "updated_at": "2025-05-13T06:25:09Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2410.13757v3"}, "publication": {"authors": ["Zichen Zhu", "Hao Tang", "Yansi Li", "Dingye Liu", "Hongshen Xu", "Kunyao Lan", "Danyang Zhang", "Yixuan Jiang", "Hao Zhou", "Chenrun Wang", "Situo Zhang", "Liangtai Sun", "Yixiao Wang", "Yuheng Sun", "Lu Chen", "Kai Yu"], "date_basis": "Explicit publisher citation_publication_date: 2025/4. Preserve supplied precision; do not substitute conference event dates.", "date_parts": [2025, 4], "doi": "10.18653/v1/2025.naacl-demo.43", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2025.naacl-demo.43/"], "identity_basis": "Same work: normalized title, ordered author correspondence and the specific contribution in the complete publisher abstract were reviewed against the retained arXiv identity.", "issue": null, "notes": "Publisher proceedings edition; retained separately from the arXiv snapshot.", "pages": "535-549", "pdf_url": "https://aclanthology.org/2025.naacl-demo.43.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "MobA: Multifaceted Memory-Enhanced Adaptive Planning for Efficient Mobile Task Automation", "type": "conference-paper", "url": "https://aclanthology.org/2025.naacl-demo.43/", "venue": "Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (System Demonstrations)", "volume": null}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:35Z", "role": "primary", "sha256": "b3e67984f59a1b05ccd946eb54e2d37b0bb78ab0ca796fbb6d544793dafea3e0", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2410.00467%2C2410.00689%2C2410.02810%2C2410.02907%2C2410.05243%2C2410.06703%2C2410.08164%2C2410.09006%2C2410.09024%2C2410.11843%2C2410.11871%2C2410.11872%2C2410.13232%2C2410.13757%2C2410.13824%2C2410.13825%2C2410.13886%2C2410.14803%2C2410.15164%2C2410.16464%2C2410.17236%2C2410.17401%2C2410.17520%2C2410.17883%2C2410.18603%2C2410.18963%2C2410.18967%2C2410.19054%2C2410.19100%2C2410.19461%2C2410.19609%2C2410.20745%2C2410.22552%2C2410.23218%2C2410.24024%2C2411.00820%2C2411.02337%2C2411.02391%2C2411.04468%2C2411.04890%2C2411.06559%2C2411.07618%2C2411.10323%2C2411.13451%2C2411.13591%2C2411.15004%2C2411.17465%2C2411.18279%2C2412.00402%2C2412.01268&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:09:53Z", "role": "primary", "sha256": "a873b34264be14f64599c0897219ff2796ab74d111df8a2fdc667fd22ad87028", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.publisher", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue"], "url": "https://aclanthology.org/2025.naacl-demo.43/"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:35Z", "role": "primary", "sha256": "b3e67984f59a1b05ccd946eb54e2d37b0bb78ab0ca796fbb6d544793dafea3e0", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2410.00467%2C2410.00689%2C2410.02810%2C2410.02907%2C2410.05243%2C2410.06703%2C2410.08164%2C2410.09006%2C2410.09024%2C2410.11843%2C2410.11871%2C2410.11872%2C2410.13232%2C2410.13757%2C2410.13824%2C2410.13825%2C2410.13886%2C2410.14803%2C2410.15164%2C2410.16464%2C2410.17236%2C2410.17401%2C2410.17520%2C2410.17883%2C2410.18603%2C2410.18963%2C2410.18967%2C2410.19054%2C2410.19100%2C2410.19461%2C2410.19609%2C2410.20745%2C2410.22552%2C2410.23218%2C2410.24024%2C2411.00820%2C2411.02337%2C2411.02391%2C2411.04468%2C2411.04890%2C2411.06559%2C2411.07618%2C2411.10323%2C2411.13451%2C2411.13591%2C2411.15004%2C2411.17465%2C2411.18279%2C2412.00402%2C2412.01268&max_results=50"}], "id": "arxiv:2410.13757", "identifiers": {"arxiv": "2410.13757"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:35Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2410.13757"}, {"rel": "publication", "url": "https://aclanthology.org/2025.naacl-demo.43/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2025.naacl-demo.43.pdf"}], "name": "MobA: Multifaceted Memory-Enhanced Adaptive Planning for Efficient Mobile Task Automation", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-10-17", "relations": [], "review_status": "curated", "scope": "core", "summary": "Combines adaptive mobile-task planning, reflection-based recovery, and multiple memory components, evaluated with MobBench and AndroidArena.", "topics": ["memory", "planning", "mobile-navigation"], "updated_at": "2025-05-13", "url": "https://arxiv.org/abs/2410.13757"}
{"aliases": [], "authors": ["Lawrence Jang", "Yinheng Li", "Dan Zhao", "Charles Ding", "Justin Lin", "Paul Pu Liang", "Rogerio Bonatti", "Kazuhito Koishida"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6f413943bd5c2a395ddb5cf38a1a5ad61754b35a9eb5de43e8d980bd83e4adff", "submitted_at": "2024-10-24T19:03:01Z", "updated_at": "2025-02-15T05:19:38Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2410.19100v3"}, "publication": {"authors": ["Lawrence Jang", "Yinheng Li", "Dan Zhao", "Charles Ding", "Justin Lin", "Paul Pu Liang", "Rogerio Bonatti", "Kazuhito Koishida"], "date_basis": "Explicit publisher citation_publication_date: 2025-05-01. Preserve supplied precision; do not substitute conference event dates.", "date_parts": [2025, 5, 1], "doi": null, "event_dates": null, "evidence_urls": ["https://proceedings.iclr.cc/paper_files/paper/2025/hash/5b555804d495321df2e3208cc27f4fbc-Abstract-Conference.html"], "identity_basis": "Same work: normalized title, ordered author correspondence and the specific contribution in the complete publisher abstract were reviewed against the retained arXiv identity.", "issue": null, "notes": "Publisher proceedings edition; retained separately from the arXiv snapshot.", "pages": "36934-36958", "pdf_url": "https://proceedings.iclr.cc/paper_files/paper/2025/file/5b555804d495321df2e3208cc27f4fbc-Paper-Conference.pdf", "publisher": null, "series": null, "title": "VideoWebArena:  Evaluating Long Context Multimodal Agents with Video Understanding Web Tasks", "type": "conference-paper", "url": "https://proceedings.iclr.cc/paper_files/paper/2025/hash/5b555804d495321df2e3208cc27f4fbc-Abstract-Conference.html", "venue": "International Conference on Learning Representations", "volume": "2025"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:35Z", "role": "primary", "sha256": "b3e67984f59a1b05ccd946eb54e2d37b0bb78ab0ca796fbb6d544793dafea3e0", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2410.00467%2C2410.00689%2C2410.02810%2C2410.02907%2C2410.05243%2C2410.06703%2C2410.08164%2C2410.09006%2C2410.09024%2C2410.11843%2C2410.11871%2C2410.11872%2C2410.13232%2C2410.13757%2C2410.13824%2C2410.13825%2C2410.13886%2C2410.14803%2C2410.15164%2C2410.16464%2C2410.17236%2C2410.17401%2C2410.17520%2C2410.17883%2C2410.18603%2C2410.18963%2C2410.18967%2C2410.19054%2C2410.19100%2C2410.19461%2C2410.19609%2C2410.20745%2C2410.22552%2C2410.23218%2C2410.24024%2C2411.00820%2C2411.02337%2C2411.02391%2C2411.04468%2C2411.04890%2C2411.06559%2C2411.07618%2C2411.10323%2C2411.13451%2C2411.13591%2C2411.15004%2C2411.17465%2C2411.18279%2C2412.00402%2C2412.01268&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:10:02Z", "role": "primary", "sha256": "fdb42289bd00ad80cb51b66a01b91e13e60f134e809bb6c46181b4024ccabad2", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue", "details.publication.volume"], "url": "https://proceedings.iclr.cc/paper_files/paper/2025/hash/5b555804d495321df2e3208cc27f4fbc-Abstract-Conference.html"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:35Z", "role": "primary", "sha256": "b3e67984f59a1b05ccd946eb54e2d37b0bb78ab0ca796fbb6d544793dafea3e0", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2410.00467%2C2410.00689%2C2410.02810%2C2410.02907%2C2410.05243%2C2410.06703%2C2410.08164%2C2410.09006%2C2410.09024%2C2410.11843%2C2410.11871%2C2410.11872%2C2410.13232%2C2410.13757%2C2410.13824%2C2410.13825%2C2410.13886%2C2410.14803%2C2410.15164%2C2410.16464%2C2410.17236%2C2410.17401%2C2410.17520%2C2410.17883%2C2410.18603%2C2410.18963%2C2410.18967%2C2410.19054%2C2410.19100%2C2410.19461%2C2410.19609%2C2410.20745%2C2410.22552%2C2410.23218%2C2410.24024%2C2411.00820%2C2411.02337%2C2411.02391%2C2411.04468%2C2411.04890%2C2411.06559%2C2411.07618%2C2411.10323%2C2411.13451%2C2411.13591%2C2411.15004%2C2411.17465%2C2411.18279%2C2412.00402%2C2412.01268&max_results=50"}], "id": "arxiv:2410.19100", "identifiers": {"arxiv": "2410.19100"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:35Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2410.19100"}, {"rel": "publication", "url": "https://proceedings.iclr.cc/paper_files/paper/2025/hash/5b555804d495321df2e3208cc27f4fbc-Abstract-Conference.html"}, {"rel": "publication_pdf", "url": "https://proceedings.iclr.cc/paper_files/paper/2025/file/5b555804d495321df2e3208cc27f4fbc-Paper-Conference.pdf"}], "name": "VideoWebArena: Evaluating Long Context Multimodal Agents with Video Understanding Web Tasks", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-10-24", "relations": [], "review_status": "curated", "scope": "core", "summary": "Evaluates whether web agents can retain procedural skills and factual information from long video tutorials and use them during tasks.", "topics": ["evaluation", "video", "memory"], "updated_at": "2025-02-15", "url": "https://arxiv.org/abs/2410.19100"}
{"aliases": [], "authors": ["Yuhao Yang", "Yue Wang", "Dongxu Li", "Ziyang Luo", "Bei Chen", "Chao Huang", "Junnan Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "9b49d45aa539c224846a7e14cee9b537d032228e19e86a9cb3cdd9f5004aa429", "submitted_at": "2024-12-20T07:16:57Z", "updated_at": "2025-07-08T08:49:17Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2412.16256v2"}, "publication": {"authors": ["Yuhao Yang", "Yue Wang", "Dongxu Li", "Ziyang Luo", "Bei Chen", "Chao Huang", "Junnan Li"], "date_basis": "Explicit publisher citation_publication_date: 2025/7. Preserve supplied precision; do not substitute conference event dates.", "date_parts": [2025, 7], "doi": "10.18653/v1/2025.findings-acl.1152", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2025.findings-acl.1152/"], "identity_basis": "Same work: normalized title, ordered author correspondence and the specific contribution in the complete publisher abstract were reviewed against the retained arXiv identity.", "issue": null, "notes": "Publisher proceedings edition; retained separately from the arXiv snapshot.", "pages": "22418-22433", "pdf_url": "https://aclanthology.org/2025.findings-acl.1152.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "Aria-UI: Visual Grounding for GUI Instructions", "type": "conference-paper", "url": "https://aclanthology.org/2025.findings-acl.1152/", "venue": "Findings of the Association for Computational Linguistics: ACL 2025", "volume": null}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2412.04454%2C2412.04626%2C2412.05467%2C2412.09362%2C2412.09605%2C2412.10342%2C2412.10742%2C2412.10840%2C2412.13103%2C2412.13194%2C2412.13501%2C2412.14161%2C2412.14470%2C2412.16256%2C2412.17589%2C2412.18116%2C2412.18426%2C2412.19723%2C2501.01149%2C2501.04575%2C2501.07572%2C2501.11733%2C2501.12326%2C2501.13896%2C2501.16150%2C2502.01600%2C2502.01977%2C2502.02982%2C2502.06395%2C2502.07949%2C2502.08047%2C2502.08226%2C2502.11357%2C2502.13053%2C2502.13923%2C2502.14282%2C2502.18906%2C2503.00401%2C2503.02268%2C2503.03196%2C2503.04730%2C2503.05143%2C2503.06470%2C2503.09263%2C2503.11069%2C2503.11170%2C2503.12687%2C2503.15661%2C2503.15937%2C2503.16465&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:09:57Z", "role": "primary", "sha256": "e24a63769d700caa2d27341dd2634b990cdbc53c7cf557817b3679dc63faef6f", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.publisher", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue"], "url": "https://aclanthology.org/2025.findings-acl.1152/"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2412.04454%2C2412.04626%2C2412.05467%2C2412.09362%2C2412.09605%2C2412.10342%2C2412.10742%2C2412.10840%2C2412.13103%2C2412.13194%2C2412.13501%2C2412.14161%2C2412.14470%2C2412.16256%2C2412.17589%2C2412.18116%2C2412.18426%2C2412.19723%2C2501.01149%2C2501.04575%2C2501.07572%2C2501.11733%2C2501.12326%2C2501.13896%2C2501.16150%2C2502.01600%2C2502.01977%2C2502.02982%2C2502.06395%2C2502.07949%2C2502.08047%2C2502.08226%2C2502.11357%2C2502.13053%2C2502.13923%2C2502.14282%2C2502.18906%2C2503.00401%2C2503.02268%2C2503.03196%2C2503.04730%2C2503.05143%2C2503.06470%2C2503.09263%2C2503.11069%2C2503.11170%2C2503.12687%2C2503.15661%2C2503.15937%2C2503.16465&max_results=50"}], "id": "arxiv:2412.16256", "identifiers": {"arxiv": "2412.16256"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:38Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2412.16256"}, {"rel": "publication", "url": "https://aclanthology.org/2025.findings-acl.1152/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2025.findings-acl.1152.pdf"}], "name": "Aria-UI: Visual Grounding for GUI Instructions", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-12-20", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces a vision-based grounding model trained with diverse synthetic instructions and action histories for context-sensitive GUI targeting.", "topics": ["grounding", "memory"], "updated_at": "2025-07-08", "url": "https://arxiv.org/abs/2412.16256"}
{"aliases": [], "authors": ["Zhenhailong Wang", "Haiyang Xu", "Junyang Wang", "Xi Zhang", "Ming Yan", "Ji Zhang", "Fei Huang", "Heng Ji"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "5cf28f084cac7fc38e5c2dc4654115356d2d57f16009aad8925e071b8005be9c", "submitted_at": "2025-01-20T20:35:46Z", "updated_at": "2025-01-28T16:58:02Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2501.11733v2"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2412.04454%2C2412.04626%2C2412.05467%2C2412.09362%2C2412.09605%2C2412.10342%2C2412.10742%2C2412.10840%2C2412.13103%2C2412.13194%2C2412.13501%2C2412.14161%2C2412.14470%2C2412.16256%2C2412.17589%2C2412.18116%2C2412.18426%2C2412.19723%2C2501.01149%2C2501.04575%2C2501.07572%2C2501.11733%2C2501.12326%2C2501.13896%2C2501.16150%2C2502.01600%2C2502.01977%2C2502.02982%2C2502.06395%2C2502.07949%2C2502.08047%2C2502.08226%2C2502.11357%2C2502.13053%2C2502.13923%2C2502.14282%2C2502.18906%2C2503.00401%2C2503.02268%2C2503.03196%2C2503.04730%2C2503.05143%2C2503.06470%2C2503.09263%2C2503.11069%2C2503.11170%2C2503.12687%2C2503.15661%2C2503.15937%2C2503.16465&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2412.04454%2C2412.04626%2C2412.05467%2C2412.09362%2C2412.09605%2C2412.10342%2C2412.10742%2C2412.10840%2C2412.13103%2C2412.13194%2C2412.13501%2C2412.14161%2C2412.14470%2C2412.16256%2C2412.17589%2C2412.18116%2C2412.18426%2C2412.19723%2C2501.01149%2C2501.04575%2C2501.07572%2C2501.11733%2C2501.12326%2C2501.13896%2C2501.16150%2C2502.01600%2C2502.01977%2C2502.02982%2C2502.06395%2C2502.07949%2C2502.08047%2C2502.08226%2C2502.11357%2C2502.13053%2C2502.13923%2C2502.14282%2C2502.18906%2C2503.00401%2C2503.02268%2C2503.03196%2C2503.04730%2C2503.05143%2C2503.06470%2C2503.09263%2C2503.11069%2C2503.11170%2C2503.12687%2C2503.15661%2C2503.15937%2C2503.16465&max_results=50"}], "id": "arxiv:2501.11733", "identifiers": {"arxiv": "2501.11733"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:38Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2501.11733"}], "name": "Mobile-Agent-E: Self-Evolving Mobile Assistant for Complex Tasks", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-01-20", "relations": [], "review_status": "curated", "scope": "core", "summary": "Uses hierarchical mobile agents and persistent tips and reusable action shortcuts to improve complex tasks from accumulated experience.", "topics": ["mobile-navigation", "memory", "long-horizon"], "updated_at": "2025-01-28", "url": "https://arxiv.org/abs/2501.11733"}
{"aliases": [], "authors": ["Junyang Wang", "Haiyang Xu", "Xi Zhang", "Ming Yan", "Ji Zhang", "Fei Huang", "Jitao Sang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d723cb764deb78b194ff0199c7f1b921a12b6a41655c07b06ae88c3d0af58358", "submitted_at": "2025-02-24T12:51:23Z", "updated_at": "2025-06-03T08:37:03Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2502.17110v3"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:45:04Z", "role": "primary", "sha256": "bab88dfed88ca440c231aa768530374a0f19498777a70921481a904fddd46bb8", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2303.17760%2C2310.03720%2C2312.11190%2C2403.02502%2C2406.12639%2C2409.18060%2C2410.00201%2C2501.01702%2C2501.11425%2C2502.02533%2C2502.02955%2C2502.12110%2C2502.15760%2C2502.17110%2C2502.19328%2C2503.02197%2C2503.02403%2C2503.02682%2C2503.06580%2C2503.07365%2C2503.09572%2C2503.14021%2C2503.24180%2C2504.02357%2C2504.11675%2C2504.13936&max_results=26"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:45:04Z", "role": "primary", "sha256": "bab88dfed88ca440c231aa768530374a0f19498777a70921481a904fddd46bb8", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2303.17760%2C2310.03720%2C2312.11190%2C2403.02502%2C2406.12639%2C2409.18060%2C2410.00201%2C2501.01702%2C2501.11425%2C2502.02533%2C2502.02955%2C2502.12110%2C2502.15760%2C2502.17110%2C2502.19328%2C2503.02197%2C2503.02403%2C2503.02682%2C2503.06580%2C2503.07365%2C2503.09572%2C2503.14021%2C2503.24180%2C2504.02357%2C2504.11675%2C2504.13936&max_results=26"}], "id": "arxiv:2502.17110", "identifiers": {"arxiv": "2502.17110"}, "kind": "paper", "last_verified_at": "2026-10-16T06:45:04Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2502.17110"}], "name": "Mobile-Agent-V: A Video-Guided Approach for Effortless and Efficient Operational Knowledge Injection in Mobile Automation", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced. The authors identify arXiv:2505.13887 as an accidental duplicate and direct future updates to this record.", "published_at": "2025-02-24", "relations": [], "review_status": "curated", "scope": "core", "summary": "Extracts operational guidance from video for mobile task execution and introduces Mobile-Knowledge to evaluate the contribution of external knowledge.", "topics": ["mobile-navigation", "video", "memory"], "updated_at": "2025-06-03", "url": "https://arxiv.org/abs/2502.17110"}
{"aliases": [], "authors": ["Wenjia Jiang", "Yangyang Zhuang", "Chenxi Song", "Xu Yang", "Joey Tianyi Zhou", "Chi Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d396e4e249fc2aa1f4d53eadac38c3c814aae3f0208921b3388d5b009e6037b3", "submitted_at": "2025-03-04T04:34:09Z", "updated_at": "2025-04-15T02:32:21Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2503.02268v3"}}, "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:2503.02268", "identifiers": {"arxiv": "2503.02268"}, "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/2503.02268"}], "name": "AppAgentX: Evolving GUI Agents as Proficient Smartphone 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": "2025-03-04", "relations": [], "review_status": "curated", "scope": "core", "summary": "Records interaction history and converts repeated action sequences into reusable high-level shortcuts for mobile GUI tasks.", "topics": ["memory", "efficiency"], "updated_at": "2025-04-15", "url": "https://arxiv.org/abs/2503.02268"}
{"aliases": [], "authors": ["Weimin Xiong", "Yifan Song", "Qingxiu Dong", "Bingchan Zhao", "Feifan Song", "Xun Wang", "Sujian Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "3395107d8c7db3562bf75a3c22fee55d3da0564bd2f413491bc0d381fc601323", "submitted_at": "2025-03-04T14:54:45Z", "updated_at": "2025-09-10T16:45:42Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2503.02682v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:45:04Z", "role": "primary", "sha256": "bab88dfed88ca440c231aa768530374a0f19498777a70921481a904fddd46bb8", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2303.17760%2C2310.03720%2C2312.11190%2C2403.02502%2C2406.12639%2C2409.18060%2C2410.00201%2C2501.01702%2C2501.11425%2C2502.02533%2C2502.02955%2C2502.12110%2C2502.15760%2C2502.17110%2C2502.19328%2C2503.02197%2C2503.02403%2C2503.02682%2C2503.06580%2C2503.07365%2C2503.09572%2C2503.14021%2C2503.24180%2C2504.02357%2C2504.11675%2C2504.13936&max_results=26"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:45:04Z", "role": "primary", "sha256": "bab88dfed88ca440c231aa768530374a0f19498777a70921481a904fddd46bb8", "supports": ["name", "kind", "summary", "topics"], "url": "https://export.arxiv.org/api/query?id_list=2303.17760%2C2310.03720%2C2312.11190%2C2403.02502%2C2406.12639%2C2409.18060%2C2410.00201%2C2501.01702%2C2501.11425%2C2502.02533%2C2502.02955%2C2502.12110%2C2502.15760%2C2502.17110%2C2502.19328%2C2503.02197%2C2503.02403%2C2503.02682%2C2503.06580%2C2503.07365%2C2503.09572%2C2503.14021%2C2503.24180%2C2504.02357%2C2504.11675%2C2504.13936&max_results=26"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T13:06:43Z", "role": "primary", "sha256": "e634a3e576c94f1f7f5409d28a5c4c80c46141f1763c81337bb45f19793815f6", "supports": ["environments", "scope", "notes"], "url": "https://arxiv.org/html/2503.02682"}], "id": "arxiv:2503.02682", "identifiers": {"arxiv": "2503.02682"}, "kind": "paper", "last_verified_at": "2026-10-16T06:45:04Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2503.02682"}], "name": "MPO: Boosting LLM Agents with Meta Plan Optimization", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced. The full paper was also inspected to confirm WebShop evaluation; the broader agent-learning method is classified as adjacent.", "published_at": "2025-03-04", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "Refines reusable high-level meta plans with execution feedback to guide language-agent planning across interactive environments.", "topics": ["planning", "memory"], "updated_at": "2025-09-10", "url": "https://arxiv.org/abs/2503.02682"}
{"aliases": [], "authors": ["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": ["Yiqiao Jin", "Stefano Petrangeli", "Yu Shen", "Gang Wu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "0888d0da1c5da9cf9956fd8b145a09f463aa432a60b142c4aea8d20b4aa84c4d", "submitted_at": "2025-03-26T20:41:24Z", "updated_at": "2025-03-26T20:41:24Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2503.20978v1"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:42Z", "role": "primary", "sha256": "afe4e75ffb8a5c44f4c088a90ffa5f53553036f514aa768f5aee0c5aa38d731e", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2503.16788%2C2503.17709%2C2503.18492%2C2503.19537%2C2503.20978%2C2503.21620%2C2503.23434%2C2504.01382%2C2504.07981%2C2504.10445%2C2504.10458%2C2504.11257%2C2504.11281%2C2504.12679%2C2504.13805%2C2504.13865%2C2504.14239%2C2504.18575%2C2504.19838%2C2504.20464%2C2505.10924%2C2505.11891%2C2505.12299%2C2505.12632%2C2505.13227%2C2505.13887%2C2505.13909%2C2505.14289%2C2505.15810%2C2505.16282%2C2505.16827%2C2505.18829%2C2505.19897%2C2505.21496%2C2505.21499%2C2505.21936%2C2505.22648%2C2505.23762%2C2505.24838%2C2506.00618%2C2506.01391%2C2506.01968%2C2506.02456%2C2506.03143%2C2506.03569%2C2506.04135%2C2506.04614%2C2506.07672%2C2506.08012%2C2506.08972&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:42Z", "role": "primary", "sha256": "afe4e75ffb8a5c44f4c088a90ffa5f53553036f514aa768f5aee0c5aa38d731e", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2503.16788%2C2503.17709%2C2503.18492%2C2503.19537%2C2503.20978%2C2503.21620%2C2503.23434%2C2504.01382%2C2504.07981%2C2504.10445%2C2504.10458%2C2504.11257%2C2504.11281%2C2504.12679%2C2504.13805%2C2504.13865%2C2504.14239%2C2504.18575%2C2504.19838%2C2504.20464%2C2505.10924%2C2505.11891%2C2505.12299%2C2505.12632%2C2505.13227%2C2505.13887%2C2505.13909%2C2505.14289%2C2505.15810%2C2505.16282%2C2505.16827%2C2505.18829%2C2505.19897%2C2505.21496%2C2505.21499%2C2505.21936%2C2505.22648%2C2505.23762%2C2505.24838%2C2506.00618%2C2506.01391%2C2506.01968%2C2506.02456%2C2506.03143%2C2506.03569%2C2506.04135%2C2506.04614%2C2506.07672%2C2506.08012%2C2506.08972&max_results=50"}], "id": "arxiv:2503.20978", "identifiers": {"arxiv": "2503.20978"}, "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/2503.20978"}], "name": "ScreenLLM: Stateful Screen Schema for Efficient Action Understanding and Prediction", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-03-26", "relations": [], "review_status": "curated", "scope": "core", "summary": "Represents interface interactions with a stateful screen schema to model user behavior and support action understanding and prediction.", "topics": ["screen-understanding", "memory"], "updated_at": "2025-03-26", "url": "https://arxiv.org/abs/2503.20978"}
{"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": ["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": ["Bin Xie", "Rui Shao", "Gongwei Chen", "Kaiwen Zhou", "Yinchuan Li", "Jie Liu", "Min Zhang", "Liqiang Nie"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "641001c49fccac1e1c0659fe765eba5a30cefd2a7ff5d00c821f79092c45a1df", "submitted_at": "2025-05-22T16:01:06Z", "updated_at": "2025-05-22T16:01:06Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2505.16827v1"}, "publication": {"authors": ["Bin Xie", "Rui Shao", "Gongwei Chen", "Kaiwen Zhou", "Yinchuan Li", "Jie Liu", "Min Zhang", "Liqiang Nie"], "date_basis": "Explicit publisher citation_publication_date: 2025/7. Preserve supplied precision; do not substitute conference event dates.", "date_parts": [2025, 7], "doi": "10.18653/v1/2025.acl-long.282", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2025.acl-long.282/"], "identity_basis": "Same work: normalized title, ordered author correspondence and the specific contribution in the complete publisher abstract were reviewed against the retained arXiv identity.", "issue": null, "notes": "Publisher proceedings edition; retained separately from the arXiv snapshot.", "pages": "5650-5667", "pdf_url": "https://aclanthology.org/2025.acl-long.282.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "GUI-explorer: Autonomous Exploration and Mining of Transition-aware Knowledge for GUI Agent", "type": "conference-paper", "url": "https://aclanthology.org/2025.acl-long.282/", "venue": "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)", "volume": null}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:09:59Z", "role": "primary", "sha256": "40babce7d310777fa65a498c170320124d4d4941f2cb46ad0ff04aae3583e02f", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.publisher", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue"], "url": "https://aclanthology.org/2025.acl-long.282/"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:54Z", "role": "primary", "sha256": "4953a8e0f029016aef668bce98cec4d2873a7ae49c1fe33c4311fa5eb8d24089", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=700&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2505.16827", "identifiers": {"arxiv": "2505.16827"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2505.16827"}, {"rel": "publication", "url": "https://aclanthology.org/2025.acl-long.282/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2025.acl-long.282.pdf"}], "name": "GUI-explorer: Autonomous Exploration and Mining of Transition-aware Knowledge for GUI Agent", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-05-22", "relations": [], "review_status": "curated", "scope": "core", "summary": "GUI-explorer generates exploration goals from screenshots and interface structure, then mines observation-action-outcome triples for app-specific transition knowledge. The resulting training-free agent uses this knowledge during mobile task execution, evaluated on SPA-Bench and AndroidWorld.", "topics": ["mobile-navigation", "data-generation", "memory", "planning"], "updated_at": "2025-05-22", "url": "https://arxiv.org/abs/2505.16827"}
{"aliases": [], "authors": ["Ziyun Zhang", "Xinyi Liu", "Xiaoyi Zhang", "Jun Wang", "Gang Chen", "Yan Lu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "17c0dac4cbb7c3b7b94496dc225db061a0dc3bda0a34ef40c58ea451524fada7", "submitted_at": "2025-05-28T04:32:05Z", "updated_at": "2025-11-03T08:44:04Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2505.21964v2"}}, "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:2505.21964", "identifiers": {"arxiv": "2505.21964"}, "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/2505.21964"}], "name": "UI-Evol: Automatic Knowledge Evolving for Computer Use Agents", "notes": "Reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2025-05-28", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces UI-Evol, which extracts action sequences from agent interactions and critiques external guidance against those traces. Evaluation on OSWorld examines how revised operational knowledge affects execution and behavioral consistency.", "topics": ["memory", "self-reflection", "robustness"], "updated_at": "2025-11-03", "url": "https://arxiv.org/abs/2505.21964"}
{"aliases": [], "authors": ["Linqiang Guo", "Wei Liu", "Yi Wen Heng", "Tse-Hsun", "Chen", "Yang Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6f0ec81429fb39e56414d3bfe621df6c6adf461e63fc1a1c8886ea2b6850b8ee", "submitted_at": "2025-05-29T16:08:51Z", "updated_at": "2025-11-19T16:48:16Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2505.23596v3"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T07:54:03Z", "role": "primary", "sha256": "13d5969845070cd2877dc51ae5300c2d623e5daef00c8a1552ffa46af3f721a5", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2401.05507%2C2406.07275%2C2409.16427%2C2410.00079%2C2503.07701%2C2503.22738%2C2505.15859%2C2505.17735%2C2505.23596%2C2507.05257%2C2507.06134%2C2507.21953%2C2507.23336%2C2508.06960%2C2508.08997%2C2509.02473%2C2509.23694%2C2510.09116%2C2510.14438%2C2511.18423%2C2512.12818%2C2512.13564%2C2601.10758%2C2601.12449%2C2601.21841%2C2603.01152%2C2603.19993%2C2603.20576%2C2604.04503%2C2604.17562%2C2604.23941%2C2605.16265%2C2606.06448%2C2606.06462%2C2606.29193%2C2607.01647%2C2607.13157&max_results=37"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T07:54:03Z", "role": "primary", "sha256": "13d5969845070cd2877dc51ae5300c2d623e5daef00c8a1552ffa46af3f721a5", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2401.05507%2C2406.07275%2C2409.16427%2C2410.00079%2C2503.07701%2C2503.22738%2C2505.15859%2C2505.17735%2C2505.23596%2C2507.05257%2C2507.06134%2C2507.21953%2C2507.23336%2C2508.06960%2C2508.08997%2C2509.02473%2C2509.23694%2C2510.09116%2C2510.14438%2C2511.18423%2C2512.12818%2C2512.13564%2C2601.10758%2C2601.12449%2C2601.21841%2C2603.01152%2C2603.19993%2C2603.20576%2C2604.04503%2C2604.17562%2C2604.23941%2C2605.16265%2C2606.06448%2C2606.06462%2C2606.29193%2C2607.01647%2C2607.13157&max_results=37"}], "id": "arxiv:2505.23596", "identifiers": {"arxiv": "2505.23596"}, "kind": "paper", "last_verified_at": "2026-10-16T07:54:03Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2505.23596"}], "name": "Agent-SAMA: State-Aware Mobile Assistant", "notes": "Hugging Face discovery used the label MAPLE for this ID; the record follows the current primary arXiv title, Agent-SAMA. The secondary-title discrepancy is not treated as proof of a historical rename. Experiments not independently reproduced.", "published_at": "2025-05-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "Models mobile application execution as a finite-state machine to support planning, verification, and recovery across interface transitions.", "topics": ["planning", "memory", "mobile-navigation"], "updated_at": "2025-11-19", "url": "https://arxiv.org/abs/2505.23596"}
{"aliases": [], "authors": ["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": ["Yuquan Xie", "Zaijing Li", "Rui Shao", "Gongwei Chen", "Kaiwen Zhou", "Yinchuan Li", "Dongmei Jiang", "Liqiang Nie"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "9b353606ce29a5bc60d263e86827a825af46883e25da5bd290b48affa9dd3838", "submitted_at": "2025-06-12T06:21:19Z", "updated_at": "2025-06-12T06:21:19Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2506.10387v1"}}, "environments": ["android", "web", "cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2506.10387", "identifiers": {"arxiv": "2506.10387"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2506.10387"}], "name": "Mirage-1: Augmenting and Updating GUI Agent with Hierarchical Multimodal Skills", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-06-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "Mirage-1 organizes interaction trajectories into execution, core, and meta-skills, then uses them to guide Monte Carlo tree search during online GUI tasks. It introduces AndroidLH for long-horizon evaluation and tests the approach on Android and web-navigation benchmarks.", "topics": ["planning", "memory", "long-horizon", "mobile-navigation", "web-navigation"], "updated_at": "2025-06-12", "url": "https://arxiv.org/abs/2506.10387"}
{"aliases": [], "authors": ["Atharv Singh Patlan", "Ashwin Hebbar", "Pramod Viswanath", "Prateek Mittal"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2d346389013d465f578417a7fca80ee527f5e7b582e93eb73dae148f11479971", "submitted_at": "2025-06-18T14:29:02Z", "updated_at": "2025-06-18T14:29:02Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2506.17318v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2506.17318", "identifiers": {"arxiv": "2506.17318"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2506.17318"}], "name": "Context manipulation attacks : Web agents are susceptible to corrupted memory", "notes": "Reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2025-06-18", "relations": [], "review_status": "curated", "scope": "core", "summary": "Studies context manipulation in web agents by altering stored task representations rather than webpage prompts alone. Evaluation on Browser-use and Agent-E examines memory-directed attacks and the implications for secure context handling.", "topics": ["safety-security", "memory", "evaluation"], "updated_at": "2025-06-18", "url": "https://arxiv.org/abs/2506.17318"}
{"aliases": [], "authors": ["Xinzge Gao", "Chuanrui Hu", "Bin Chen", "Teng Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "aaa62e24a7a269fea49471f17028a3d5d8068d934e450f53bc002ae80fb7cb03", "submitted_at": "2025-06-22T20:17:46Z", "updated_at": "2025-06-22T20:17:46Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2506.18158v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2506.18158", "identifiers": {"arxiv": "2506.18158"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2506.18158"}], "name": "Chain-of-Memory: Enhancing GUI Agents for Cross-Application Navigation", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-06-22", "relations": [], "review_status": "curated", "scope": "core", "summary": "Chain-of-Memory explicitly records action descriptions and task-relevant screen information in short- and long-term memory for GUI navigation. GUI Odyssey-CoM provides 111,000 screen-action pairs with memory annotations for learning and evaluating information retention across applications.", "topics": ["memory", "state-tracking", "cross-app", "data-generation"], "updated_at": "2025-06-22", "url": "https://arxiv.org/abs/2506.18158"}
{"aliases": [], "authors": ["Zihao Sun", "Zijing Shi", "Ling Chen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "1745ec9cf10c465c0f389f885df6db502f1884cca271bccea1b45acd2c787af9", "submitted_at": "2025-07-01T16:43:57Z", "updated_at": "2026-09-24T12:58:22Z", "version": 4, "versioned_url": "https://arxiv.org/abs/2507.00938v4"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2507.00938", "identifiers": {"arxiv": "2507.00938"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2507.00938"}], "name": "WebArxiv: A Reproducible Benchmark for Evaluating Multimodal Web Agents on arXiv 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": "2025-07-01", "relations": [], "review_status": "curated", "scope": "core", "summary": "WebArxiv provides 510 static-snapshot tasks with deterministic answers for scholarly browsing, including constrained paper retrieval, content extraction, and cross-paper comparison. The study also tests dynamic retrieval of interaction history to address repetitive or incomplete agent reasoning.", "topics": ["evaluation", "web-navigation", "memory"], "updated_at": "2026-09-24", "url": "https://arxiv.org/abs/2507.00938"}
{"aliases": [], "authors": ["Yi Kong", "Dianxi Shi", "Guoli Yang", "Zhang ke-di", "Chenlin Huang", "Xiaopeng Li", "Songchang Jin"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "197dbb0e2734f30349260444fca80a8ac830a5d4000a677d6417f46509737475", "submitted_at": "2025-07-29T16:05:32Z", "updated_at": "2025-07-29T16:05:32Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2507.21953v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T07:54:03Z", "role": "primary", "sha256": "13d5969845070cd2877dc51ae5300c2d623e5daef00c8a1552ffa46af3f721a5", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2401.05507%2C2406.07275%2C2409.16427%2C2410.00079%2C2503.07701%2C2503.22738%2C2505.15859%2C2505.17735%2C2505.23596%2C2507.05257%2C2507.06134%2C2507.21953%2C2507.23336%2C2508.06960%2C2508.08997%2C2509.02473%2C2509.23694%2C2510.09116%2C2510.14438%2C2511.18423%2C2512.12818%2C2512.13564%2C2601.10758%2C2601.12449%2C2601.21841%2C2603.01152%2C2603.19993%2C2603.20576%2C2604.04503%2C2604.17562%2C2604.23941%2C2605.16265%2C2606.06448%2C2606.06462%2C2606.29193%2C2607.01647%2C2607.13157&max_results=37"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T07:54:03Z", "role": "primary", "sha256": "13d5969845070cd2877dc51ae5300c2d623e5daef00c8a1552ffa46af3f721a5", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2401.05507%2C2406.07275%2C2409.16427%2C2410.00079%2C2503.07701%2C2503.22738%2C2505.15859%2C2505.17735%2C2505.23596%2C2507.05257%2C2507.06134%2C2507.21953%2C2507.23336%2C2508.06960%2C2508.08997%2C2509.02473%2C2509.23694%2C2510.09116%2C2510.14438%2C2511.18423%2C2512.12818%2C2512.13564%2C2601.10758%2C2601.12449%2C2601.21841%2C2603.01152%2C2603.19993%2C2603.20576%2C2604.04503%2C2604.17562%2C2604.23941%2C2605.16265%2C2606.06448%2C2606.06462%2C2606.29193%2C2607.01647%2C2607.13157&max_results=37"}], "id": "arxiv:2507.21953", "identifiers": {"arxiv": "2507.21953"}, "kind": "paper", "last_verified_at": "2026-10-16T07:54:03Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2507.21953"}], "name": "MapAgent: Trajectory-Constructed Memory-Augmented Planning for Mobile Task Automation", "notes": "Editorial summary based on the primary paper abstract; experiments and released artifacts have not been independently reproduced.", "published_at": "2025-07-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "Builds reusable page memories from past mobile trajectories and retrieves them to support task planning and execution.", "topics": ["memory", "planning", "mobile-navigation"], "updated_at": "2025-07-29", "url": "https://arxiv.org/abs/2507.21953"}
{"aliases": [], "authors": ["Tanvir Bhathal", "Asanshay Gupta"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "67da98d98fec647de9b482fabd8c7861063c0fdb91f1ec303fe8df759ee0f69d", "submitted_at": "2025-08-23T11:02:59Z", "updated_at": "2025-08-23T11:02:59Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2508.16987v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2508.16987", "identifiers": {"arxiv": "2508.16987"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2508.16987"}], "name": "WebSight: A Vision-First Architecture for Robust Web Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-08-23", "relations": [], "review_status": "curated", "scope": "core", "summary": "WebSight uses a vision-language model trained on web-interface examples within a modular agent with planning, action, verification, and episodic memory components. It evaluates element clicking on Showdown Clicks and full browsing tasks on WebVoyager using visual observations.", "topics": ["web-navigation", "grounding", "planning", "memory"], "updated_at": "2025-08-23", "url": "https://arxiv.org/abs/2508.16987"}
{"aliases": [], "authors": ["Xin Wang", "Zhiyao Cui", "Hao Li", "Ya Zeng", "Chenxu Wang", "Ruiqi Song", "Yihang Chen", "Kun Shao", "Qiaosheng Zhang", "Jinzhuo Liu", "Siyue Ren", "Shuyue Hu", "Zhen Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "57d5fc98a2eb2232f237d71b34060df0acef89cfeaa4bc247580a4dda3223e0b", "submitted_at": "2025-08-25T13:57:02Z", "updated_at": "2025-08-25T13:57:02Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2508.18040v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-18T10:01:10.946679+00:00", "role": "primary", "sha256": "594b5261dd2039777d36c46cdb8525d973ad57a154b902ff1a171bcbed2c64cc", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22+OR+ti%3A%22mobile+agent%22+OR+ti%3A%22UI+agent%22+OR+ti%3A%22desktop+agent%22+OR+ti%3A%22screen+grounding%22%29+AND+submittedDate%3A%5B200001010000+TO+202610180000%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:01:10.946679+00:00", "role": "primary", "sha256": "594b5261dd2039777d36c46cdb8525d973ad57a154b902ff1a171bcbed2c64cc", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22+OR+ti%3A%22mobile+agent%22+OR+ti%3A%22UI+agent%22+OR+ti%3A%22desktop+agent%22+OR+ti%3A%22screen+grounding%22%29+AND+submittedDate%3A%5B200001010000+TO+202610180000%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:18:36.598801+00:00", "role": "primary", "sha256": "3c3474b771af7bdd500e6aaa91b83d39ec0d455e59bb83df906e7a94700ad2b7", "supports": ["summary", "scope", "environments", "notes"], "url": "https://arxiv.org/pdf/2508.18040v1"}], "id": "arxiv:2508.18040", "identifiers": {"arxiv": "2508.18040"}, "kind": "paper", "last_verified_at": "2026-10-18T10:01:10.946679+00:00", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2508.18040"}], "name": "PerPilot: Personalizing VLM-based Mobile Agents via Memory and Exploration", "notes": "Personalization depends on the validity of stored or inferred user information and retrieval access. This is instruction completion with interaction, not demonstrated permanent model-weight personalization. No specific operating system or reproduced result is asserted. Basis: complete retained abstract and source-reviewed sections of the exact-version paper (The Framework of PerPilot; Personalization Perception; Personalization Completion; PerInstruct Data Construction); not an entire-paper reading or reproduction claim.", "published_at": "2025-08-25", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces PerPilot and the PerInstruct dataset for completing user-specific mobile instructions through memory retrieval, app exploration, and clarification when missing context remains unresolved.", "topics": ["memory", "planning", "human-in-the-loop", "mobile-navigation"], "updated_at": "2025-08-25", "url": "https://arxiv.org/abs/2508.18040"}
{"aliases": [], "authors": ["Ziyi Guan", "Jason Chun Lok Li", "Zhijian Hou", "Pingping Zhang", "Donglai Xu", "Yuzhi Zhao", "Mengyang Wu", "Jinpeng Chen", "Thanh-Toan Nguyen", "Pengfei Xian", "Wenao Ma", "Shengchao Qin", "Graziano Chesi", "Ngai Wong"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "c22f63bb5de22dd922ba37ea45df9404def017fee492e8ade03c83ab8b4f9b58", "submitted_at": "2025-08-30T05:32:32Z", "updated_at": "2025-08-30T05:32:32Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2509.00366v1"}}, "environments": ["android", "web", "desktop", "cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2509.00366", "identifiers": {"arxiv": "2509.00366"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2509.00366"}], "name": "KG-RAG: Enhancing GUI Agent Decision-Making via Knowledge Graph-Driven Retrieval-Augmented Generation", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-08-30", "relations": [], "review_status": "curated", "scope": "core", "summary": "KG-RAG converts UI transition graphs into a searchable vector store and uses intent-guided retrieval to construct actionable navigation paths. It introduces KG-Android-Bench and KG-Harmony-Bench and evaluates transfer from mobile apps to web and desktop tasks.", "topics": ["memory", "planning", "mobile-navigation", "evaluation"], "updated_at": "2025-08-30", "url": "https://arxiv.org/abs/2509.00366"}
{"aliases": [], "authors": ["Jaewoo Ahn", "Junseo Kim", "Heeseung Yun", "Jaehyeon Son", "Dongmin Park", "Jaewoong Cho", "Gunhee Kim"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "3bf86652aaeb252972f3010be38b75b78dfaa9a8063c480c80d0cea47397c2e7", "submitted_at": "2025-09-01T01:33:16Z", "updated_at": "2025-10-15T10:33:27Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2509.01052v2"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2509.01052", "identifiers": {"arxiv": "2509.01052"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2509.01052"}], "name": "FlashAdventure: A Benchmark for GUI Agents Solving Full Story Arcs in Diverse Adventure Games", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-09-01", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "FlashAdventure evaluates GUI agents on completing story arcs in 34 adventure games. It pairs an automated gameplay judge with COAST, a framework that keeps long-term clue memory to plan and complete sequential objectives.", "topics": ["evaluation", "long-horizon", "memory", "planning"], "updated_at": "2025-10-15", "url": "https://arxiv.org/abs/2509.01052"}
{"aliases": [], "authors": ["Gowen Loo", "Chang Liu", "Qinghong Yin", "Xiang Chen", "Jiawei Chen", "Jingyuan Zhang", "Yu Tian"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a8b2db00f62afe350e55e2266b3627de9b0cf089178b0cf80eb13b06275eee15", "submitted_at": "2025-09-04T05:22:42Z", "updated_at": "2025-09-04T05:22:42Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2509.03891v1"}}, "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:2509.03891", "identifiers": {"arxiv": "2509.03891"}, "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/2509.03891"}], "name": "MobileRAG: Enhancing Mobile Agent with Retrieval-Augmented Generation", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-09-04", "relations": [], "review_status": "curated", "scope": "core", "summary": "MobileRAG combines retrieval and memory components to support mobile tasks that need external knowledge or long action sequences. MobileRAG-Eval is introduced to study these requirements on complex real-world app tasks.", "topics": ["memory", "mobile-navigation", "evaluation", "long-horizon"], "updated_at": "2025-09-04", "url": "https://arxiv.org/abs/2509.03891"}
{"aliases": [], "authors": ["Liangtao Lin", "Zhaomeng Zhu", "Tianwei Zhang", "Yonggang Wen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2c1a4019df77278943829e16b155771abb1c8ca30726044cf5969227e0437f5f", "submitted_at": "2025-09-17T05:14:11Z", "updated_at": "2025-09-17T05:14:11Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2509.13704v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2509.13704", "identifiers": {"arxiv": "2509.13704"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2509.13704"}], "name": "InfraMind: A Novel Exploration-based GUI Agentic Framework for Mission-critical Industrial Management", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-09-17", "relations": [], "review_status": "curated", "scope": "core", "summary": "InfraMind explores industrial management GUIs using virtual-machine snapshots and builds memory for planning and hierarchical state recognition. Its framework also studies knowledge distillation and safety checks on open-source and commercial data-center management platforms.", "topics": ["industry", "planning", "memory", "safety-security"], "updated_at": "2025-09-17", "url": "https://arxiv.org/abs/2509.13704"}
{"aliases": [], "authors": ["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": ["Ran Xu", "Kaixin Ma", "Wenhao Yu", "Hongming Zhang", "Joyce C. Ho", "Carl Yang", "Dong Yu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e1c6fd3e84212bac4c4c8341ffb3478c97becce08f92dac89e5de77ad42cd07d", "submitted_at": "2025-09-29T02:04:20Z", "updated_at": "2025-09-29T02:04:20Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2509.24183v1"}}, "environments": ["web", "android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:20:58+00:00", "role": "primary", "sha256": "fe9f70994c119a85cb8075461174e9699c423d79094e611f1ac95a51666600c1", "supports": ["summary", "environments", "scope"], "url": "https://arxiv.org/html/2509.24183v1"}], "id": "arxiv:2509.24183", "identifiers": {"arxiv": "2509.24183"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2509.24183"}], "name": "Retrieval-augmented GUI Agents with Generative Guidelines", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-09-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "RAG-GUI trains a small guidance adapter to filter and summarize retrieved web tutorials while keeping the acting agent fixed. Supervised and rejection-sampling stages prepare the adapter for inference-time assistance, tested on offline web/mobile tasks and interactive Android tasks.", "topics": ["memory", "planning", "imitation-learning", "web-navigation", "mobile-navigation"], "updated_at": "2025-09-29", "url": "https://arxiv.org/abs/2509.24183"}
{"aliases": [], "authors": ["Siru Ouyang", "Jun Yan", "I-Hung Hsu", "Yanfei Chen", "Ke Jiang", "Zifeng Wang", "Rujun Han", "Long T. Le", "Samira Daruki", "Xiangru Tang", "Vishy Tirumalashetty", "George Lee", "Mahsan Rofouei", "Hangfei Lin", "Jiawei Han", "Chen-Yu Lee", "Tomas Pfister"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d984883b65895d02debd54edbd198f0aadf8bd295da869680c3ebbe08d2e9492", "submitted_at": "2025-09-29T17:51:03Z", "updated_at": "2026-03-16T20:49:28Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2509.25140v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-18T10:54:12Z", "role": "primary", "sha256": "5f5bae19420f79231516d4bcf604ea18d4f930fc7790cde5a3e7fcb8d41a1126", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2508.19461%2C2508.19493%2C2509.21782%2C2509.25140%2C2510.03485%2C2510.07809%2C2510.14276%2C2510.27140%2C2510.27210%2C2511.13248%2C2512.08417%2C2601.07263%2C2602.13516%2C2603.00476%2C2603.11862%2C2603.18762%2C2603.23791%2C2604.00986%2C2604.16515%2C2604.16548%2C2604.24348%2C2605.05509%2C2605.08876%2C2605.20833%2C2605.24785%2C2607.00333%2C2607.18659%2C2607.25255%2C2608.00202%2C2608.08939%2C2608.24913%2C2608.29596&max_results=32"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:54:12Z", "role": "primary", "sha256": "5f5bae19420f79231516d4bcf604ea18d4f930fc7790cde5a3e7fcb8d41a1126", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2508.19461%2C2508.19493%2C2509.21782%2C2509.25140%2C2510.03485%2C2510.07809%2C2510.14276%2C2510.27140%2C2510.27210%2C2511.13248%2C2512.08417%2C2601.07263%2C2602.13516%2C2603.00476%2C2603.11862%2C2603.18762%2C2603.23791%2C2604.00986%2C2604.16515%2C2604.16548%2C2604.24348%2C2605.05509%2C2605.08876%2C2605.20833%2C2605.24785%2C2607.00333%2C2607.18659%2C2607.25255%2C2608.00202%2C2608.08939%2C2608.24913%2C2608.29596&max_results=32"}], "id": "arxiv:2509.25140", "identifiers": {"arxiv": "2509.25140"}, "kind": "paper", "last_verified_at": "2026-10-18T10:54:12Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2509.25140"}], "name": "ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-09-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "Distills generalizable reasoning strategies from both successful and failed agent trajectories into a persistent memory bank for test-time retrieval.", "topics": ["memory", "self-reflection", "web-navigation"], "updated_at": "2026-03-16", "url": "https://arxiv.org/abs/2509.25140"}
{"aliases": [], "authors": ["Zikang Liu", "Junyi Li", "Wayne Xin Zhao", "Dawei Gao", "Yaliang Li", "Ji-rong Wen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "277315ecc1a341c0a511bce84b87190fc0aeeb2306666ea8d4746aa8e9b39984", "submitted_at": "2025-10-01T01:48:39Z", "updated_at": "2025-10-04T15:23:55Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2510.00413v2"}}, "environments": ["web", "cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2510.00413", "identifiers": {"arxiv": "2510.00413"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2510.00413"}], "name": "PAL-UI: Planning with Active Look-back for Vision-Based GUI Agents", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-10-01", "relations": [], "review_status": "curated", "scope": "core", "summary": "PAL-UI combines observation/action summaries with selective retrieval of historical screenshots during planning. It trains mobile navigation models from step-level supervision and evaluates transfer to web navigation without additional training.", "topics": ["memory", "planning", "interaction-history", "mobile-navigation"], "updated_at": "2025-10-04", "url": "https://arxiv.org/abs/2510.00413"}
{"aliases": [], "authors": ["Hadi Nekoei", "Aman Jaiswal", "Patrice Bechard", "Oleh Shliazhko", "Orlando Marquez Ayala", "Mathieu Reymond", "Massimo Caccia", "Alexandre Drouin", "Sarath Chandar", "Alexandre Lacoste"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "0d4bd517ada9303e765827b3badfa2ee432bc5511ff4ee0cc42d937735e1c8de", "submitted_at": "2025-10-05T21:34:42Z", "updated_at": "2026-02-20T19:34:32Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2510.04373v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2510.04373", "identifiers": {"arxiv": "2510.04373"}, "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/2510.04373"}], "name": "JEF-Hinter: Leveraging Offline Knowledge for Improving Web Agents Adaptation", "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": "2025-10-05", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces JEF-Hinter, which distills successful and failed trajectories into compact episodic hints and retrieves them for the current agent state. Web benchmark experiments examine guidance from decisive steps rather than full demonstration replay.", "topics": ["memory", "interaction-history", "web-navigation"], "updated_at": "2026-02-20", "url": "https://arxiv.org/abs/2510.04373"}
{"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": ["Wenyi Wu", "Kun Zhou", "Ruoxin Yuan", "Vivian Yu", "Stephen Wang", "Zhiting Hu", "Biwei Huang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "43c41dfbef8b76bbdccd7c435694cee8788b116db278122f64a22e7498bd5a5a", "submitted_at": "2025-10-10T06:16:45Z", "updated_at": "2025-10-10T06:16:45Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2510.09038v1"}}, "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:2510.09038", "identifiers": {"arxiv": "2510.09038"}, "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/2510.09038"}], "name": "Auto-scaling Continuous Memory for GUI Agent", "notes": "Reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2025-10-10", "relations": [], "review_status": "curated", "scope": "core", "summary": "Encodes GUI trajectories into compact continuous embeddings that can be retrieved directly into an agent’s input. An automated environment-discovery and rollout pipeline expands this memory for long tasks and unfamiliar interfaces.", "topics": ["memory", "interaction-history", "efficiency"], "updated_at": "2025-10-10", "url": "https://arxiv.org/abs/2510.09038"}
{"aliases": [], "authors": ["Tao Yu", "Zhengbo Zhang", "Zhiheng Lyu", "Junhao Gong", "Hongzhu Yi", "Xinming Wang", "Yuxuan Zhou", "Jiabing Yang", "Ping Nie", "Yan Huang", "Wenhu Chen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "fe982af6e7285c21e13dcd4565255b501b401b188da64069d9cfe2fab7891f88", "submitted_at": "2025-10-12T15:43:37Z", "updated_at": "2025-10-14T08:54:57Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2510.10666v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2510.10666", "identifiers": {"arxiv": "2510.10666"}, "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/2510.10666"}], "name": "BrowserAgent: Building Web Agents with Human-Inspired Web Browsing Actions", "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": "2025-10-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "Trains BrowserAgent to use Playwright-based scrolling, clicking and typing directly on webpages for information-seeking tasks. Supervised and rejection fine-tuning are paired with explicit memory for conclusions carried across steps.", "topics": ["web-navigation", "imitation-learning", "memory"], "updated_at": "2025-10-14", "url": "https://arxiv.org/abs/2510.10666"}
{"aliases": [], "authors": ["Kai Mei", "Jiang Guo", "Shuaichen Chang", "Mingwen Dong", "Dongkyu Lee", "Xing Niu", "Jiarong Jiang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "7c37dae372655e8ef556163744e47ed42db3533175ca7d1db7560033774ed2a0", "submitted_at": "2025-10-13T19:52:04Z", "updated_at": "2026-03-09T15:25:16Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2510.11892v2"}}, "environments": ["desktop", "web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2510.11892", "identifiers": {"arxiv": "2510.11892"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2510.11892"}], "name": "R-WoM: Retrieval-augmented World Model For Computer-use Agents", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-10-13", "relations": [], "review_status": "curated", "scope": "core", "summary": "R-WoM evaluates how well world models predict nearby GUI states, align complete procedures, and recognize milestone transitions. It retrieves external tutorials to guide simulated future interactions, with experiments on subsets of OSWorld and WebArena.", "topics": ["world-models", "memory", "planning", "evaluation"], "updated_at": "2026-03-09", "url": "https://arxiv.org/abs/2510.11892"}
{"aliases": [], "authors": ["Mathieu Andreux", "Märt Bakler", "Yanael Barbier", "Hamza Benchekroun", "Emilien Biré", "Antoine Bonnet", "Riaz Bordie", "Nathan Bout", "Matthias Brunel", "Aleix Cambray", "Pierre-Louis Cedoz", "Antoine Chassang", "Gautier Cloix", "Ethan Connelly", "Alexandra Constantinou", "Ramzi De Coster", "Hubert de la Jonquiere", "Aurélien Delfosse", "Maxime Delpit", "Alexis Deprez", "Augustin Derupti", "Mathieu Diaz", "Shannon D'Souza", "Julie Dujardin", "Abai Edmund", "Michael Eickenberg", "Armand Fatalot", "Wissem Felissi", "Isaac Herring", "Xavier Koegler", "Erwan Le Jumeau de Kergaradec", "Aurélien Lac", "Maxime Langevin", "Corentin Lauverjat", "Antonio Loison", "Avshalom Manevich", "Axel Moyal", "Axel Nguyen Kerbel", "Marinela Parovic", "Julien Revelle", "Guillaume Richard", "Mats Richter", "Ronan Riochet", "María Santos", "Romain Savidan", "Laurent Sifre", "Maxime Theillard", "Marc Thibault", "Ivan Valentini", "Tony Wu", "Laura Yie", "Kai Yuan", "Jevgenij Zubovskij"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "de25494af1bc2ceaa760a21da51803aaf89d9d2d1961887cfd0023fb87f28f63", "submitted_at": "2025-10-22T18:21:52Z", "updated_at": "2025-10-24T11:52:29Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2510.19949v2"}}, "environments": ["web", "desktop", "android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2510.19949", "identifiers": {"arxiv": "2510.19949"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2510.19949"}], "name": "Surfer 2: The Next Generation of Cross-Platform Computer Use Agents", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-10-22", "relations": [], "review_status": "curated", "scope": "core", "summary": "Surfer2 combines hierarchical context management with separate planning and execution, self-verification, and recovery for visual computer use. It evaluates the framework across web, desktop, and Android tasks without task-specific fine-tuning.", "topics": ["planning", "memory", "long-horizon", "robustness"], "updated_at": "2025-10-24", "url": "https://arxiv.org/abs/2510.19949"}
{"aliases": [], "authors": ["Yangqin Jiang", "Chao Huang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "09bd73dff01641593f31f7d40effd60df42e3d9460ffa010f76c00a32be3593b", "submitted_at": "2025-10-24T20:23:12Z", "updated_at": "2026-10-05T01:17:12Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2510.22009v3"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:52Z", "role": "primary", "sha256": "8623b0ff7ef9d2e8ba484cd67359e7b8817e78e9ca22b8f27432c2533f5b90a0", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2510.01670%2C2510.04673%2C2510.10073%2C2510.12200%2C2510.14621%2C2510.16051%2C2510.17790%2C2510.18488%2C2510.18596%2C2510.19488%2C2510.19949%2C2510.20286%2C2510.22009%2C2510.24168%2C2510.24411%2C2510.24563%2C2511.00810%2C2511.04307%2C2511.07332%2C2511.09157%2C2511.12254%2C2511.13087%2C2511.15567%2C2511.16590%2C2511.19663%2C2511.20597%2C2511.20766%2C2512.08529%2C2512.10371%2C2512.10962%2C2512.12634%2C2512.14014%2C2512.15431%2C2512.16295%2C2512.16501%2C2512.19432%2C2512.22894%2C2512.24965%2C2601.02399%2C2601.06031%2C2601.07181%2C2601.07262%2C2601.07779%2C2601.08406%2C2601.09923%2C2601.11044%2C2601.11868%2C2601.12822%2C2601.13060%2C2601.15876&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:52Z", "role": "primary", "sha256": "8623b0ff7ef9d2e8ba484cd67359e7b8817e78e9ca22b8f27432c2533f5b90a0", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2510.01670%2C2510.04673%2C2510.10073%2C2510.12200%2C2510.14621%2C2510.16051%2C2510.17790%2C2510.18488%2C2510.18596%2C2510.19488%2C2510.19949%2C2510.20286%2C2510.22009%2C2510.24168%2C2510.24411%2C2510.24563%2C2511.00810%2C2511.04307%2C2511.07332%2C2511.09157%2C2511.12254%2C2511.13087%2C2511.15567%2C2511.16590%2C2511.19663%2C2511.20597%2C2511.20766%2C2512.08529%2C2512.10371%2C2512.10962%2C2512.12634%2C2512.14014%2C2512.15431%2C2512.16295%2C2512.16501%2C2512.19432%2C2512.22894%2C2512.24965%2C2601.02399%2C2601.06031%2C2601.07181%2C2601.07262%2C2601.07779%2C2601.08406%2C2601.09923%2C2601.11044%2C2601.11868%2C2601.12822%2C2601.13060%2C2601.15876&max_results=50"}], "id": "arxiv:2510.22009", "identifiers": {"arxiv": "2510.22009"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:52Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2510.22009"}], "name": "OpenPhone: Mobile Agentic Foundation Models", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-10-24", "relations": [], "review_status": "curated", "scope": "core", "summary": "OpenPhone trains a compact mobile policy with supervised learning followed by GRPO on synthetic GUI data, incorporating reasoning and memory management. Its device-cloud design handles routine work locally and delegates complex subtasks, with online evaluation on AndroidLab.", "topics": ["reinforcement-learning", "memory", "efficiency", "mobile-navigation"], "updated_at": "2026-10-05", "url": "https://arxiv.org/abs/2510.22009"}
{"aliases": [], "authors": ["Weihua Cheng", "Junming Liu", "Yifei Sun", "Botian Shi", "Yirong Chen", "Ding Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d80312fbd8f57cdc53cc10e03f2c68e2152d06945e5089967dceda50c02b9d4b", "submitted_at": "2025-10-28T08:19:58Z", "updated_at": "2026-04-14T04:34:20Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2510.24168v3"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "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:2510.24168", "identifiers": {"arxiv": "2510.24168"}, "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/2510.24168"}], "name": "MGA: Memory-Driven GUI Agent for Observation-Centric Interaction", "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": "2025-10-28", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces MGA, which separates task-agnostic screen-state reading from decision making and connects steps through structured state memory. Validated state deltas replace full trajectory replay, evaluated on OSWorld and application tasks.", "topics": ["memory", "state-tracking", "long-horizon"], "updated_at": "2026-04-14", "url": "https://arxiv.org/abs/2510.24168"}
{"aliases": [], "authors": ["Yujian Liu", "Ze Wang", "Hao Chen", "Ximeng Sun", "Xiaodong Yu", "Jialian Wu", "Jiang Liu", "Emad Barsoum", "Zicheng Liu", "Shiyu Chang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a91413c1e80869860d411402ab49f6e5968787b3bca83f0a9066a0e9c2f75d40", "submitted_at": "2025-11-06T07:29:02Z", "updated_at": "2025-11-06T07:29:02Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2511.04137v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2511.04137", "identifiers": {"arxiv": "2511.04137"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2511.04137"}], "name": "Learning from Online Videos at Inference Time for Computer-Use Agents", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-11-06", "relations": [], "review_status": "curated", "scope": "core", "summary": "This work retrieves online video tutorials and converts them into structured GUI action demonstrations with short objectives. A two-stage selection process supplies a relevant local trajectory as in-context guidance at each execution step.", "topics": ["video", "memory", "planning", "data-generation"], "updated_at": "2025-11-06", "url": "https://arxiv.org/abs/2511.04137"}
{"aliases": [], "authors": ["Ziwei Wang", "Leyang Yang", "Xiaoxuan Tang", "Sheng Zhou", "Dajun Chen", "Wei Jiang", "Yong Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d0f61f09ce52b1e66fa81479483ea419aaa16e6b27ce331d3c1039927075fd91", "submitted_at": "2025-11-12T09:06:25Z", "updated_at": "2025-11-12T09:06:25Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2511.09127v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2511.09127", "identifiers": {"arxiv": "2511.09127"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2511.09127"}], "name": "History-Aware Reasoning for GUI Agents", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-11-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "HAR-GUI combines episodic reasoning and short-term memory to use earlier interactions when selecting GUI actions. Its training constructs reflective scenarios and correction guidance and applies a hybrid reinforcement-learning reward to the 3B policy.", "topics": ["interaction-history", "memory", "self-reflection", "reinforcement-learning"], "updated_at": "2025-11-12", "url": "https://arxiv.org/abs/2511.09127"}
{"aliases": [], "authors": ["Yuxiang Zhou", "Jichang Li", "Yanhao Zhang", "Haonan Lu", "Guanbin Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "ede08df1c10d8eeede43af75d26608b4917dd83becde23f28cd3adf335f79736", "submitted_at": "2025-11-15T15:22:42Z", "updated_at": "2026-03-12T08:45:07Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2511.12254v3"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:52Z", "role": "primary", "sha256": "8623b0ff7ef9d2e8ba484cd67359e7b8817e78e9ca22b8f27432c2533f5b90a0", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2510.01670%2C2510.04673%2C2510.10073%2C2510.12200%2C2510.14621%2C2510.16051%2C2510.17790%2C2510.18488%2C2510.18596%2C2510.19488%2C2510.19949%2C2510.20286%2C2510.22009%2C2510.24168%2C2510.24411%2C2510.24563%2C2511.00810%2C2511.04307%2C2511.07332%2C2511.09157%2C2511.12254%2C2511.13087%2C2511.15567%2C2511.16590%2C2511.19663%2C2511.20597%2C2511.20766%2C2512.08529%2C2512.10371%2C2512.10962%2C2512.12634%2C2512.14014%2C2512.15431%2C2512.16295%2C2512.16501%2C2512.19432%2C2512.22894%2C2512.24965%2C2601.02399%2C2601.06031%2C2601.07181%2C2601.07262%2C2601.07779%2C2601.08406%2C2601.09923%2C2601.11044%2C2601.11868%2C2601.12822%2C2601.13060%2C2601.15876&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:52Z", "role": "primary", "sha256": "8623b0ff7ef9d2e8ba484cd67359e7b8817e78e9ca22b8f27432c2533f5b90a0", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2510.01670%2C2510.04673%2C2510.10073%2C2510.12200%2C2510.14621%2C2510.16051%2C2510.17790%2C2510.18488%2C2510.18596%2C2510.19488%2C2510.19949%2C2510.20286%2C2510.22009%2C2510.24168%2C2510.24411%2C2510.24563%2C2511.00810%2C2511.04307%2C2511.07332%2C2511.09157%2C2511.12254%2C2511.13087%2C2511.15567%2C2511.16590%2C2511.19663%2C2511.20597%2C2511.20766%2C2512.08529%2C2512.10371%2C2512.10962%2C2512.12634%2C2512.14014%2C2512.15431%2C2512.16295%2C2512.16501%2C2512.19432%2C2512.22894%2C2512.24965%2C2601.02399%2C2601.06031%2C2601.07181%2C2601.07262%2C2601.07779%2C2601.08406%2C2601.09923%2C2601.11044%2C2601.11868%2C2601.12822%2C2601.13060%2C2601.15876&max_results=50"}], "id": "arxiv:2511.12254", "identifiers": {"arxiv": "2511.12254"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:52Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2511.12254"}], "name": "Mobile-Agent-RAG: Driving Smart Multi-Agent Coordination with Contextual Knowledge Empowerment for Long-Horizon Mobile 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": "2025-11-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "Uses separate retrieval systems for high-level mobile-task planning and precise interface operations, with a benchmark for extended cross-app tasks.", "topics": ["mobile-navigation", "memory", "long-horizon"], "updated_at": "2026-03-12", "url": "https://arxiv.org/abs/2511.12254"}
{"aliases": [], "authors": ["Genglin Liu", "Shijie Geng", "Sha Li", "Hejie Cui", "Sarah Zhang", "Xin Liu", "Tianyi Liu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "8a105701d9265bacd3971f71f34a0e7ed315b6e54e9a7e675588c50713fcc4de", "submitted_at": "2025-11-17T05:38:50Z", "updated_at": "2026-07-22T21:19:00Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2511.12997v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2511.12997", "identifiers": {"arxiv": "2511.12997"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2511.12997"}], "name": "WebCoach: Self-Evolving Web Agents with Cross-Session Memory Guidance", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2025-11-17", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces WebCoach, which condenses navigation logs into persistent episodic memory and retrieves experiences by similarity and recency. Runtime hooks selectively supply advice to browser agents, evaluated on WebVoyager without retraining the underlying models.", "topics": ["memory", "interaction-history", "web-navigation"], "updated_at": "2026-07-22", "url": "https://arxiv.org/abs/2511.12997"}
{"aliases": [], "authors": ["Xurui Zhou", "Gongwei Chen", "Yuquan Xie", "Zaijing Li", "Kaiwen Zhou", "Shuai Wang", "Shuo Yang", "Zhuotao Tian", "Rui Shao"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "3a6eb58f9b6349ab9cc33444d8ea82c8f17653cc4a7f4b19ffc94e3b4fb894b3", "submitted_at": "2025-12-01T15:06:45Z", "updated_at": "2026-03-08T10:18:14Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2512.01763v2"}}, "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:2512.01763", "identifiers": {"arxiv": "2512.01763"}, "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/2512.01763"}], "name": "HiconAgent: History Context-aware Policy Optimization for GUI Agents", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-12-01", "relations": [], "review_status": "curated", "scope": "core", "summary": "HiconAgent trains GUI policies to use history through variable-length context sampling and paired policy updates with compressed and complete histories. Its compression retains past actions while removing historical observations, and an alignment loss couples the two branches.", "topics": ["interaction-history", "memory", "efficiency"], "updated_at": "2026-03-08", "url": "https://arxiv.org/abs/2512.01763"}
{"aliases": [], "authors": ["Shizuo Tian", "Hao Wen", "Yuxuan Chen", "Jiacheng Liu", "Shanhui Zhao", "Guohong Liu", "Ju Ren", "Yunxin Liu", "Yuanchun Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "72b2db237a52ae31eaba58183a2777e79379aa8e849d18de5d8633daf27ccbf9", "submitted_at": "2025-12-11T07:37:38Z", "updated_at": "2026-05-08T09:40:02Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2512.10371v2"}}, "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", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2512.10371", "identifiers": {"arxiv": "2512.10371"}, "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/2512.10371"}], "name": "AgentProg: Empowering Long-Horizon GUI Agents with Program-Guided Context Management", "notes": "Reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2025-12-11", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces AgentProg, organizing mobile-agent interaction history as a program with variables and control flow to decide what context to retain. A global belief state handles partial observability, with evaluation on AndroidWorld and extended long-horizon tasks.", "topics": ["interaction-history", "memory", "long-horizon"], "updated_at": "2026-05-08", "url": "https://arxiv.org/abs/2512.10371"}
{"aliases": [], "authors": ["Runze Li", "Yuwen Zhai", "Bo Xu", "LiWu Xu", "Nian Shi", "Wei Zhang", "Ran Lin", "Liang Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "03cdf8514fa9f036b4614a880d842d49bb9cb0a821858b7215784fba209faa62", "submitted_at": "2025-12-22T13:42:18Z", "updated_at": "2026-04-10T06:20:00Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2512.19396v3"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:47Z", "role": "primary", "sha256": "ae98e77b8bb125e45277aacec768a047e7bb0e13f33ca9ff6ca7c19e29e56cca", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2512.19396", "identifiers": {"arxiv": "2512.19396"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:47Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2512.19396"}], "name": "EchoTrail-GUI: Building Actionable Memory for GUI Agents via Critic-Guided Self-Exploration", "notes": "The annotation was reviewed against the complete retained primary abstract. Experiments and released artifacts have not been independently reproduced.", "published_at": "2025-12-22", "relations": [], "review_status": "curated", "scope": "core", "summary": "EchoTrail-GUI builds a memory of successful GUI trajectories through autonomous exploration filtered by a reward model. It retrieves relevant trajectories as in-context guidance for new tasks and evaluates the approach on AndroidWorld and AndroidLab.", "topics": ["memory", "data-generation", "reward-models", "mobile-navigation"], "updated_at": "2026-04-10", "url": "https://arxiv.org/abs/2512.19396"}
{"aliases": [], "authors": ["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": ["Bowen Yang", "Kaiming Jin", "Zhenyu Wu", "Zhaoyang Liu", "Qiushi Sun", "Zehao Li", "JingJing Xie", "Zhoumianze Liu", "Fangzhi Xu", "Kanzhi Cheng", "Qingyun Li", "Yian Wang", "Yu Qiao", "Zun Wang", "Zichen Ding"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "cfdaabaf3c6df80eb718e51dc666c54fb2ecf38ec520f6a71ccf5e5ce54248c8", "submitted_at": "2026-01-12T17:55:51Z", "updated_at": "2026-01-12T17:55:51Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2601.07779v1"}}, "environments": ["desktop", "web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2601.07779", "identifiers": {"arxiv": "2601.07779"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2601.07779"}], "name": "OS-Symphony: A Holistic Framework for Robust and Generalist Computer-Using Agent", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-01-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "OS-Symphony coordinates milestone-based reflection and memory with tools that build visually aligned tutorials through browser interaction. The framework evaluates trajectory correction and adaptation on online computer-use benchmarks including OSWorld.", "topics": ["memory", "self-reflection", "planning", "tool-integration"], "updated_at": "2026-01-12", "url": "https://arxiv.org/abs/2601.07779"}
{"aliases": [], "authors": ["Wenyuan Zhang", "Xinghua Zhang", "Haiyang Yu", "Shuaiyi Nie", "Bingli Wu", "Juwei Yue", "Tingwen Liu", "Yongbin Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "73df8c5b417c3387530b7730be3aed8b69374c06fdd3573f56147b19359c6912", "submitted_at": "2026-01-13T14:48:34Z", "updated_at": "2026-04-15T14:59:17Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2601.08605v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2601.08605", "identifiers": {"arxiv": "2601.08605"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2601.08605"}], "name": "ExpSeek: Self-Triggered Experience Seeking for Web Agents", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-01-13", "relations": [], "review_status": "curated", "scope": "core", "summary": "ExpSeek uses step-level entropy signals to decide when a web agent should request experience-based guidance. It supplies guidance tailored to the current step and evaluates the intervention across web-agent benchmarks.", "topics": ["memory", "planning", "web-navigation", "evaluation"], "updated_at": "2026-04-15", "url": "https://arxiv.org/abs/2601.08605"}
{"aliases": [], "authors": ["Yibo Lyu", "Gongwei Chen", "Rui Shao", "Weili Guan", "Liqiang Nie"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d8b697b91942291403491581a61c135c6bfc8b0f3c2a9b9bab0f926af7208fce", "submitted_at": "2026-01-14T17:12:48Z", "updated_at": "2026-05-13T02:54:19Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2601.09636v2"}}, "environments": ["android"], "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:2601.09636", "identifiers": {"arxiv": "2601.09636"}, "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/2601.09636"}], "name": "PersonalAlign: Hierarchical Implicit Intent Alignment for Personalized GUI Agent with Long-Term User-Centric Records", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-01-14", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces AndroidIntent to evaluate how GUI agents infer omitted preferences and recurring routines from long-term user records. Its HIM-Agent maintains hierarchical memory of preferences and routines for personalized execution and proactive suggestions.", "topics": ["memory", "evaluation", "mobile-navigation"], "updated_at": "2026-05-13", "url": "https://arxiv.org/abs/2601.09636"}
{"aliases": [], "authors": ["Mingxian Yu", "Siqi Luo", "Xu Chen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "4ce0909d1c9081f099ae0dc0b0182e0442fd71fd58c62b13d486d154afc2187e", "submitted_at": "2026-01-24T11:35:37Z", "updated_at": "2026-01-24T11:35:37Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2601.17418v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:32:55Z", "role": "primary", "sha256": "e6061c8dd9c00a70fcf1d4325b671c50a6912063ecbe2b6359175ae4622983a0", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2511.13091%2C2511.21398%2C2511.22235%2C2512.00756%2C2512.00846%2C2512.01174%2C2512.01763%2C2512.02423%2C2512.04105%2C2512.05941%2C2512.07725%2C2512.09396%2C2512.11611%2C2512.12692%2C2512.17221%2C2512.19396%2C2512.21302%2C2512.22009%2C2512.22047%2C2512.22322%2C2512.23128%2C2601.02371%2C2601.02439%2C2601.03641%2C2601.04035%2C2601.04126%2C2601.04500%2C2601.05787%2C2601.06899%2C2601.08605%2C2601.09636%2C2601.09770%2C2601.11631%2C2601.12349%2C2601.16356%2C2601.17418%2C2601.18197%2C2601.18305%2C2601.19199%2C2601.20732%2C2601.21352%2C2601.21872%2C2601.21961%2C2601.22149%2C2601.22491%2C2601.22528%2C2602.02468%2C2602.05354%2C2602.06090%2C2602.06351&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:32:55Z", "role": "primary", "sha256": "e6061c8dd9c00a70fcf1d4325b671c50a6912063ecbe2b6359175ae4622983a0", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2511.13091%2C2511.21398%2C2511.22235%2C2512.00756%2C2512.00846%2C2512.01174%2C2512.01763%2C2512.02423%2C2512.04105%2C2512.05941%2C2512.07725%2C2512.09396%2C2512.11611%2C2512.12692%2C2512.17221%2C2512.19396%2C2512.21302%2C2512.22009%2C2512.22047%2C2512.22322%2C2512.23128%2C2601.02371%2C2601.02439%2C2601.03641%2C2601.04035%2C2601.04126%2C2601.04500%2C2601.05787%2C2601.06899%2C2601.08605%2C2601.09636%2C2601.09770%2C2601.11631%2C2601.12349%2C2601.16356%2C2601.17418%2C2601.18197%2C2601.18305%2C2601.19199%2C2601.20732%2C2601.21352%2C2601.21872%2C2601.21961%2C2601.22149%2C2601.22491%2C2601.22528%2C2602.02468%2C2602.05354%2C2602.06090%2C2602.06351&max_results=50"}], "id": "arxiv:2601.17418", "identifiers": {"arxiv": "2601.17418"}, "kind": "paper", "last_verified_at": "2026-10-16T06:32:55Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2601.17418"}], "name": "GraphPilot: GUI Task Automation with One-Step LLM Reasoning Powered by Knowledge Graph", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-01-24", "relations": [], "review_status": "curated", "scope": "core", "summary": "GraphPilot builds app-specific knowledge graphs from explored pages, elements and transitions, then uses them to plan action sequences. It requests additional interface observations when uncertain and checks plans against graph transition rules before correction and execution.", "topics": ["mobile-navigation", "planning", "memory", "efficiency"], "updated_at": "2026-01-24", "url": "https://arxiv.org/abs/2601.17418"}
{"aliases": [], "authors": ["Libo Sun", "Jiwen Zhang", "Siyuan Wang", "Zhongyu Wei"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "1930b4efca7ee38fc6cd1ceb283c14ed9b1388e7b758b951ba9469ed3067c3e0", "submitted_at": "2026-01-27T04:58:48Z", "updated_at": "2026-01-31T15:17:38Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2601.19199v2"}}, "environments": ["android"], "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:2601.19199", "identifiers": {"arxiv": "2601.19199"}, "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/2601.19199"}], "name": "MAGNET: Towards Adaptive GUI Agents with Memory-Driven 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": "2026-01-27", "relations": [], "review_status": "curated", "scope": "core", "summary": "MAGNET maintains separate memories for stable interface functions and task procedures across changing layouts and workflows. It updates frequently used knowledge and evaluates adaptation in AndroidWorld and offline distribution-shift settings.", "topics": ["memory", "grounding", "robustness", "evaluation"], "updated_at": "2026-01-31", "url": "https://arxiv.org/abs/2601.19199"}
{"aliases": [], "authors": ["Sijia Li", "Xiaoyu Tan", "Shahir Ali", "Niels Schmidt", "Gengchen Ma", "Xihe Qiu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "223b9a915e5ea9849644318e1e5426e64ad8655ab46d0b007c30e1967130dc50", "submitted_at": "2026-01-27T07:46:05Z", "updated_at": "2026-01-27T07:46:05Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2601.19306v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-18T10:01:04.629294+00:00", "role": "primary", "sha256": "53521b0a80976320b10efcf7840629e4a2085119710324dd48db2cfd9c5d9cf6", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22+OR+ti%3A%22mobile+agent%22+OR+ti%3A%22UI+agent%22+OR+ti%3A%22desktop+agent%22+OR+ti%3A%22screen+grounding%22%29+AND+submittedDate%3A%5B200001010000+TO+202610180000%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:01:04.629294+00:00", "role": "primary", "sha256": "53521b0a80976320b10efcf7840629e4a2085119710324dd48db2cfd9c5d9cf6", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22+OR+ti%3A%22mobile+agent%22+OR+ti%3A%22UI+agent%22+OR+ti%3A%22desktop+agent%22+OR+ti%3A%22screen+grounding%22%29+AND+submittedDate%3A%5B200001010000+TO+202610180000%5D&start=500&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:14:20.444793+00:00", "role": "primary", "sha256": "ef3bbc8adcc147239ea910e0c59904d7a26ee1610a7a4d1613be2cb845098536", "supports": ["summary", "scope", "environments", "notes"], "url": "https://arxiv.org/pdf/2601.19306v1"}], "id": "arxiv:2601.19306", "identifiers": {"arxiv": "2601.19306"}, "kind": "paper", "last_verified_at": "2026-10-18T10:01:04.629294+00:00", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2601.19306"}], "name": "Curiosity Driven Knowledge Retrieval for Mobile Agents", "notes": "The curiosity signal is a model-based proxy, not a calibrated probability of failure. Accurate external app information is a dependency, and AndroidWorld is the reported evaluation environment. No performance numbers were independently reproduced. Basis: complete retained abstract and source-reviewed sections of the exact-version paper (3.1 Problem Formulation; 3.2 Curiosity-Driven Information Gain Estimation; 3.3 Retrieving and Embedding External Knowledge); not an entire-paper reading or reproduction claim.", "published_at": "2026-01-27", "relations": [], "review_status": "curated", "scope": "core", "summary": "Uses execution uncertainty to trigger retrieval from application documentation, code, and past trajectories, organizing the results into structured AppCards for Android agent planning.", "topics": ["memory", "planning", "mobile-navigation"], "updated_at": "2026-01-27", "url": "https://arxiv.org/abs/2601.19306"}
{"aliases": [], "authors": ["Tianyi Chen", "Yinheng Li", "Michael Solodko", "Sen Wang", "Nan Jiang", "Tingyuan Cui", "Junheng Hao", "Jongwoo Ko", "Sara Abdali", "Leon Xu", "Suzhen Zheng", "Hao Fan", "Pashmina Cameron", "Justin Wagle", "Kazuhito Koishida"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "5a3efa3cd0add437e327ea49896a3ee794daa178664dfb4096aba66fed27554b", "submitted_at": "2026-01-28T23:38:25Z", "updated_at": "2026-02-02T23:11:55Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2601.21123v2"}}, "environments": ["desktop", "windows"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2601.21123", "identifiers": {"arxiv": "2601.21123"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2601.21123"}], "name": "CUA-Skill: Develop Skills for Computer Using Agent", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-01-28", "relations": [], "review_status": "curated", "scope": "core", "summary": "CUA-Skill encodes Windows application knowledge in parameterized skills and execution/composition graphs. Its agent retrieves and instantiates these skills and uses memory for failure recovery, with end-to-end evaluation on WindowsAgentArena.", "topics": ["tool-integration", "planning", "memory", "automation"], "updated_at": "2026-02-02", "url": "https://arxiv.org/abs/2601.21123"}
{"aliases": [], "authors": ["Hongze Mi", "Yibo Feng", "WenJie Lu", "Song Cao", "Jinyuan Li", "Yanming Li", "Xuelin Zhang", "Haotian Luo", "Songyang Peng", "He Cui", "Tengfei Tian", "Jun Fang", "Hua Chai", "Naiqiang Tan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "c0b3dbaacdf640a1314a40688133f7386c0eb8b901c43393ad21ec887814a606", "submitted_at": "2026-01-30T04:01:21Z", "updated_at": "2026-01-30T04:01:21Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2601.22528v1"}}, "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:2601.22528", "identifiers": {"arxiv": "2601.22528"}, "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/2601.22528"}], "name": "Darwinian Memory: A Training-Free Self-Regulating Memory System for GUI Agent Evolution", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-01-30", "relations": [], "review_status": "curated", "scope": "core", "summary": "Darwinian Memory splits GUI trajectories into reusable units and updates the memory by retaining useful paths and pruning poor ones. It evaluates this training-free memory mechanism on long multi-application tasks.", "topics": ["memory", "training-free", "cross-app", "long-horizon"], "updated_at": "2026-01-30", "url": "https://arxiv.org/abs/2601.22528"}
{"aliases": [], "authors": ["Aiden Yiliu Li", "Xinyue Hao", "Shilong Liu", "Mengdi Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a5b730fa81936d6a1b2b0a8b74a522a22de39b977929165c44a5050e490c301f", "submitted_at": "2026-02-02T18:50:07Z", "updated_at": "2026-02-02T18:50:07Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2602.02468v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2602.02468", "identifiers": {"arxiv": "2602.02468"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2602.02468"}], "name": "Avenir-Web: Human-Experience-Imitating Multimodal Web Agents with Mixture of Grounding Experts", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-02-02", "relations": [], "review_status": "curated", "scope": "core", "summary": "Avenir-Web combines multiple grounding specialists with procedural experience, a task checklist and adaptive memory. It evaluates this planning and execution framework on live Online-Mind2Web tasks.", "topics": ["grounding", "planning", "memory", "web-navigation"], "updated_at": "2026-02-02", "url": "https://arxiv.org/abs/2602.02468"}
{"aliases": [], "authors": ["Han Xiao", "Guozhi Wang", "Hao Wang", "Shilong Liu", "Yuxiang Chai", "Yue Pan", "Yufeng Zhou", "Xiaoxin Chen", "Yafei Wen", "Hongsheng Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "f51eb0b7102a37fcc705c53640c5a1d30fcc115cedcdca4a00493ed45b9d5507", "submitted_at": "2026-02-05T16:21:43Z", "updated_at": "2026-02-05T16:21:43Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2602.05832v1"}}, "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:2602.05832", "identifiers": {"arxiv": "2602.05832"}, "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/2602.05832"}], "name": "UI-Mem: Self-Evolving Experience Memory for Online Reinforcement Learning in Mobile GUI Agents", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-02-05", "relations": [], "review_status": "curated", "scope": "core", "summary": "UI-Mem stores workflows, skills and failure patterns as reusable templates to guide online GUI reinforcement learning. It varies guidance within rollout groups and continually updates the memory so an unguided policy can learn from guided experiences.", "topics": ["memory", "reinforcement-learning", "cross-app"], "updated_at": "2026-02-05", "url": "https://arxiv.org/abs/2602.05832"}
{"aliases": [], "authors": ["Guangyi Liu", "Pengxiang Zhao", "Yaozhen Liang", "Qinyi Luo", "Shunye Tang", "Yuxiang Chai", "Weifeng Lin", "Han Xiao", "WenHao Wang", "Siheng Chen", "Zhengxi Lu", "Gao Wu", "Hao Wang", "Liang Liu", "Yong Liu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "0692d6936a3e3c20a7ad61c69ae770f5669c3783ff837092254f8b3fb170cd0a", "submitted_at": "2026-02-03T17:01:59Z", "updated_at": "2026-08-27T06:48:18Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2602.06075v3"}}, "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:2602.06075", "identifiers": {"arxiv": "2602.06075"}, "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/2602.06075"}], "name": "MemGUI-Bench: Benchmarking Memory of Mobile GUI Agents in Dynamic Environments", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-02-03", "relations": [], "review_status": "curated", "scope": "core", "summary": "MemGUI-Bench evaluates short-term retention and cross-session experience reuse in mobile GUI agents using paired tasks and repeated-attempt protocols. Its staged evaluator measures memory fidelity, learning and execution efficiency.", "topics": ["memory", "evaluation", "mobile-navigation", "cross-app"], "updated_at": "2026-08-27", "url": "https://arxiv.org/abs/2602.06075"}
{"aliases": [], "authors": ["Deyang Jiang", "Jing Huang", "Xuanle Zhao", "Lei Chen", "Liming Zheng", "Fanfan Liu", "Haibo Qiu", "Peng Shi", "Zhixiong Zeng"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "b5080050b85e38c8c03078c96c206e92d4ed63d106d55983563d51fcbc9f66c5", "submitted_at": "2026-02-10T11:16:57Z", "updated_at": "2026-02-10T11:16:57Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2602.09662v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:56Z", "role": "primary", "sha256": "c22be2d5c6a84b2c367759ce2b8f41f5a178fccd0534e832802c9d57b40918e8", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2601.16206%2C2601.17722%2C2601.18113%2C2601.18842%2C2601.20162%2C2601.20335%2C2601.20380%2C2601.20650%2C2601.21123%2C2601.22781%2C2602.00575%2C2602.01576%2C2602.01725%2C2602.02419%2C2602.02548%2C2602.02995%2C2602.03255%2C2602.03792%2C2602.05429%2C2602.05832%2C2602.06075%2C2602.06391%2C2602.07153%2C2602.07624%2C2602.08235%2C2602.08995%2C2602.09082%2C2602.09222%2C2602.09310%2C2602.09662%2C2602.10139%2C2602.11514%2C2602.11724%2C2602.11750%2C2602.14093%2C2602.14276%2C2602.14296%2C2602.14721%2C2602.16855%2C2602.17049%2C2602.17365%2C2602.20502%2C2602.21858%2C2602.22190%2C2603.05044%2C2603.05295%2C2603.07978%2C2603.08533%2C2603.10178%2C2603.12138&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:56Z", "role": "primary", "sha256": "c22be2d5c6a84b2c367759ce2b8f41f5a178fccd0534e832802c9d57b40918e8", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2601.16206%2C2601.17722%2C2601.18113%2C2601.18842%2C2601.20162%2C2601.20335%2C2601.20380%2C2601.20650%2C2601.21123%2C2601.22781%2C2602.00575%2C2602.01576%2C2602.01725%2C2602.02419%2C2602.02548%2C2602.02995%2C2602.03255%2C2602.03792%2C2602.05429%2C2602.05832%2C2602.06075%2C2602.06391%2C2602.07153%2C2602.07624%2C2602.08235%2C2602.08995%2C2602.09082%2C2602.09222%2C2602.09310%2C2602.09662%2C2602.10139%2C2602.11514%2C2602.11724%2C2602.11750%2C2602.14093%2C2602.14276%2C2602.14296%2C2602.14721%2C2602.16855%2C2602.17049%2C2602.17365%2C2602.20502%2C2602.21858%2C2602.22190%2C2603.05044%2C2603.05295%2C2603.07978%2C2603.08533%2C2603.10178%2C2603.12138&max_results=50"}], "id": "arxiv:2602.09662", "identifiers": {"arxiv": "2602.09662"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:56Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2602.09662"}], "name": "TreeCUA: Efficiently Scaling GUI Automation with Tree-Structured Verifiable Evolution", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and released artifacts have not been independently reproduced.", "published_at": "2026-02-10", "relations": [], "review_status": "curated", "scope": "core", "summary": "TreeCUA organizes GUI exploration trajectories as reusable trees and balances deeper tasks against broader coverage. Its data pipeline combines action verification, memory-backed recovery and a DPO variant that uses information from adjacent branches.", "topics": ["data-generation", "planning", "memory", "efficiency"], "updated_at": "2026-02-10", "url": "https://arxiv.org/abs/2602.09662"}
{"aliases": [], "authors": ["Zexue He", "Yu Wang", "Churan Zhi", "Yuanzhe Hu", "Tzu-Ping Chen", "Lang Yin", "Ze Chen", "Tong Arthur Wu", "Siru Ouyang", "Zihan Wang", "Jiaxin Pei", "Julian McAuley", "Yejin Choi", "Alex Pentland"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2317a94d54b4458a1b7c5e92754aeaefcbf270aabfa0bd8ea333c673cd4596b3", "submitted_at": "2026-02-18T09:49:14Z", "updated_at": "2026-09-17T22:55:03Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2602.16313v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T12:31:12Z", "role": "primary", "sha256": "c8c95278aa5dcc6a8f2eda3f3b44d61ec1f35adcb6a4c6c9ab328691a1a4f102", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2601.13462%2C2601.14232%2C2601.15331%2C2601.17548%2C2601.18418%2C2601.18491%2C2601.18827%2C2601.20171%2C2601.20613%2C2601.20742%2C2601.21233%2C2601.21253%2C2601.22037%2C2601.22154%2C2601.22511%2C2601.22607%2C2602.01550%2C2602.03117%2C2602.03361%2C2602.05665%2C2602.05746%2C2602.05877%2C2602.06052%2C2602.07398%2C2602.07458%2C2602.07918%2C2602.09007%2C2602.09463%2C2602.09514%2C2602.10122%2C2602.10525%2C2602.11224%2C2602.12268%2C2602.12430%2C2602.13318%2C2602.13543%2C2602.13692%2C2602.14038%2C2602.14337%2C2602.14364%2C2602.16313%2C2602.16317%2C2602.16666%2C2602.16901%2C2602.16930%2C2602.17547%2C2602.17646%2C2602.18846%2C2602.19320%2C2602.19594&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:31:12Z", "role": "primary", "sha256": "c8c95278aa5dcc6a8f2eda3f3b44d61ec1f35adcb6a4c6c9ab328691a1a4f102", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2601.13462%2C2601.14232%2C2601.15331%2C2601.17548%2C2601.18418%2C2601.18491%2C2601.18827%2C2601.20171%2C2601.20613%2C2601.20742%2C2601.21233%2C2601.21253%2C2601.22037%2C2601.22154%2C2601.22511%2C2601.22607%2C2602.01550%2C2602.03117%2C2602.03361%2C2602.05665%2C2602.05746%2C2602.05877%2C2602.06052%2C2602.07398%2C2602.07458%2C2602.07918%2C2602.09007%2C2602.09463%2C2602.09514%2C2602.10122%2C2602.10525%2C2602.11224%2C2602.12268%2C2602.12430%2C2602.13318%2C2602.13543%2C2602.13692%2C2602.14038%2C2602.14337%2C2602.14364%2C2602.16313%2C2602.16317%2C2602.16666%2C2602.16901%2C2602.16930%2C2602.17547%2C2602.17646%2C2602.18846%2C2602.19320%2C2602.19594&max_results=50"}], "id": "arxiv:2602.16313", "identifiers": {"arxiv": "2602.16313"}, "kind": "paper", "last_verified_at": "2026-10-17T12:31:12Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2602.16313"}], "name": "MemoryArena: Benchmarking Agent Memory in Interdependent Multi-Session Agentic Tasks", "notes": "Reviewed against the complete retained official abstract. Experimental findings and release availability have not been independently reproduced or verified.", "published_at": "2026-02-18", "relations": [], "review_status": "curated", "scope": "core", "summary": "MemoryArena links tasks across sessions so that agents must distill earlier interaction experience and use it in later decisions. Its evaluation includes web navigation alongside planning, search and formal reasoning, testing memory through subsequent actions rather than recall alone.", "topics": ["memory", "interaction-history", "planning", "evaluation", "long-horizon"], "updated_at": "2026-09-17", "url": "https://arxiv.org/abs/2602.16313"}
{"aliases": [], "authors": ["Haiyang Xu", "Xi Zhang", "Haowei Liu", "Junyang Wang", "Zhaozai Zhu", "Shengjie Zhou", "Xuhao Hu", "Feiyu Gao", "Junjie Cao", "Zihua Wang", "Zhiyuan Chen", "Jitong Liao", "Qi Zheng", "Jiahui Zeng", "Ze Xu", "Shuai Bai", "Junyang Lin", "Jingren Zhou", "Ming Yan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "eeaddd8a32bdd4b53356ca43e849ea3697ac3d40a05b32e6c930dc445c243898", "submitted_at": "2026-02-15T01:52:19Z", "updated_at": "2026-02-15T01:52:19Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2602.16855v1"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2602.16855", "identifiers": {"arxiv": "2602.16855"}, "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/2602.16855"}], "name": "Mobile-Agent-v3.5: Multi-platform Fundamental GUI Agents", "notes": "Summary and topic tags reviewed against the primary abstract. Reported experiments, task counts, and artifact availability are not independently reproduced.", "published_at": "2026-02-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "Presents GUI-Owl-1.5 variants and a hybrid data pipeline for multimodal GUI interaction, tool use, memory, and multi-platform tasks.", "topics": ["tool-integration", "memory", "reinforcement-learning"], "updated_at": "2026-02-15", "url": "https://arxiv.org/abs/2602.16855"}
{"aliases": [], "authors": ["Serin Kim", "Sangam Lee", "Dongha Lee"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "28a23bf22920f8c33317e9d16f59d7085f188e79157301ff573abe773cf562ce", "submitted_at": "2026-02-19T01:54:26Z", "updated_at": "2026-05-27T12:15:49Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2602.17003v3"}}, "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:2602.17003", "identifiers": {"arxiv": "2602.17003"}, "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/2602.17003"}], "name": "Persona2Web: Benchmarking Personalized Web Agents for Contextual Reasoning with User History", "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": "Persona2Web tests whether web agents can resolve ambiguous requests using preferences implicit in a user’s history. Its reasoning-aware evaluation compares agent architectures and history-access strategies on personalized tasks on the open web.", "topics": ["evaluation", "memory", "web-navigation"], "updated_at": "2026-05-27", "url": "https://arxiv.org/abs/2602.17003"}
{"aliases": [], "authors": ["Hongbin Zhong", "Fazle Faisal", "Luis França", "Tanakorn Leesatapornwongsa", "Adriana Szekeres", "Kexin Rong", "Suman Nath"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e179c833d43a88098d576e31e726eb3e4289f728659697364d0b448282df3fc1", "submitted_at": "2026-02-24T03:03:18Z", "updated_at": "2026-09-28T17:40:20Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2602.20502v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2602.20502", "identifiers": {"arxiv": "2602.20502"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:43Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2602.20502"}], "name": "ActionEngine: From Reactive to Programmatic Web Agents via State Machine Memory", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-02-24", "relations": [], "review_status": "curated", "scope": "core", "summary": "ActionEngine explores web applications to build an updateable state-machine memory of GUI states, operations and transitions. It generates complete task programs from that memory, using reactive repair to handle changed interfaces or missing knowledge.", "topics": ["memory", "planning", "efficiency", "web-navigation"], "updated_at": "2026-09-28", "url": "https://arxiv.org/abs/2602.20502"}
{"aliases": [], "authors": ["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": ["Yiping Xie", "Song Chen", "Jingxuan Xing", "Wei Jiang", "Zekun Zhu", "Yingyao Wang", "Pi Bu", "Jun Song", "Yuning Jiang", "Bo Zheng"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "db31c6ca303f0c7dc43f24dd1eea69621b1e557af33122386945154e8236ab78", "submitted_at": "2026-03-09T16:04:08Z", "updated_at": "2026-03-31T03:08:51Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2603.08533v2"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2603.08533", "identifiers": {"arxiv": "2603.08533"}, "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/2603.08533"}], "name": "SecAgent: Efficient Mobile GUI Agent with Semantic Context", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced. Mobile use is explicit, but the abstract does not identify a mobile operating system.", "published_at": "2026-03-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "Presents a Chinese mobile GUI dataset and navigation benchmark together with SecAgent, which compresses previous screenshots and actions into semantic summaries. The agent is trained with supervised and reinforcement fine-tuning to use this compact interaction context.", "topics": ["mobile-navigation", "memory", "reinforcement-learning"], "updated_at": "2026-03-31", "url": "https://arxiv.org/abs/2603.08533"}
{"aliases": [], "authors": ["Xunzhuo Liu", "Bowei He", "Xue Liu", "Andy Luo", "Haichen Zhang", "Huamin Chen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "274deff6cacfb5f1b7b620a86abe7ef6496b673d929de39376a9ad42dd3c4c02", "submitted_at": "2026-03-13T09:21:25Z", "updated_at": "2026-03-13T09:21:25Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2603.12823v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:43Z", "role": "primary", "sha256": "c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=400&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2603.12823", "identifiers": {"arxiv": "2603.12823"}, "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/2603.12823"}], "name": "Adaptive Vision-Language Model Routing for Computer Use Agents", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-03-13", "relations": [], "review_status": "curated", "scope": "core", "summary": "Adaptive VLM Routing estimates GUI-action difficulty and small-model confidence to select a model under a predicted reliability threshold. Grounding and agent-routing benchmarks assess the cost–accuracy trade-off, with prior interaction context used for warm agents.", "topics": ["efficiency", "grounding", "memory", "evaluation"], "updated_at": "2026-03-13", "url": "https://arxiv.org/abs/2603.12823"}
{"aliases": [], "authors": ["Yibo Shi", "Jungang Li", "Linghao Zhang", "Zihao Dongfang", "Biao Wu", "Sicheng Tao", "Yibo Yan", "Chenxi Qin", "Weiting Liu", "Zhixin Lin", "Hanqian Li", "Yu Huang", "Song Dai", "Yonghua Hei", "Yue Ding", "Xiang Li", "Shikang Wang", "Chengdong Xu", "Jingqi Liu", "Xueying Ma", "Zhiwen Zheng", "Xiaofei Zhang", "Bincheng Wang", "Nichen Yang", "Jie Wu", "Lihua Tian", "Chen Li", "Xuming Hu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "0ce6da91e59ac57fa24ff991ca023b84448c54d477cacb2d6fded9bf527dd29d", "submitted_at": "2026-03-19T02:45:21Z", "updated_at": "2026-03-19T02:45:21Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2603.18429v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2603.18429", "identifiers": {"arxiv": "2603.18429"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:39Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2603.18429"}], "name": "AndroTMem: From Interaction Trajectories to Anchored Memory in Long-Horizon GUI Agents", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2026-03-19", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces AndroTMem-Bench for Android tasks whose later actions depend on earlier UI states. Anchored State Memory retains causally linked intermediate states for subgoal-specific retrieval and evaluates them against full-history and summary-based memory.", "topics": ["memory", "long-horizon", "evaluation"], "updated_at": "2026-03-19", "url": "https://arxiv.org/abs/2603.18429"}
{"aliases": [], "authors": ["Akash Ghosh", "Tajamul Ashraf", "Rishu Kumar Singh", "Numan Saeed", "Sriparna Saha", "Xiuying Chen", "Salman Khan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "8fb6a294616e1be46c5da05f60cda7d43c05d8a1ce71116c5e30cb94507cd56f", "submitted_at": "2026-03-25T10:25:48Z", "updated_at": "2026-03-25T10:25:48Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2603.24157v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T12:31:16Z", "role": "primary", "sha256": "9289bc047b59e93f75703b72f5e92d9277e115238427519c343df550220e351b", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2602.19870%2C2602.22769%2C2602.23166%2C2602.23320%2C2603.00198%2C2603.00842%2C2603.00991%2C2603.01050%2C2603.01145%2C2603.02236%2C2603.02795%2C2603.03278%2C2603.04549%2C2603.06713%2C2603.07106%2C2603.07670%2C2603.07853%2C2603.08706%2C2603.09716%2C2603.10002%2C2603.10664%2C2603.11139%2C2603.11320%2C2603.12483%2C2603.13026%2C2603.13247%2C2603.15714%2C2603.16065%2C2603.16289%2C2603.16839%2C2603.17216%2C2603.17378%2C2603.19313%2C2603.19685%2C2603.21489%2C2603.22527%2C2603.23638%2C2603.24157%2C2603.24639%2C2603.27646%2C2603.28069%2C2603.28407%2C2603.29418%2C2603.29847%2C2604.00528%2C2604.00886%2C2604.01194%2C2604.01438%2C2604.01560%2C2604.02985&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:31:16Z", "role": "primary", "sha256": "9289bc047b59e93f75703b72f5e92d9277e115238427519c343df550220e351b", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2602.19870%2C2602.22769%2C2602.23166%2C2602.23320%2C2603.00198%2C2603.00842%2C2603.00991%2C2603.01050%2C2603.01145%2C2603.02236%2C2603.02795%2C2603.03278%2C2603.04549%2C2603.06713%2C2603.07106%2C2603.07670%2C2603.07853%2C2603.08706%2C2603.09716%2C2603.10002%2C2603.10664%2C2603.11139%2C2603.11320%2C2603.12483%2C2603.13026%2C2603.13247%2C2603.15714%2C2603.16065%2C2603.16289%2C2603.16839%2C2603.17216%2C2603.17378%2C2603.19313%2C2603.19685%2C2603.21489%2C2603.22527%2C2603.23638%2C2603.24157%2C2603.24639%2C2603.27646%2C2603.28069%2C2603.28407%2C2603.29418%2C2603.29847%2C2604.00528%2C2604.00886%2C2604.01194%2C2604.01438%2C2604.01560%2C2604.02985&max_results=50"}], "id": "arxiv:2603.24157", "identifiers": {"arxiv": "2603.24157"}, "kind": "paper", "last_verified_at": "2026-10-17T12:31:16Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2603.24157"}], "name": "CarePilot: A Multi-Agent Framework for Long-Horizon Computer Task Automation in Healthcare", "notes": "Reviewed against the complete retained official abstract. Experimental findings and release availability have not been independently reproduced or verified.", "published_at": "2026-03-25", "relations": [], "review_status": "curated", "scope": "core", "summary": "CareFlow benchmarks long healthcare-software workflows, while CarePilot coordinates action proposals, execution checks and corrective feedback using short- and long-term memory. Its actor consumes visual interfaces and system state to select semantic actions.", "topics": ["automation", "planning", "memory", "evaluation", "long-horizon", "workflow"], "updated_at": "2026-03-25", "url": "https://arxiv.org/abs/2603.24157"}
{"aliases": [], "authors": ["Wei Zou", "Mingwen Dong", "Miguel Romero Calvo", "Shuaichen Chang", "Jiang Guo", "Dongkyu Lee", "Xing Niu", "Xiaofei Ma", "Yanjun Qi", "Jiarong Jiang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a37f4bdc495e10e4ae301ebff6faa28cdfa6de152d4619c051083b5dce41e6ba", "submitted_at": "2026-04-03T01:25:12Z", "updated_at": "2026-04-07T14:45:15Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2604.02623v2"}}, "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:2604.02623", "identifiers": {"arxiv": "2604.02623"}, "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/2604.02623"}], "name": "Poison Once, Exploit Forever: Environment-Injected Memory Poisoning Attacks on Web Agents", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2026-04-03", "relations": [], "review_status": "curated", "scope": "core", "summary": "Studies persistent web-agent memory contamination introduced through environmental observations instead of direct memory access. Experiments on WebArena and VisualWebArena examine cross-session effects and how interaction failures alter susceptibility.", "topics": ["safety-security", "memory", "web-navigation"], "updated_at": "2026-04-07", "url": "https://arxiv.org/abs/2604.02623"}
{"aliases": [], "authors": ["Tiantian He", "Yihang Chen", "Keyue Jiang", "Ka Yiu Lee", "Kaiwen Zhou", "Kun Shao", "Shuai Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "dd2bc0ee1955dbc8e98a9306a3d6260251cbab89d36ac117e4dc49b6fa99766e", "submitted_at": "2026-04-10T18:46:05Z", "updated_at": "2026-04-10T18:46:05Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2604.09815v1"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2604.09815", "identifiers": {"arxiv": "2604.09815"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:39Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2604.09815"}], "name": "EE-MCP: Self-Evolving MCP-GUI Agents via Automated Environment Generation and Experience Learning", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-04-10", "relations": [], "review_status": "curated", "scope": "core", "summary": "EE-MCP studies application-dependent mixtures of GUI actions and MCP calls using generated environments, trajectory collection and filtered training. An experience bank extracts rules from trajectory comparisons for inference-time improvement, distinct from model distillation.", "topics": ["tool-integration", "data-generation", "imitation-learning", "memory"], "updated_at": "2026-04-10", "url": "https://arxiv.org/abs/2604.09815"}
{"aliases": [], "authors": ["Zhaoyang Wang", "Qianhui Wu", "Xuchao Zhang", "Chaoyun Zhang", "Wenlin Yao", "Fazle Elahi Faisal", "Baolin Peng", "Si Qin", "Suman Nath", "Qingwei Lin", "Chetan Bansal", "Dongmei Zhang", "Saravan Rajmohan", "Jianfeng Gao", "Huaxiu Yao"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "bf709a3d7096bc07218206edb1accfdda64036a39ea770e637b3e073dd9a76c8", "submitted_at": "2026-04-14T21:48:15Z", "updated_at": "2026-08-31T07:02:14Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2604.13318v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2604.13318", "identifiers": {"arxiv": "2604.13318"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:39Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2604.13318"}], "name": "WebXSkill: Skill Learning for Autonomous Web Agents", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-04-14", "relations": [], "review_status": "curated", "scope": "core", "summary": "WebXSkill pairs parameterized action programs with step-level language guidance, mining reusable skills from synthetic browser-agent traces. A URL-based graph supports retrieval, and deployment alternates between direct program execution and guided native planning.", "topics": ["memory", "planning", "web-navigation", "automation"], "updated_at": "2026-08-31", "url": "https://arxiv.org/abs/2604.13318"}
{"aliases": [], "authors": ["Zhengxi Lu", "Fei Tang", "Guangyi Liu", "Kaitao Song", "Xu Tan", "Jin Ma", "Wenqi Zhang", "Weiming Lu", "Jun Xiao", "Yueting Zhuang", "Yongliang Shen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d6e81b448c9c22e2fd25dce519db3fdc0585b0265b94d448751d7a37ed910dfa", "submitted_at": "2026-04-15T12:55:46Z", "updated_at": "2026-04-15T12:55:46Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2604.13822v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:32:00Z", "role": "primary", "sha256": "4d09ebfb093e43a4c44adef86c8fa8374f90d907fc33bd8e12a1315e62d5a583", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2603.12823%2C2603.14707%2C2603.15039%2C2603.17357%2C2603.18429%2C2603.19191%2C2603.24440%2C2603.25226%2C2603.25864%2C2603.26211%2C2603.26648%2C2603.29318%2C2604.01535%2C2604.02947%2C2604.04399%2C2604.05477%2C2604.06126%2C2604.06182%2C2604.06367%2C2604.07429%2C2604.08455%2C2604.08516%2C2604.08523%2C2604.09155%2C2604.09574%2C2604.09587%2C2604.09815%2C2604.09937%2C2604.11259%2C2604.11784%2C2604.12284%2C2604.13019%2C2604.13822%2C2604.14113%2C2604.14262%2C2604.14956%2C2604.15093%2C2604.15309%2C2604.17817%2C2604.17821%2C2604.18860%2C2604.24964%2C2604.25380%2C2604.25562%2C2604.27151%2C2604.27253%2C2604.27776%2C2605.00642%2C2605.06393%2C2605.06664&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:32:00Z", "role": "primary", "sha256": "4d09ebfb093e43a4c44adef86c8fa8374f90d907fc33bd8e12a1315e62d5a583", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?id_list=2603.12823%2C2603.14707%2C2603.15039%2C2603.17357%2C2603.18429%2C2603.19191%2C2603.24440%2C2603.25226%2C2603.25864%2C2603.26211%2C2603.26648%2C2603.29318%2C2604.01535%2C2604.02947%2C2604.04399%2C2604.05477%2C2604.06126%2C2604.06182%2C2604.06367%2C2604.07429%2C2604.08455%2C2604.08516%2C2604.08523%2C2604.09155%2C2604.09574%2C2604.09587%2C2604.09815%2C2604.09937%2C2604.11259%2C2604.11784%2C2604.12284%2C2604.13019%2C2604.13822%2C2604.14113%2C2604.14262%2C2604.14956%2C2604.15093%2C2604.15309%2C2604.17817%2C2604.17821%2C2604.18860%2C2604.24964%2C2604.25380%2C2604.25562%2C2604.27151%2C2604.27253%2C2604.27776%2C2605.00642%2C2605.06393%2C2605.06664&max_results=50"}], "id": "arxiv:2604.13822", "identifiers": {"arxiv": "2604.13822"}, "kind": "paper", "last_verified_at": "2026-10-16T06:32:00Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2604.13822"}], "name": "UI-Copilot: Advancing Long-Horizon GUI Automation via Tool-Integrated Policy Optimization", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-04-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "UI-Copilot separates GUI execution from on-demand memory retrieval and numerical assistance. Its training procedure treats tool selection and multi-turn task execution as distinct optimization targets while keeping persistent observations outside transient context.", "topics": ["tool-integration", "memory", "planning", "long-horizon", "mobile-navigation"], "updated_at": "2026-04-15", "url": "https://arxiv.org/abs/2604.13822"}
{"aliases": [], "authors": ["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": ["Zehao Lin", "Xixuan Hao", "Renyu Fu", "Shaobo Cui", "Kai Chen", "Chunyu Li", "Zhiyu Li", "Feiyu Xiong"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "fb75ea30086ecf069d8adda1513a6e906cfcfcedcd4cab01d0986a850aa8009a", "submitted_at": "2026-04-17T06:28:22Z", "updated_at": "2026-09-22T07:33:24Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2604.16548v3"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-18T10:54:12Z", "role": "primary", "sha256": "5f5bae19420f79231516d4bcf604ea18d4f930fc7790cde5a3e7fcb8d41a1126", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2508.19461%2C2508.19493%2C2509.21782%2C2509.25140%2C2510.03485%2C2510.07809%2C2510.14276%2C2510.27140%2C2510.27210%2C2511.13248%2C2512.08417%2C2601.07263%2C2602.13516%2C2603.00476%2C2603.11862%2C2603.18762%2C2603.23791%2C2604.00986%2C2604.16515%2C2604.16548%2C2604.24348%2C2605.05509%2C2605.08876%2C2605.20833%2C2605.24785%2C2607.00333%2C2607.18659%2C2607.25255%2C2608.00202%2C2608.08939%2C2608.24913%2C2608.29596&max_results=32"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:54:12Z", "role": "primary", "sha256": "5f5bae19420f79231516d4bcf604ea18d4f930fc7790cde5a3e7fcb8d41a1126", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2508.19461%2C2508.19493%2C2509.21782%2C2509.25140%2C2510.03485%2C2510.07809%2C2510.14276%2C2510.27140%2C2510.27210%2C2511.13248%2C2512.08417%2C2601.07263%2C2602.13516%2C2603.00476%2C2603.11862%2C2603.18762%2C2603.23791%2C2604.00986%2C2604.16515%2C2604.16548%2C2604.24348%2C2605.05509%2C2605.08876%2C2605.20833%2C2605.24785%2C2607.00333%2C2607.18659%2C2607.25255%2C2608.00202%2C2608.08939%2C2608.24913%2C2608.29596&max_results=32"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T13:00:08.054642+00:00", "role": "primary", "sha256": "f795535d55a4def0f2ce7dfc41a9d1d22ad38eddf48fb4f1feed98b6b5b66721", "supports": ["summary", "scope", "notes"], "url": "https://arxiv.org/html/2604.16548v3"}], "id": "arxiv:2604.16548", "identifiers": {"arxiv": "2604.16548"}, "kind": "survey", "last_verified_at": "2026-10-18T10:54:12Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2604.16548"}], "name": "A Survey on Long-Term Memory Security in LLM Agents: Attacks, Defenses, and Governance Across the Memory Lifecycle", "notes": "Adjacent memory-security survey. Full-text Sections 3.1 and 3.3 discuss poisoned web observations persisted into later agent actions; the work is broader than GUI agents and does not establish a platform implementation.", "published_at": "2026-04-17", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "Organizes security risks and defenses for persistent agent memory into six lifecycle phases and proposes five verifiable governance primitives.", "topics": ["memory", "safety-security", "survey"], "updated_at": "2026-09-22", "url": "https://arxiv.org/abs/2604.16548"}
{"aliases": [], "authors": ["Jackson Vonderhorst", "Kuangshi Ai", "Haichao Miao", "Shusen Liu", "Chaoli Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "8e7b07a811df7e74a28f91a5b6c59c69b7ba066bd6c64452a299e678a3187733", "submitted_at": "2026-04-30T15:22:28Z", "updated_at": "2026-06-26T20:31:32Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2604.27996v3"}}, "environments": ["unspecified"], "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:2604.27996", "identifiers": {"arxiv": "2604.27996"}, "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/2604.27996"}], "name": "Exploring LLM Agent Designs and Interaction Modalities for Scientific Visualization", "notes": "Reviewed against the complete retained official abstract. Experimental findings and release availability have not been independently reproduced or verified.", "published_at": "2026-04-30", "relations": [], "review_status": "curated", "scope": "core", "summary": "This study compares specialized, computer-use and coding agents on scientific visualization workflows. It examines GUI, CLI, code and tool-call interaction modes, including how persistent memory changes performance across repeated trials.", "topics": ["science", "evaluation", "tool-integration", "memory"], "updated_at": "2026-06-26", "url": "https://arxiv.org/abs/2604.27996"}
{"aliases": [], "authors": ["Xiaoming Ren", "Ru Zhen", "Chao Li", "Yang Song", "Qiuxia Hou", "Yanhao Zhang", "Peng Liu", "Qi Qi", "Quanlong Zheng", "Qi Wu", "Zhenyi Liao", "Binqiang Pan", "Haobo Ji", "Haonan Lu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2ece93c00c5556dfe50ddd0ba3f52b725f34d1bee18ea3565951bb78343d8ed0", "submitted_at": "2026-05-07T06:58:34Z", "updated_at": "2026-05-21T06:55:40Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2605.05765v2"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-18T10:00:51.952206+00:00", "role": "primary", "sha256": "5cd7fb95d5ee8e8c473e0c9d5b892bf50ee425604c6a320e8e9a12eff17120a1", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+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=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:00:51.952206+00:00", "role": "primary", "sha256": "5cd7fb95d5ee8e8c473e0c9d5b892bf50ee425604c6a320e8e9a12eff17120a1", "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=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:14:10.101459+00:00", "role": "primary", "sha256": "80375cd810dcda2e9ccca2c6fbee44730de8449972036c548bde312566b20673", "supports": ["summary", "scope", "environments", "notes"], "url": "https://arxiv.org/pdf/2605.05765v2"}], "id": "arxiv:2605.05765", "identifiers": {"arxiv": "2605.05765"}, "kind": "paper", "last_verified_at": "2026-10-18T10:00:51.952206+00:00", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2605.05765"}], "name": "X-OmniClaw Technical Report: A Unified Mobile Agent for Multimodal Understanding and Interaction", "notes": "The report presents demonstrations. Its behavior-cloning component captures runtime skills; it does not establish policy-weight imitation training, and literal trace replay is described as future work. No general benchmark result or release completeness is inferred. Basis: complete retained abstract and source-reviewed sections of the exact-version paper (2 Framework; 3 Omni Perception; 4 Omni Memory; 5.1 Omni Action in App Ecosystem; 5.2 Trajectory Cloned Execution); not an entire-paper reading or reproduction claim.", "published_at": "2026-05-07", "relations": [], "review_status": "curated", "scope": "core", "summary": "Describes an Android agent architecture combining multimodal perception, working and personal memory, hybrid XML/visual grounding, and reusable skills captured from user interactions.", "topics": ["mobile-navigation", "memory", "grounding", "workflow"], "updated_at": "2026-05-21", "url": "https://arxiv.org/abs/2605.05765"}
{"aliases": [], "authors": ["Guibin Zhang", "Yaohui Ling", "Fanci Meng", "Kun Wang", "Shuicheng Yan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "729ff6b37da62bb8a9a911865554520e3b29219d9735e087c3186705428d7b05", "submitted_at": "2026-05-10T04:31:23Z", "updated_at": "2026-05-10T04:31:23Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2605.09317v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2605.09317", "identifiers": {"arxiv": "2605.09317"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:39Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2605.09317"}], "name": "Mem-W: Latent Memory-Native GUI Agents", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2026-05-10", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces Mem-W, which compresses previous trajectories and current-session segments into latent memory tokens used directly by the GUI policy. Self-distillation and outcome supervision train this memory representation for web and mobile navigation.", "topics": ["memory", "long-horizon", "imitation-learning"], "updated_at": "2026-05-10", "url": "https://arxiv.org/abs/2605.09317"}
{"aliases": [], "authors": ["Zerui Qin", "Sheng Yue", "Xingyuan Hua", "Yongjian Fu", "Ju Ren"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "388eac2721929fc8211f40cc4edde61be0b6e6145fa60fbaa444ec80822fcc36", "submitted_at": "2026-05-12T15:48:44Z", "updated_at": "2026-05-12T15:48:44Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2605.12294v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:39Z", "role": "primary", "sha256": "27c33576ed43c7bc63d4b290f6558553bd1d3aa112a7358138f4045e1bc14918", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=300&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2605.12294", "identifiers": {"arxiv": "2605.12294"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:39Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2605.12294"}], "name": "Executable Agentic Memory for GUI Agent", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-05-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "Executable Agentic Memory stores GUI routines in a knowledge graph built through state-aware exploration and action-group mining. A learned value model guides tree search over this executable memory for retrieval-based planning, with AndroidWorld and efficiency evaluation.", "topics": ["memory", "planning", "long-horizon", "efficiency"], "updated_at": "2026-05-12", "url": "https://arxiv.org/abs/2605.12294"}
{"aliases": [], "authors": ["Di Wu", "Zixiang Ji", "Asmi Kawatkar", "Bryan Kwan", "Jia-Chen Gu", "Nanyun Peng", "Kai-Wei Chang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "5893d1d9205776667a1e1d89720ec3baf55fafe80cc9c6201ca97088f44a10ac", "submitted_at": "2026-05-12T17:59:34Z", "updated_at": "2026-05-12T17:59:34Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2605.12493v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T12:31:23Z", "role": "primary", "sha256": "9a71e95803ce58c462c84e2b45a891555e31d1fb409e643bb3ca4675a44c96e6", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2605.12233%2C2605.12493%2C2605.12978%2C2605.13119%2C2605.14133%2C2605.16909%2C2605.17260%2C2605.17324%2C2605.17448%2C2605.17656%2C2605.17721%2C2605.19447%2C2605.19587%2C2605.20894%2C2605.22642%2C2605.23204%2C2605.25233%2C2605.25901%2C2605.26086%2C2605.26144%2C2605.26252%2C2605.26646%2C2605.27209%2C2605.27705%2C2605.27898%2C2605.28158%2C2605.28255%2C2605.29341%2C2605.29463%2C2605.29801%2C2605.30010%2C2605.30052%2C2605.30126%2C2605.30434%2C2605.30907%2C2605.30914%2C2605.31264%2C2605.31557%2C2606.00530%2C2606.00543%2C2606.00856%2C2606.01057%2C2606.01961%2C2606.02540%2C2606.03103%2C2606.03889%2C2606.03895%2C2606.04315%2C2606.04455%2C2606.05525&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:31:23Z", "role": "primary", "sha256": "9a71e95803ce58c462c84e2b45a891555e31d1fb409e643bb3ca4675a44c96e6", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2605.12233%2C2605.12493%2C2605.12978%2C2605.13119%2C2605.14133%2C2605.16909%2C2605.17260%2C2605.17324%2C2605.17448%2C2605.17656%2C2605.17721%2C2605.19447%2C2605.19587%2C2605.20894%2C2605.22642%2C2605.23204%2C2605.25233%2C2605.25901%2C2605.26086%2C2605.26144%2C2605.26252%2C2605.26646%2C2605.27209%2C2605.27705%2C2605.27898%2C2605.28158%2C2605.28255%2C2605.29341%2C2605.29463%2C2605.29801%2C2605.30010%2C2605.30052%2C2605.30126%2C2605.30434%2C2605.30907%2C2605.30914%2C2605.31264%2C2605.31557%2C2606.00530%2C2606.00543%2C2606.00856%2C2606.01057%2C2606.01961%2C2606.02540%2C2606.03103%2C2606.03889%2C2606.03895%2C2606.04315%2C2606.04455%2C2606.05525&max_results=50"}], "id": "arxiv:2605.12493", "identifiers": {"arxiv": "2605.12493"}, "kind": "paper", "last_verified_at": "2026-10-17T12:31:23Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2605.12493"}], "name": "LongMemEval-V2: Evaluating Long-Term Agent Memory Toward Experienced Colleagues", "notes": "Reviewed against the complete retained official abstract. Experimental findings and release availability have not been independently reproduced or verified.", "published_at": "2026-05-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "LongMemEval-V2 asks memory systems to recover environment knowledge from long web-agent histories for question answering. Its AgentRunbook variants contrast indexed knowledge pools with coding-agent retrieval from stored trajectory files.", "topics": ["memory", "interaction-history", "state-tracking", "workflow", "evaluation"], "updated_at": "2026-05-12", "url": "https://arxiv.org/abs/2605.12493"}
{"aliases": [], "authors": ["Shilong Jin", "Lanjun Wang", "Zhuosheng Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "52c1bf4c614ac05a55f2c2c66d3ac4eeb1bf31743814a45eb9f0ce064b6baaab", "submitted_at": "2026-05-16T08:51:57Z", "updated_at": "2026-05-16T08:51:57Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2605.16883v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2605.16883", "identifiers": {"arxiv": "2605.16883"}, "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/2605.16883"}], "name": "SE-GA: Memory-Augmented Self-Evolution for GUI Agents", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-05-16", "relations": [], "review_status": "curated", "scope": "core", "summary": "SE-GA retrieves hierarchical memories during inference and uses the collected experience in a separate policy-improvement pipeline. Its evaluation includes offline grounding/action tasks and online Android interaction, which measure different capabilities.", "topics": ["memory", "interaction-history", "mobile-navigation", "grounding", "evaluation"], "updated_at": "2026-05-16", "url": "https://arxiv.org/abs/2605.16883"}
{"aliases": [], "authors": ["Ziyun Zeng", "Hang Hua", "Bocheng Zou", "Mu Cai", "Rogerio Feris", "Jiebo Luo"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d96c8844274f79adea167169a0853e3a7cd82fee4a8a666913f97898a0836e7c", "submitted_at": "2026-05-18T16:57:36Z", "updated_at": "2026-05-18T16:57:36Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2605.18652v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2605.18652", "identifiers": {"arxiv": "2605.18652"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2605.18652"}], "name": "MementoGUI: Learning Agentic Multimodal Memory Control for Long-Horizon GUI Agents", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-05-18", "relations": [], "review_status": "curated", "scope": "core", "summary": "MementoGUI learns a controller for selecting, compressing and retrieving textual and region-level visual memories while leaving the GUI-agent backbone fixed. It also converts computer-use traces into controller training data and evaluates semantic actions, progress and memory consistency on long-horizon tasks.", "topics": ["memory", "interaction-history", "long-horizon", "data-generation", "evaluation"], "updated_at": "2026-05-18", "url": "https://arxiv.org/abs/2605.18652"}
{"aliases": [], "authors": ["Yubo Li", "Yidi Miao", "Yuntian Shen", "Yuxin Liu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e64d36e920a33851aacf93138a2081d0470424dfec6a8fd11915928660824851", "submitted_at": "2026-05-24T00:07:25Z", "updated_at": "2026-05-26T04:25:32Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2605.24785v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-18T10:54:12Z", "role": "primary", "sha256": "5f5bae19420f79231516d4bcf604ea18d4f930fc7790cde5a3e7fcb8d41a1126", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2508.19461%2C2508.19493%2C2509.21782%2C2509.25140%2C2510.03485%2C2510.07809%2C2510.14276%2C2510.27140%2C2510.27210%2C2511.13248%2C2512.08417%2C2601.07263%2C2602.13516%2C2603.00476%2C2603.11862%2C2603.18762%2C2603.23791%2C2604.00986%2C2604.16515%2C2604.16548%2C2604.24348%2C2605.05509%2C2605.08876%2C2605.20833%2C2605.24785%2C2607.00333%2C2607.18659%2C2607.25255%2C2608.00202%2C2608.08939%2C2608.24913%2C2608.29596&max_results=32"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:54:12Z", "role": "primary", "sha256": "5f5bae19420f79231516d4bcf604ea18d4f930fc7790cde5a3e7fcb8d41a1126", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2508.19461%2C2508.19493%2C2509.21782%2C2509.25140%2C2510.03485%2C2510.07809%2C2510.14276%2C2510.27140%2C2510.27210%2C2511.13248%2C2512.08417%2C2601.07263%2C2602.13516%2C2603.00476%2C2603.11862%2C2603.18762%2C2603.23791%2C2604.00986%2C2604.16515%2C2604.16548%2C2604.24348%2C2605.05509%2C2605.08876%2C2605.20833%2C2605.24785%2C2607.00333%2C2607.18659%2C2607.25255%2C2608.00202%2C2608.08939%2C2608.24913%2C2608.29596&max_results=32"}], "id": "arxiv:2605.24785", "identifiers": {"arxiv": "2605.24785"}, "kind": "paper", "last_verified_at": "2026-10-18T10:54:12Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2605.24785"}], "name": "PANDO: Efficient Multimodal AI Agents via Online Skill Distillation", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-05-24", "relations": [], "review_status": "curated", "scope": "core", "summary": "Analyzes inefficiency in VisualWebArena trajectories and distills reusable online skills to reduce repeated actions and token overhead in multimodal web agents.", "topics": ["efficiency", "memory", "web-navigation"], "updated_at": "2026-05-26", "url": "https://arxiv.org/abs/2605.24785"}
{"aliases": [], "authors": ["Runxi Huang", "Liyu Zhang", "Shengzhong Liu", "Xiaomin Ouyang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "97467a9a1d3c456a0078e96ebb128efa43412a3139a7a68beb4e6dd7b5f0674d", "submitted_at": "2026-05-26T04:53:53Z", "updated_at": "2026-05-26T04:53:53Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2605.26546v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2605.26546", "identifiers": {"arxiv": "2605.26546"}, "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/2605.26546"}], "name": "MobileExplorer: Accelerating On-Device Inference for Mobile GUI Agents via Online Exploration", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-05-26", "relations": [], "review_status": "curated", "scope": "core", "summary": "MobileExplorer probes relevant UI elements while an on-device model is reasoning, then supplies the exploration trace as compact context. A two-level rollback mechanism restores the working state when simple backtracking fails.", "topics": ["mobile-navigation", "efficiency", "memory", "robustness"], "updated_at": "2026-05-26", "url": "https://arxiv.org/abs/2605.26546"}
{"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": ["Yuxiang Chai", "Han Xiao", "Xinyu Fu", "Jinpeng Chen", "Rui Liu", "Hongsheng Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "461790f351a2816004027cc58d07f79ee7c670ed5064c59b1651a4efb84ff2bf", "submitted_at": "2026-05-28T07:49:09Z", "updated_at": "2026-05-28T07:49:09Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2605.29534v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2605.29534", "identifiers": {"arxiv": "2605.29534"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2605.29534"}], "name": "UI-KOBE: Knowledge-Oriented Behavior Exploration for Lightweight Graph-Guided GUI Agents", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-05-28", "relations": [], "review_status": "curated", "scope": "core", "summary": "UI-KOBE explores mobile applications to build graphs of UI states and executable transitions. A lightweight agent then matches its current screenshot to the graph and chooses a transition, local action, completion signal or fallback action.", "topics": ["mobile-navigation", "planning", "state-tracking", "memory", "efficiency"], "updated_at": "2026-05-28", "url": "https://arxiv.org/abs/2605.29534"}
{"aliases": [], "authors": ["Jiaxi Li", "Ke Deng", "Yun Wang", "Jingyuan Huang", "Yucheng Shi", "Qiaoyu Tan", "Jin Lu", "Ninghao Liu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6d891956bde9b8672859fcfcd2d54dfd3e4720964c25d77cc28bc1200ef3a56c", "submitted_at": "2026-06-03T03:11:50Z", "updated_at": "2026-06-03T03:11:50Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.04391v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2606.04391", "identifiers": {"arxiv": "2606.04391"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2606.04391"}], "name": "Online Skill Learning for Web Agents via State-Grounded Dynamic Retrieval", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-06-03", "relations": [], "review_status": "curated", "scope": "core", "summary": "State-Grounded Dynamic Retrieval extracts reusable sub-procedures from completed web trajectories and represents them as paired text and code. Skills are retrieved against both the current webpage state and the task goal at intermediate execution steps.", "topics": ["web-navigation", "memory", "interaction-history", "planning", "workflow"], "updated_at": "2026-06-03", "url": "https://arxiv.org/abs/2606.04391"}
{"aliases": [], "authors": ["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": ["Radeen Mostafa", "Sawradip Saha"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d29dec745923e652654521f65a32f919a99d4577992006efc1a5e469bf775d81", "submitted_at": "2026-06-08T12:18:22Z", "updated_at": "2026-06-08T12:18:22Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.09399v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:02Z", "role": "primary", "sha256": "e436adf2735cb537ee4b24aa0c3f36930a50c2f683cc6868e7c40df602bdb6b8", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2604.00892%2C2604.01221%2C2604.01676%2C2604.02345%2C2604.03465%2C2604.05157%2C2604.06134%2C2604.06240%2C2604.06995%2C2604.07277%2C2604.07831%2C2604.07929%2C2604.08005%2C2604.09581%2C2604.10261%2C2604.10577%2C2604.10988%2C2604.11201%2C2604.18847%2C2604.21375%2C2604.24441%2C2604.26020%2C2605.00551%2C2605.02168%2C2605.07505%2C2605.10347%2C2605.12294%2C2605.15040%2C2605.16883%2C2605.27134%2C2605.28258%2C2606.01790%2C2606.02031%2C2606.02208%2C2606.03203%2C2606.03236%2C2606.03951%2C2606.04391%2C2606.04627%2C2606.04701%2C2606.05233%2C2606.05597%2C2606.05697%2C2606.06560%2C2606.06708%2C2606.07027%2C2606.07074%2C2606.09399%2C2606.09447%2C2606.09764&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:02Z", "role": "primary", "sha256": "e436adf2735cb537ee4b24aa0c3f36930a50c2f683cc6868e7c40df602bdb6b8", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?id_list=2604.00892%2C2604.01221%2C2604.01676%2C2604.02345%2C2604.03465%2C2604.05157%2C2604.06134%2C2604.06240%2C2604.06995%2C2604.07277%2C2604.07831%2C2604.07929%2C2604.08005%2C2604.09581%2C2604.10261%2C2604.10577%2C2604.10988%2C2604.11201%2C2604.18847%2C2604.21375%2C2604.24441%2C2604.26020%2C2605.00551%2C2605.02168%2C2605.07505%2C2605.10347%2C2605.12294%2C2605.15040%2C2605.16883%2C2605.27134%2C2605.28258%2C2606.01790%2C2606.02031%2C2606.02208%2C2606.03203%2C2606.03236%2C2606.03951%2C2606.04391%2C2606.04627%2C2606.04701%2C2606.05233%2C2606.05597%2C2606.05697%2C2606.06560%2C2606.06708%2C2606.07027%2C2606.07074%2C2606.09399%2C2606.09447%2C2606.09764&max_results=50"}], "id": "arxiv:2606.09399", "identifiers": {"arxiv": "2606.09399"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:02Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2606.09399"}], "name": "RunAgent SuperBrowser: A Theory of Autonomous Web Navigation Grounded in Human Browsing Behaviour", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-06-08", "relations": [], "review_status": "curated", "scope": "core", "summary": "SUPERBROWSER combines screenshot-based target proposals with separate orchestration, planning and action roles. A bounded ledger and context-eviction process retain recent actions, task facts and checkpoints during browser execution.", "topics": ["web-navigation", "grounding", "planning", "memory", "efficiency"], "updated_at": "2026-06-08", "url": "https://arxiv.org/abs/2606.09399"}
{"aliases": [], "authors": ["Lawrence Keunho Jang", "Mareks Woodside", "Geronimo Carom", "Andrew Keunwoo Jang", "Jing Yu Koh", "Ruslan Salakhutdinov"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "495420bd87ba06ae2ec1763865b1145271e964b4bd40145c8e591653a1581c42", "submitted_at": "2026-06-08T17:27:13Z", "updated_at": "2026-06-08T17:27:13Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.09764v1"}}, "environments": ["ios"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:02Z", "role": "primary", "sha256": "e436adf2735cb537ee4b24aa0c3f36930a50c2f683cc6868e7c40df602bdb6b8", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2604.00892%2C2604.01221%2C2604.01676%2C2604.02345%2C2604.03465%2C2604.05157%2C2604.06134%2C2604.06240%2C2604.06995%2C2604.07277%2C2604.07831%2C2604.07929%2C2604.08005%2C2604.09581%2C2604.10261%2C2604.10577%2C2604.10988%2C2604.11201%2C2604.18847%2C2604.21375%2C2604.24441%2C2604.26020%2C2605.00551%2C2605.02168%2C2605.07505%2C2605.10347%2C2605.12294%2C2605.15040%2C2605.16883%2C2605.27134%2C2605.28258%2C2606.01790%2C2606.02031%2C2606.02208%2C2606.03203%2C2606.03236%2C2606.03951%2C2606.04391%2C2606.04627%2C2606.04701%2C2606.05233%2C2606.05597%2C2606.05697%2C2606.06560%2C2606.06708%2C2606.07027%2C2606.07074%2C2606.09399%2C2606.09447%2C2606.09764&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:02Z", "role": "primary", "sha256": "e436adf2735cb537ee4b24aa0c3f36930a50c2f683cc6868e7c40df602bdb6b8", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2604.00892%2C2604.01221%2C2604.01676%2C2604.02345%2C2604.03465%2C2604.05157%2C2604.06134%2C2604.06240%2C2604.06995%2C2604.07277%2C2604.07831%2C2604.07929%2C2604.08005%2C2604.09581%2C2604.10261%2C2604.10577%2C2604.10988%2C2604.11201%2C2604.18847%2C2604.21375%2C2604.24441%2C2604.26020%2C2605.00551%2C2605.02168%2C2605.07505%2C2605.10347%2C2605.12294%2C2605.15040%2C2605.16883%2C2605.27134%2C2605.28258%2C2606.01790%2C2606.02031%2C2606.02208%2C2606.03203%2C2606.03236%2C2606.03951%2C2606.04391%2C2606.04627%2C2606.04701%2C2606.05233%2C2606.05597%2C2606.05697%2C2606.06560%2C2606.06708%2C2606.07027%2C2606.07074%2C2606.09399%2C2606.09447%2C2606.09764&max_results=50"}], "id": "arxiv:2606.09764", "identifiers": {"arxiv": "2606.09764"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:02Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2606.09764"}], "name": "iOSWorld: A Benchmark for Personally Intelligent Phone Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-06-08", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces a native iOS simulator benchmark with persistent user data, multi-app tasks, and personalization-dependent phone workflows.", "topics": ["evaluation", "cross-app", "memory"], "updated_at": "2026-06-08", "url": "https://arxiv.org/abs/2606.09764"}
{"aliases": [], "authors": ["Jayoo Hwang", "Xiaowen Zhang", "Vedant Padwal"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "386bd82be97d5cbe64fdb3478e427362e710d78d423c9e52661a8a96ea2fdecf", "submitted_at": "2026-06-09T04:53:19Z", "updated_at": "2026-06-09T04:53:19Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.10423v1"}}, "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:2606.10423", "identifiers": {"arxiv": "2606.10423"}, "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/2606.10423"}], "name": "WebChallenger: A Reliable and Efficient Generalist Web Agent", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-06-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "WebChallenger builds a hierarchical DOM-derived page representation for selective observation, reusable site maps and compound actions. The framework combines these mechanisms without model fine-tuning and evaluates generalization across multiple web-agent benchmarks.", "topics": ["web-navigation", "memory", "efficiency", "training-free", "workflow"], "updated_at": "2026-06-09", "url": "https://arxiv.org/abs/2606.10423"}
{"aliases": [], "authors": ["Jaewoo Lee", "Zaid Khan", "Archiki Prasad", "Justin Chih-Yao Chen", "Supriyo Chakraborty", "Kartik Balasubramaniam", "Sambit Sahu", "Elias Stengel-Eskin", "Hyunji Lee", "Mohit Bansal"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a8ed399c8b489a80c5313a00232d2ecca0fbc8228b1f56745c02d4fedb6f6908", "submitted_at": "2026-06-09T16:39:10Z", "updated_at": "2026-06-09T16:39:10Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.11078v1"}}, "environments": ["web", "desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2606.11078", "identifiers": {"arxiv": "2606.11078"}, "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/2606.11078"}], "name": "A History-Aware Visually Grounded Critic for Computer Use Agents", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-06-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "HiViG combines a compact history of completed actions with a visually grounded critic that checks proposed coordinates before execution. Its evaluations separate history-related planning failures from errors caused by targeting the wrong screen element.", "topics": ["interaction-history", "memory", "grounding", "reward-models", "long-horizon", "evaluation"], "updated_at": "2026-06-09", "url": "https://arxiv.org/abs/2606.11078"}
{"aliases": [], "authors": ["Seoyoung Choi", "Minseok Ko", "Hyunseok Lee", "Kunwoong Kim", "Woomin Song", "Chanseok Jeon", "Jinwoo Shin"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "1fb87da27071ab21519599b7d16359a38118aa358f620de94e66b15cad040828", "submitted_at": "2026-06-12T04:39:01Z", "updated_at": "2026-06-12T04:39:01Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.14106v1"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2606.14106", "identifiers": {"arxiv": "2606.14106"}, "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/2606.14106"}], "name": "Naive Visual Memory is Not Enough: A Failure-Mode Study of GUI Agents", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-06-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "This study finds that full-image visual memory can reduce some state errors while worsening action-level failures. Its AGMem alternative stores local crops associated with successful actions or recovery and evaluates their use in desktop GUI tasks.", "topics": ["memory", "interaction-history", "grounding", "robustness", "evaluation"], "updated_at": "2026-06-12", "url": "https://arxiv.org/abs/2606.14106"}
{"aliases": [], "authors": ["Sina Hajimiri", "Masih Aminbeidokhti", "Jose Dolz", "Ismail Ben Ayed", "Issam H. Laradji", "Spandana Gella", "Nicolas Gontier"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "50f1acd5727f84946cf95d181f270dbc4f46f15d24339e6cb257fea1b23b21a0", "submitted_at": "2026-06-12T23:30:14Z", "updated_at": "2026-08-30T17:49:13Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2606.15017v2"}}, "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:2606.15017", "identifiers": {"arxiv": "2606.15017"}, "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/2606.15017"}], "name": "Are Online Skill and Memory Modules Always Worth Their Tokens? A Budget-Constrained Study of Web Agents", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-06-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "Compares online skill and memory augmentations with a vanilla web agent given the same total inference-token budget. Across WebArena and WorkArena settings, it studies when module overhead removes aggregate gains and how rerun variability changes conclusions.", "topics": ["memory", "efficiency", "evaluation", "robustness"], "updated_at": "2026-08-30", "url": "https://arxiv.org/abs/2606.15017"}
{"aliases": [], "authors": ["Bojie Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "19fd8d8363ef6a7c58e42309f25dea1b9c5b6bca09b5d823f8a7cc7a80d61f0f", "submitted_at": "2026-06-16T13:40:21Z", "updated_at": "2026-06-16T13:40:21Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.17929v1"}}, "environments": ["cross-platform"], "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:2606.17929", "identifiers": {"arxiv": "2606.17929"}, "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/2606.17929"}], "name": "PreAct: Computer-Using Agents that Get Faster on Repeated Tasks", "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": "2026-06-16", "relations": [], "review_status": "curated", "scope": "core", "summary": "PreAct compiles successful screen-interaction runs into guarded state-machine programs for repeated tasks. It validates programs from clean states before storage and returns control to the agent when replayed screen expectations fail, with tests across mobile, desktop and web.", "topics": ["recording", "memory", "efficiency", "automation"], "updated_at": "2026-06-16", "url": "https://arxiv.org/abs/2606.17929"}
{"aliases": [], "authors": ["Ziyan Jiang", "Li An", "Yujian Liu", "Jiabao Ji", "Qiucheng Wu", "Jacob Andreas", "Yang Zhang", "Shiyu Chang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2605cedb4359aa0cf1e694455903871ed334dd37b7b8e15edf31bac9d5e51b13", "submitted_at": "2026-06-16T19:57:07Z", "updated_at": "2026-06-16T19:57:07Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.18448v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2606.18448", "identifiers": {"arxiv": "2606.18448"}, "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/2606.18448"}], "name": "VISUALSKILL: Multimodal Skills for Computer-Use Agents", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-06-16", "relations": [], "review_status": "curated", "scope": "core", "summary": "VISUALSKILL organizes application-specific text and figures into hierarchical skills retrieved on demand through an MCP tool. Skills combine authored documentation with live UI exploration, and matched multimodal/text-only comparisons test their utility for interface identification and workflow-state verification.", "topics": ["memory", "tool-integration", "screen-understanding", "long-horizon"], "updated_at": "2026-06-16", "url": "https://arxiv.org/abs/2606.18448"}
{"aliases": [], "authors": ["Guangyi Liu", "Gao Wu", "Congxiao Liu", "Pengxiang Zhao", "Liang Liu", "Mading Li", "Qi Zhang", "Mengyan Wang", "Liang Guo", "Yong Liu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "cbbb1bd57057e465ce9d991d3f6e2bb52112fc133a5846fc8cb6ce51eddc2121", "submitted_at": "2026-06-18T08:26:09Z", "updated_at": "2026-06-18T08:26:09Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.19926v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2606.19926", "identifiers": {"arxiv": "2606.19926"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2606.19926"}], "name": "MemGUI-Agent: An End-to-End Long-Horizon Mobile GUI Agent with Proactive Context Management", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-06-18", "relations": [], "review_status": "curated", "scope": "core", "summary": "MemGUI-Agent makes context management an action chosen by the same policy that controls the mobile UI. Structured history, UI-state and recent-step fields are trained from annotated trajectories to retain cross-app facts while limiting context growth.", "topics": ["memory", "interaction-history", "long-horizon", "imitation-learning", "data-generation"], "updated_at": "2026-06-18", "url": "https://arxiv.org/abs/2606.19926"}
{"aliases": [], "authors": ["Yuexing Hao", "Xiaomin Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "b370491b37d5bf25c624f2bac654e438459c8084b5da292fb18a0bf9142adaa2", "submitted_at": "2026-06-18T15:25:42Z", "updated_at": "2026-06-18T15:25:42Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.20363v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2606.20363", "identifiers": {"arxiv": "2606.20363"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:36Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2606.20363"}], "name": "Automating SKILL.md Generation for Computer-Using Agents via Interaction Trajectory Mining", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-06-18", "relations": [], "review_status": "curated", "scope": "core", "summary": "Segments and clusters GUI trajectories into inspectable skills, then trains a skill-aware policy from those annotations. Its diagnostic results distinguish readable source-domain clusters from reliable transfer and report limited or absent downstream gains.", "topics": ["memory", "data-generation", "reinforcement-learning", "evaluation"], "updated_at": "2026-06-18", "url": "https://arxiv.org/abs/2606.20363"}
{"aliases": [], "authors": ["Bojie Li", "Noah Shi"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "54de4e93f4f3e39d904b4661a193765e02c29f8ec7787264aee5616579c32210", "submitted_at": "2026-06-28T15:59:31Z", "updated_at": "2026-06-28T15:59:31Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.29472v1"}}, "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"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%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:2606.29472", "identifiers": {"arxiv": "2606.29472"}, "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/2606.29472"}], "name": "Agent-Computer Observation Interfaces Enable Dynamic 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": "2026-06-28", "relations": [], "review_status": "curated", "scope": "core", "summary": "The Agent-Computer Observation Interface separates continuous perception from discrete computer actions using captured frames, audio transcription and persistent visual narration. DynaCU-Bench tests these components on dynamic browser tasks and static controls, with model-specific ablations showing that additional observations can also hurt.", "topics": ["evaluation", "video", "screen-understanding", "memory", "training-free"], "updated_at": "2026-06-28", "url": "https://arxiv.org/abs/2606.29472"}
{"aliases": [], "authors": ["Yixiong Chen", "Xinyi Bai", "Alan Yuille"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "0473ced7881ed2007aa49c2767860c33719d5bf1c798dfe4771a4061b13d655e", "submitted_at": "2026-07-12T07:05:31Z", "updated_at": "2026-07-12T07:05:31Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2607.10608v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T12:31:32Z", "role": "primary", "sha256": "5128028eea81e3ae80715cb038513012598c0c81a7cbfe8c0d4d6adb6c00e518", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2607.03726%2C2607.05120%2C2607.05465%2C2607.05743%2C2607.06008%2C2607.06595%2C2607.07403%2C2607.07744%2C2607.08252%2C2607.08964%2C2607.10455%2C2607.10608%2C2607.12406%2C2607.12756%2C2607.13085%2C2607.13594%2C2607.13884%2C2607.13988%2C2607.15899%2C2607.16352%2C2607.16636%2C2607.16900%2C2607.17044%2C2607.20064%2C2607.20121%2C2607.21522%2C2607.22738%2C2607.23722%2C2607.24720%2C2607.26300%2C2607.26769%2C2607.27881%2C2607.28229%2C2607.28272%2C2607.28430%2C2608.00548%2C2608.02163%2C2608.03223%2C2608.03451%2C2608.03499%2C2608.05013%2C2608.05430%2C2608.06811%2C2608.07873%2C2608.09268%2C2608.10357%2C2608.10669%2C2608.11274%2C2608.11434%2C2608.15089&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:31:32Z", "role": "primary", "sha256": "5128028eea81e3ae80715cb038513012598c0c81a7cbfe8c0d4d6adb6c00e518", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2607.03726%2C2607.05120%2C2607.05465%2C2607.05743%2C2607.06008%2C2607.06595%2C2607.07403%2C2607.07744%2C2607.08252%2C2607.08964%2C2607.10455%2C2607.10608%2C2607.12406%2C2607.12756%2C2607.13085%2C2607.13594%2C2607.13884%2C2607.13988%2C2607.15899%2C2607.16352%2C2607.16636%2C2607.16900%2C2607.17044%2C2607.20064%2C2607.20121%2C2607.21522%2C2607.22738%2C2607.23722%2C2607.24720%2C2607.26300%2C2607.26769%2C2607.27881%2C2607.28229%2C2607.28272%2C2607.28430%2C2608.00548%2C2608.02163%2C2608.03223%2C2608.03451%2C2608.03499%2C2608.05013%2C2608.05430%2C2608.06811%2C2608.07873%2C2608.09268%2C2608.10357%2C2608.10669%2C2608.11274%2C2608.11434%2C2608.15089&max_results=50"}], "id": "arxiv:2607.10608", "identifiers": {"arxiv": "2607.10608"}, "kind": "paper", "last_verified_at": "2026-10-17T12:31:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2607.10608"}], "name": "The Compliance Trap: Diagnosing How AI Agents Consume Conflicting Memory", "notes": "Reviewed against the complete retained official abstract. Experimental findings and release availability have not been independently reproduced or verified.", "published_at": "2026-07-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "The Entry–Propagation–Recovery framework traces where conflicting memory first changes an agent action, how the effect propagates and whether recovery occurs. WebArena and the controlled MemTrapBench separate memory-consumption failures across these stages.", "topics": ["memory", "safety-security", "robustness", "evaluation", "long-horizon"], "updated_at": "2026-07-12", "url": "https://arxiv.org/abs/2607.10608"}
{"aliases": [], "authors": ["Yunxin Li", "Jinchao Li", "Shibo Su", "Zhenran Xu", "Chenrui Zhao", "Tongshu Bian", "Xiaoman Liang", "Meishan Zhang", "Baotian Hu", "Min Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "281de2bb6e45a58efd047735accc2049deab0f8815931c86abbd38b86b6d23bf", "submitted_at": "2026-07-14T11:04:01Z", "updated_at": "2026-07-15T04:41:47Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2607.12625v2"}}, "environments": ["desktop", "android", "ios", "harmonyos", "windows"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:09Z", "role": "primary", "sha256": "65cc3856c70a0fb267f9996872e077de10e9f54821ab1ffb1d82f0a92fa11dbf", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2607.05277%2C2607.06118%2C2607.06306%2C2607.07820%2C2607.11185%2C2607.11388%2C2607.12625%2C2607.12640%2C2607.13027%2C2607.14277%2C2607.14443%2C2607.14548%2C2607.15193%2C2607.15550%2C2607.17050%2C2607.18046%2C2607.19865%2C2607.21557%2C2607.22798%2C2607.23263%2C2607.24112%2C2607.24167%2C2607.25904%2C2607.26041%2C2607.27155%2C2607.28074%2C2607.28227%2C2607.28609%2C2607.29320%2C2608.01964%2C2608.02352%2C2608.02643%2C2608.03270%2C2608.03327%2C2608.04530%2C2608.04741%2C2608.05587%2C2608.05784%2C2608.05891%2C2608.06065%2C2608.06171%2C2608.06477%2C2608.09278%2C2608.09654%2C2608.10366%2C2608.10692%2C2608.11588%2C2608.12429%2C2608.15930%2C2608.17310&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:09Z", "role": "primary", "sha256": "65cc3856c70a0fb267f9996872e077de10e9f54821ab1ffb1d82f0a92fa11dbf", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?id_list=2607.05277%2C2607.06118%2C2607.06306%2C2607.07820%2C2607.11185%2C2607.11388%2C2607.12625%2C2607.12640%2C2607.13027%2C2607.14277%2C2607.14443%2C2607.14548%2C2607.15193%2C2607.15550%2C2607.17050%2C2607.18046%2C2607.19865%2C2607.21557%2C2607.22798%2C2607.23263%2C2607.24112%2C2607.24167%2C2607.25904%2C2607.26041%2C2607.27155%2C2607.28074%2C2607.28227%2C2607.28609%2C2607.29320%2C2608.01964%2C2608.02352%2C2608.02643%2C2608.03270%2C2608.03327%2C2608.04530%2C2608.04741%2C2608.05587%2C2608.05784%2C2608.05891%2C2608.06065%2C2608.06171%2C2608.06477%2C2608.09278%2C2608.09654%2C2608.10366%2C2608.10692%2C2608.11588%2C2608.12429%2C2608.15930%2C2608.17310&max_results=50"}], "id": "arxiv:2607.12625", "identifiers": {"arxiv": "2607.12625"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:09Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2607.12625"}], "name": "KnowAct-GUIClaw: Know Deeply, Act Perfectly, Personal GUI Assistant with Self-Evolving Memory and Skill", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-07-14", "relations": [], "review_status": "curated", "scope": "core", "summary": "KnowAct-GUIClaw combines a host planner with pluggable GUI subagents, attributable experience memory and an evolving skill library. It uses user profiles and feedback for task decomposition and evaluates transfer across Android, iOS, HarmonyOS and Windows.", "topics": ["memory", "planning", "interaction-history", "mobile-navigation"], "updated_at": "2026-07-15", "url": "https://arxiv.org/abs/2607.12625"}
{"aliases": [], "authors": ["Yucheng Jiang", "Zora Zhiruo Wang", "Ruishi Chen", "Diyi Yang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a85b3a99d551fcdffdd306d839b39cdfbe7f0f7654aa9adbf81f939aa874e3b9", "submitted_at": "2026-08-20T17:57:00Z", "updated_at": "2026-08-20T17:57:00Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2608.20319v1"}}, "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:2608.20319", "identifiers": {"arxiv": "2608.20319"}, "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/2608.20319"}], "name": "Inducing Task Models from Computer-Use Traces", "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": "2026-08-20", "relations": [], "review_status": "curated", "scope": "core", "summary": "Task Model Induction disentangles concurrent activities in screenshot and mouse/keyboard traces, then constructs hierarchical goals and procedural control flow for each task. Evaluation tests reconstruction of observed execution and reuse of derived skills on held-out tasks.", "topics": ["recording", "workflow", "memory", "interaction-history"], "updated_at": "2026-08-20", "url": "https://arxiv.org/abs/2608.20319"}
{"aliases": [], "authors": ["Jiaxuan Luo", "Zhanfeng Liao", "Jiayao Teng", "Yuan Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2ea33a82ab7543a64b60ee3024c6f3e39f88b4e3f06eff94f8c790c56cf76715", "submitted_at": "2026-08-23T20:04:21Z", "updated_at": "2026-08-25T04:24:36Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2608.22577v2"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2608.22577", "identifiers": {"arxiv": "2608.22577"}, "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/2608.22577"}], "name": "CausalCache: Conditional High-Fidelity Restoration for Long-Horizon GUI Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-08-23", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces CausalCache, which selects archived screenshots to restore alongside text summaries under a fixed visual-memory budget. A history-gated adapter processes restored images, with evaluation distinguishing standard step limits from longer-horizon and mobile-transfer settings.", "topics": ["memory", "long-horizon", "efficiency"], "updated_at": "2026-08-25", "url": "https://arxiv.org/abs/2608.22577"}
{"aliases": [], "authors": ["Sanket Badhe", "Deep Shah", "Priyanka Tiwari", "Nehal Kathrotia"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "13b6833243a543eb156ce7f4d943f8a3c608b3b38b7eaa77605ca7484cdedc43", "submitted_at": "2026-08-30T06:36:42Z", "updated_at": "2026-09-28T17:25:01Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2608.29596v2"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-18T10:54:12Z", "role": "primary", "sha256": "5f5bae19420f79231516d4bcf604ea18d4f930fc7790cde5a3e7fcb8d41a1126", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2508.19461%2C2508.19493%2C2509.21782%2C2509.25140%2C2510.03485%2C2510.07809%2C2510.14276%2C2510.27140%2C2510.27210%2C2511.13248%2C2512.08417%2C2601.07263%2C2602.13516%2C2603.00476%2C2603.11862%2C2603.18762%2C2603.23791%2C2604.00986%2C2604.16515%2C2604.16548%2C2604.24348%2C2605.05509%2C2605.08876%2C2605.20833%2C2605.24785%2C2607.00333%2C2607.18659%2C2607.25255%2C2608.00202%2C2608.08939%2C2608.24913%2C2608.29596&max_results=32"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T10:54:12Z", "role": "primary", "sha256": "5f5bae19420f79231516d4bcf604ea18d4f930fc7790cde5a3e7fcb8d41a1126", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2508.19461%2C2508.19493%2C2509.21782%2C2509.25140%2C2510.03485%2C2510.07809%2C2510.14276%2C2510.27140%2C2510.27210%2C2511.13248%2C2512.08417%2C2601.07263%2C2602.13516%2C2603.00476%2C2603.11862%2C2603.18762%2C2603.23791%2C2604.00986%2C2604.16515%2C2604.16548%2C2604.24348%2C2605.05509%2C2605.08876%2C2605.20833%2C2605.24785%2C2607.00333%2C2607.18659%2C2607.25255%2C2608.00202%2C2608.08939%2C2608.24913%2C2608.29596&max_results=32"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T13:00:09.204016+00:00", "role": "primary", "sha256": "d112d0a83472b87de538d0290efdd2d0dd0d21e878acca5f383da5baa880fe88", "supports": ["scope", "environments", "notes"], "url": "https://arxiv.org/html/2608.29596v2"}], "id": "arxiv:2608.29596", "identifiers": {"arxiv": "2608.29596"}, "kind": "survey", "last_verified_at": "2026-10-18T10:54:12Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2608.29596"}], "name": "A Systematic Survey of Agentic Skills: Architecture, Lifecycle, and Security", "notes": "Adjacent broad survey of procedural agent skills. Full-text Section 4.4.2 explicitly covers desktop, mobile, and web interface skills; surveyed systems and their reported results are not independently reproduced.", "published_at": "2026-08-30", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "Surveys modular agentic skill architectures, skill lifecycles, and security risks in autonomous tool- and interface-using agents.", "topics": ["survey", "workflow", "memory", "safety-security"], "updated_at": "2026-09-28", "url": "https://arxiv.org/abs/2608.29596"}
{"aliases": [], "authors": ["Longtao Hu", "Xiao Liang", "Linchao Zhu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "78db974038e0dde9a1347132382573fa9e2aafb1b0e1a574da3d769574329be9", "submitted_at": "2026-09-04T08:29:45Z", "updated_at": "2026-09-04T08:29:45Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.04869v1"}}, "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:2609.04869", "identifiers": {"arxiv": "2609.04869"}, "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/2609.04869"}], "name": "From Interaction Traces to Persistent Skills: Online Evolution for Computer-Use Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-09-04", "relations": [], "review_status": "curated", "scope": "core", "summary": "An online skill-evolution framework turns GUI interaction traces and evaluator feedback into a versioned library of reusable procedures without updating model parameters. Controlled OSWorld comparisons examine both the benefits of shared procedural memory and its domain-dependent instability under repeated revision.", "topics": ["memory", "workflow", "evaluation", "training-free"], "updated_at": "2026-09-04", "url": "https://arxiv.org/abs/2609.04869"}
{"aliases": [], "authors": ["Bowei He", "Xiaokun Zhang", "Meng Ding", "Xue Liu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "edbe52f304634dfa06e0f232090a4fe998c7ea111418f57942976d348c4e0997", "submitted_at": "2026-08-31T08:16:56Z", "updated_at": "2026-08-31T08:16:56Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.05511v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2609.05511", "identifiers": {"arxiv": "2609.05511"}, "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/2609.05511"}], "name": "SCAFFOLD: Self-Improving Web Agents via Recursive Parametric Skill Abstraction", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-08-31", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces SCAFFOLD, which extracts executable skills from web trajectories, composes them into a hierarchy, and compresses redundant skills through behavioral checks. It periodically distills skill-guided trajectories into the policy and studies repeated improvement across web benchmarks.", "topics": ["memory", "automation", "web-navigation"], "updated_at": "2026-08-31", "url": "https://arxiv.org/abs/2609.05511"}
{"aliases": [], "authors": ["Zirui Shang", "Xin Shu", "Yang Liu", "Zhi Gao", "Xinxiao Wu", "Lifeng Fan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a0d6c92c1bbf777119fdce5e9c4c5ded744c3459f8ae0b10ccc0fc75de11b237", "submitted_at": "2026-09-06T10:50:55Z", "updated_at": "2026-09-06T10:50:55Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.06530v1"}}, "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:2609.06530", "identifiers": {"arxiv": "2609.06530"}, "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/2609.06530"}], "name": "Selective Knowledge Control for Continual GUI Agent Learning over Application Streams", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-09-06", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces activation-conditioned gradient control for GUI agents learning successive applications. It protects inactive application-specific neurons, constrains updates to shared neurons, and accumulates a compact historical state to balance adaptation with retention.", "topics": ["memory", "robustness"], "updated_at": "2026-09-06", "url": "https://arxiv.org/abs/2609.06530"}
{"aliases": [], "authors": ["Xiaolei Li", "Jialun Cao", "Zhijian Hou", "Yuzhi Zhao", "Yepang Liu", "Shing-Chi Cheung"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "3307d90ef09c16afaee160eacc9d2994d142be9b9a26ce9c4a57f61345e2e8ec", "submitted_at": "2026-09-09T11:04:53Z", "updated_at": "2026-09-09T11:04:53Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.10031v1"}}, "environments": ["android"], "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", "notes"], "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:2609.10031", "identifiers": {"arxiv": "2609.10031"}, "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/2609.10031"}], "name": "GraphDroid: Asynchronous LLM-Based Mobile App GUI Testing via History-Aware Exploration and Hybrid Intent Fulfillment", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-09-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "GraphDroid combines clustered memory of visited GUI states with asynchronous test-intent generation. A hybrid executor reserves LLM calls for complex intents and uses heuristics for simpler ones, evaluated through Android app coverage, bug discovery and crash detection.", "topics": ["evaluation", "mobile-navigation", "memory", "interaction-history", "efficiency"], "updated_at": "2026-09-09", "url": "https://arxiv.org/abs/2609.10031"}
{"aliases": [], "authors": ["Alexandru Ianta", "Eleni Stroulia"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "29b615736812b466b0df7133b7070184a1b7c8dacbe4c73fe32b878b654bb78d", "submitted_at": "2026-09-11T19:57:22Z", "updated_at": "2026-09-11T19:57:22Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.13491v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2609.13491", "identifiers": {"arxiv": "2609.13491"}, "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/2609.13491"}], "name": "Token Efficient Task Execution via Application Behavior Modeling for Web Agents", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-09-11", "relations": [], "review_status": "curated", "scope": "core", "summary": "OdoBot builds a model of web-application behavior from successful task demonstrations and uses it to guide task execution. Its Canvas LMS evaluation compares token use and completion with existing web agents.", "topics": ["memory", "efficiency", "web-navigation"], "updated_at": "2026-09-11", "url": "https://arxiv.org/abs/2609.13491"}
{"aliases": [], "authors": ["Shengjie Jin", "Zelong Sun", "Hengbo Xu", "Yanbiao Ma", "Zhiwu Lu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e8a98337941972fadb725906b9d1315a0ca5921859bd04780cacbabb0500e6a5", "submitted_at": "2026-09-14T12:22:26Z", "updated_at": "2026-09-14T12:22:26Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.15457v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:17Z", "role": "primary", "sha256": "5d7c9ff24833b254d33a41a7c4247e830e4d9c371881a1e16151bad5c71cc8b3", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2609.10297%2C2609.10451%2C2609.12394%2C2609.12404%2C2609.13287%2C2609.13491%2C2609.13548%2C2609.13873%2C2609.14239%2C2609.15134%2C2609.15307%2C2609.15364%2C2609.15457%2C2609.16251%2C2609.16635%2C2609.16732%2C2609.16900%2C2609.17653%2C2609.17885%2C2609.18411%2C2609.18690%2C2609.19125%2C2609.19523%2C2609.19843%2C2609.22083%2C2609.22724%2C2609.24277%2C2609.24890%2C2609.25769%2C2609.27273%2C2609.27307%2C2609.27353%2C2609.27461%2C2609.29892%2C2609.30186%2C2609.30604%2C2609.30887%2C2609.31957%2C2609.32036%2C2609.32130%2C2609.32166%2C2609.32351%2C2609.32385%2C2609.32600%2C2609.32679%2C2609.32750%2C2609.32915%2C2609.33304%2C2609.33646%2C2609.34113&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:17Z", "role": "primary", "sha256": "5d7c9ff24833b254d33a41a7c4247e830e4d9c371881a1e16151bad5c71cc8b3", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?id_list=2609.10297%2C2609.10451%2C2609.12394%2C2609.12404%2C2609.13287%2C2609.13491%2C2609.13548%2C2609.13873%2C2609.14239%2C2609.15134%2C2609.15307%2C2609.15364%2C2609.15457%2C2609.16251%2C2609.16635%2C2609.16732%2C2609.16900%2C2609.17653%2C2609.17885%2C2609.18411%2C2609.18690%2C2609.19125%2C2609.19523%2C2609.19843%2C2609.22083%2C2609.22724%2C2609.24277%2C2609.24890%2C2609.25769%2C2609.27273%2C2609.27307%2C2609.27353%2C2609.27461%2C2609.29892%2C2609.30186%2C2609.30604%2C2609.30887%2C2609.31957%2C2609.32036%2C2609.32130%2C2609.32166%2C2609.32351%2C2609.32385%2C2609.32600%2C2609.32679%2C2609.32750%2C2609.32915%2C2609.33304%2C2609.33646%2C2609.34113&max_results=50"}], "id": "arxiv:2609.15457", "identifiers": {"arxiv": "2609.15457"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:17Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.15457"}], "name": "AnchorGUI: Asymmetric Memory for Dual-Scale Learning in GUI Navigation", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-09-14", "relations": [], "review_status": "curated", "scope": "core", "summary": "AnchorGUI compares expected and observed UI transitions and selectively retains screenshots when they disagree. This asymmetric memory supports immediate correction and focuses cross-attempt experience distillation on likely failure steps.", "topics": ["memory", "interaction-history", "state-tracking", "self-reflection", "efficiency"], "updated_at": "2026-09-14", "url": "https://arxiv.org/abs/2609.15457"}
{"aliases": [], "authors": ["Yao Zhao", "Aditya Shanmugham", "Swastik Roy", "Yanxun Xu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e8dcd34beebb8a4660d8d7b9f4c76c3f86069926b9856d26b1c20d87dec83bb5", "submitted_at": "2026-09-15T04:56:17Z", "updated_at": "2026-09-15T04:56:17Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.16635v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2609.16635", "identifiers": {"arxiv": "2609.16635"}, "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/2609.16635"}], "name": "EchoPath: Execution-Level Replayable Memory for GUI Agents", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-09-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "EchoPath turns validated GUI trajectories into parameterized replay procedures with explicit runtime preconditions. Image-based target matching adjusts stored coordinates, while ambiguous steps fall back to bounded repair or fresh planning.", "topics": ["memory", "workflow", "automation", "grounding", "efficiency", "robustness"], "updated_at": "2026-09-15", "url": "https://arxiv.org/abs/2609.16635"}
{"aliases": [], "authors": ["Bofan Chen", "Boxuan Zhang", "Fei Tang", "Zhengxi Lu", "Yong Du", "Tongbo Chen", "Weiming Lu", "Jun Xiao", "Yueting Zhuang", "Yongliang Shen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "975662f216cc9a6b89856d4457f2125b733620e80bfac1b6424775253071d2e6", "submitted_at": "2026-09-15T17:24:06Z", "updated_at": "2026-09-15T17:24:06Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.17653v1"}}, "environments": ["android", "desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2609.17653", "identifiers": {"arxiv": "2609.17653"}, "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/2609.17653"}], "name": "Reflect, Revise, Reuse: Training-Free Skill Evolution for GUI Agents", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-09-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "EvoSkill-GUI revises reusable skill files from execution feedback through an executor–critic loop. It studies how training-free updates to localization, recovery and procedure information help subsequent mobile and desktop tasks.", "topics": ["memory", "training-free", "self-reflection", "robustness", "long-horizon", "workflow"], "updated_at": "2026-09-15", "url": "https://arxiv.org/abs/2609.17653"}
{"aliases": [], "authors": ["Yinzhu Quan", "Zefang Liu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "5e7fb5704932a99147ff7c64a82a6a4041565f3edf6506a6252f75f832834d96", "submitted_at": "2026-09-17T00:29:46Z", "updated_at": "2026-10-08T04:53:11Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2609.19523v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2609.19523", "identifiers": {"arxiv": "2609.19523"}, "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/2609.19523"}], "name": "EconSkills: Studying Skill Transfer and Retrieval for Web Agents on Live Economic Data", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-09-17", "relations": [], "review_status": "curated", "scope": "core", "summary": "EconSkills converts verified economic-data browsing traces into parameterized procedures with navigation, verification and recovery guidance. Its evaluation separates known-skill transfer from library retrieval, exposing the limits of approximate matches to new tasks.", "topics": ["memory", "web-navigation", "evaluation", "workflow"], "updated_at": "2026-10-08", "url": "https://arxiv.org/abs/2609.19523"}
{"aliases": [], "authors": ["Yan Zhang", "Daiqing Wu", "Huawen Shen", "Liang Li", "Gang Cao", "Zhi Gong", "Wei Dai", "Xiaode Zhang", "Can Ma", "Yu Zhou"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "0ae6c83c017a05c2916f31e388f6584cc2cab7b87e1eb0038c33a707bbd66e24", "submitted_at": "2026-09-23T03:38:11Z", "updated_at": "2026-09-23T03:38:11Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.27307v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2609.27307", "identifiers": {"arxiv": "2609.27307"}, "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/2609.27307"}], "name": "Learn How to Act from Your Own Interactions: On-Policy Self-Distillation for GUI Agents", "notes": "The annotation was reviewed against the complete retained official abstract. Experimental findings and artifact availability have not been independently reproduced.", "published_at": "2026-09-23", "relations": [], "review_status": "curated", "scope": "core", "summary": "GUI-SD-v2 extends on-policy self-distillation from grounding to multi-turn GUI interaction. It first strengthens responses to privileged guidance, then selectively distills step-specific reasoning and memory support for later actions.", "topics": ["imitation-learning", "memory", "interaction-history", "mobile-navigation"], "updated_at": "2026-09-23", "url": "https://arxiv.org/abs/2609.27307"}
{"aliases": [], "authors": ["Tingyu Qu", "Weigao Sun", "Yuecheng Liu", "Yucheng Zhao", "Yi Zhu", "Yifeng Ding", "Qiyi Wang", "Sihan Cao", "Pengkun Jiao", "Hanlei Xie", "Xiongwei Wu", "Qichao Wang", "Haodong Zhang", "Jiajun Liu", "Yuhao Wang", "Yuqing Xie", "Junpeng Zhao", "Long Chen", "Ming Ma", "Sihan Yang", "Ziwang Zhao", "Yanhao Jia", "Liangquan Gong", "Feida Zhu", "Yiran Zhong", "Steven Hoi"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "572ba5397fc1e4bb138febc3e7004e6e4f244e0ba8b628fb91debef4a854561f", "submitted_at": "2026-09-24T14:38:08Z", "updated_at": "2026-09-24T14:38:08Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.29892v1"}, "license_review": {"document": {"evidence_urls": ["https://arxiv.org/html/2609.29892v1"], "reason": "Record the paper-specific arXiv document declaration; it does not establish code, data or model-weight permissions."}}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:17Z", "role": "primary", "sha256": "5d7c9ff24833b254d33a41a7c4247e830e4d9c371881a1e16151bad5c71cc8b3", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2609.10297%2C2609.10451%2C2609.12394%2C2609.12404%2C2609.13287%2C2609.13491%2C2609.13548%2C2609.13873%2C2609.14239%2C2609.15134%2C2609.15307%2C2609.15364%2C2609.15457%2C2609.16251%2C2609.16635%2C2609.16732%2C2609.16900%2C2609.17653%2C2609.17885%2C2609.18411%2C2609.18690%2C2609.19125%2C2609.19523%2C2609.19843%2C2609.22083%2C2609.22724%2C2609.24277%2C2609.24890%2C2609.25769%2C2609.27273%2C2609.27307%2C2609.27353%2C2609.27461%2C2609.29892%2C2609.30186%2C2609.30604%2C2609.30887%2C2609.31957%2C2609.32036%2C2609.32130%2C2609.32166%2C2609.32351%2C2609.32385%2C2609.32600%2C2609.32679%2C2609.32750%2C2609.32915%2C2609.33304%2C2609.33646%2C2609.34113&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T15:43:20.402537Z", "role": "primary", "sha256": "d4385231cea2965a48895eeebd2dd6000265dd7b43731a0ad88b0373a21d0308", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "licenses.document"], "url": "https://arxiv.org/html/2609.29892v1"}], "id": "arxiv:2609.29892", "identifiers": {"arxiv": "2609.29892"}, "kind": "paper", "last_verified_at": "2026-10-20T15:43:20.402537Z", "licenses": {"code": null, "data": null, "document": {"basis": "The exact-v1 arXiv HTML infobox (#license-tr) declares License: CC BY 4.0 for this paper. This declaration is document-only.", "declared": "CC-BY-4.0", "retrieved_at": "2026-10-20T15:43:20.402537Z", "url": "https://arxiv.org/html/2609.29892v1"}, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.29892"}], "name": "Qwen-Planner-Agent: A Closed-Loop AI-for-AI Framework for Real-World Mobile Planner Agents", "notes": "Primarily typed tool actions and structured feedback, not pixel-coordinate GUI control; backends combine programmatic sandboxes, LLM simulation and selected real-device sessions. No specific OS is established. Model/harness updates are reviewed offline, not autonomous serving-time changes. Co-evolution scores use an evolve/development set; output-cost estimates exclude input, tools, devices and harness processing. Memory comparisons differ in history access and processing cost; user authorization relies on trusted upstream services. No independent reproduction.", "published_at": "2026-09-24", "relations": [], "review_status": "curated", "scope": "core", "summary": "Trains a mobile planner with supervised trajectories and competence-adaptive RL, then couples typed tool execution with memory, skills and offline model-harness refinement.", "topics": ["planning", "data-generation", "reinforcement-learning", "memory", "mobile-navigation", "efficiency"], "updated_at": "2026-09-24", "url": "https://arxiv.org/abs/2609.29892"}
{"aliases": [], "authors": ["Beining Wu", "Jun Huang", "Yanxiao Zhao"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "719385c97b9d174bbebc34ac4a196971af61521489f7bf032a7d807ae3c8d4c1", "submitted_at": "2026-09-26T01:13:44Z", "updated_at": "2026-09-26T01:13:44Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.32130v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2609.32130", "identifiers": {"arxiv": "2609.32130"}, "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/2609.32130"}], "name": "REBASE: Device-Cloud Experience Coherence for GUI Agents Across App Updates", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-09-26", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces REBASE, a device–cloud protocol for repairing cached GUI experience after app updates. It combines cloud replay, device-side step verification, and progressively richer failure reports to produce repairs tied to application versions.", "topics": ["memory", "robustness", "efficiency", "mobile-navigation"], "updated_at": "2026-09-26", "url": "https://arxiv.org/abs/2609.32130"}
{"aliases": [], "authors": ["Taehwan Park", "Changmin Lee", "Hayeon Lee", "Taesik Gong"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "f343a941e12c0fb8c4e12dc156e7c8b37f753d74836d8e42542d504bee883e51", "submitted_at": "2026-09-26T02:41:56Z", "updated_at": "2026-09-26T02:41:56Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.32166v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2609.32166", "identifiers": {"arxiv": "2609.32166"}, "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/2609.32166"}], "name": "PastForward: Faster On-Device GUI Agents via Computational Experience Reuse", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-09-26", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces PastForward, which reuses earlier output sequences as verified decoding proposals and starts future action computation from predicted GUI transitions. Screen checks determine whether speculative computation and cached model states can be retained.", "topics": ["memory", "efficiency", "mobile-navigation"], "updated_at": "2026-09-26", "url": "https://arxiv.org/abs/2609.32166"}
{"aliases": [], "authors": ["Jinming Hu", "Haodong Zhao", "Qi Jia", "Die Chen", "Tianhang Zhao", "Sufeng Duan", "Gongshen Liu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "ff1c4eff95094ea8ecf37b7bc621eebbd5e4a677b0e11240e802a72b9b229c7f", "submitted_at": "2026-09-26T11:59:39Z", "updated_at": "2026-09-26T11:59:39Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.32511v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-20T15:06:37Z", "role": "primary", "sha256": "a3d753815e82190ee2de6f322cd816c32f2dd391f862845305fc7e57300303a6", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28abs%3A%22desktop+agent%22+OR+abs%3A%22computer+interaction%22+OR+abs%3A%22GUI+automation%22+OR+abs%3A%22web+navigation%22+OR+abs%3A%22browser+automation%22+OR+abs%3A%22visual+grounding%22+OR+abs%3A%22screen+understanding%22+OR+abs%3A%22mobile+agents%22%29+AND+submittedDate%3A%5B202609200000+TO+202610200000%5D&start=0&max_results=200&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T15:08:56Z", "role": "primary", "sha256": "ce7210dccecb2896202639e6f7be41876fee20de1135b0d23278cead2d695339", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/html/2609.32511v1"}], "id": "arxiv:2609.32511", "identifiers": {"arxiv": "2609.32511"}, "kind": "paper", "last_verified_at": "2026-10-20T15:08:56Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.32511"}], "name": "Learning from Others, Acting for You: Cross-User Memory Sharing for LLM Agents", "notes": "Main Mind2Web evaluation uses 252 recorded tasks assigned to 20 simulated users, frozen demonstration-derived memory and task-macro step metrics; it is not a live-browser completion test. The main split withholds same-site local experience, and sharing gains shrink when local coverage rises. Mind2Web has no preference channel; preference claims come from separate service and coding settings. User-bound retrieval is not a demonstrated privacy or security guarantee. Selected sections were source-read; no independent reproduction.", "published_at": "2026-09-26", "relations": [], "review_status": "curated", "scope": "core", "summary": "ShareMem consolidates reusable experience across users while keeping preference retrieval user-bound; its web evaluation predicts target elements and actions on recorded Mind2Web states.", "topics": ["memory", "web-navigation", "action-prediction", "evaluation"], "updated_at": "2026-09-26", "url": "https://arxiv.org/abs/2609.32511"}
{"aliases": [], "authors": ["Asif Shahriar", "Md Nafiu Rahman", "Sadif Ahmed", "Farig Sadeque", "Md Rizwan Parvez"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "802342ce2c486c30c84c33c897becb0348831034d4d11f8d3d3861ec0380b65c", "submitted_at": "2026-09-26T20:09:16Z", "updated_at": "2026-09-26T20:09:16Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.32915v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:17Z", "role": "primary", "sha256": "5d7c9ff24833b254d33a41a7c4247e830e4d9c371881a1e16151bad5c71cc8b3", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2609.10297%2C2609.10451%2C2609.12394%2C2609.12404%2C2609.13287%2C2609.13491%2C2609.13548%2C2609.13873%2C2609.14239%2C2609.15134%2C2609.15307%2C2609.15364%2C2609.15457%2C2609.16251%2C2609.16635%2C2609.16732%2C2609.16900%2C2609.17653%2C2609.17885%2C2609.18411%2C2609.18690%2C2609.19125%2C2609.19523%2C2609.19843%2C2609.22083%2C2609.22724%2C2609.24277%2C2609.24890%2C2609.25769%2C2609.27273%2C2609.27307%2C2609.27353%2C2609.27461%2C2609.29892%2C2609.30186%2C2609.30604%2C2609.30887%2C2609.31957%2C2609.32036%2C2609.32130%2C2609.32166%2C2609.32351%2C2609.32385%2C2609.32600%2C2609.32679%2C2609.32750%2C2609.32915%2C2609.33304%2C2609.33646%2C2609.34113&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:17Z", "role": "primary", "sha256": "5d7c9ff24833b254d33a41a7c4247e830e4d9c371881a1e16151bad5c71cc8b3", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2609.10297%2C2609.10451%2C2609.12394%2C2609.12404%2C2609.13287%2C2609.13491%2C2609.13548%2C2609.13873%2C2609.14239%2C2609.15134%2C2609.15307%2C2609.15364%2C2609.15457%2C2609.16251%2C2609.16635%2C2609.16732%2C2609.16900%2C2609.17653%2C2609.17885%2C2609.18411%2C2609.18690%2C2609.19125%2C2609.19523%2C2609.19843%2C2609.22083%2C2609.22724%2C2609.24277%2C2609.24890%2C2609.25769%2C2609.27273%2C2609.27307%2C2609.27353%2C2609.27461%2C2609.29892%2C2609.30186%2C2609.30604%2C2609.30887%2C2609.31957%2C2609.32036%2C2609.32130%2C2609.32166%2C2609.32351%2C2609.32385%2C2609.32600%2C2609.32679%2C2609.32750%2C2609.32915%2C2609.33304%2C2609.33646%2C2609.34113&max_results=50"}], "id": "arxiv:2609.32915", "identifiers": {"arxiv": "2609.32915"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:17Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.32915"}], "name": "AgentTell: Behavioural Side-Channel Leakage in Browser-Use Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-09-26", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces AgentTell, a benchmark for privacy leakage through browser actions after an agent learns a private fact on another website. It examines whether agents choose revealing options despite nondisclosure instructions and whether their final reports accurately describe that behavior.", "topics": ["safety-security", "evaluation", "memory"], "updated_at": "2026-09-26", "url": "https://arxiv.org/abs/2609.32915"}
{"aliases": [], "authors": ["Fuyao Zhang", "Xuan Wang", "Zherui Li", "Jiaming Zhang", "Longtao Huang", "Wei Yang Bryan Lim"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "563661718e6e64d62f0601bbc5c214828219b9495a729ae685ba1ae9599332f3", "submitted_at": "2026-09-27T07:15:57Z", "updated_at": "2026-09-27T07:15:57Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.33304v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2609.33304", "identifiers": {"arxiv": "2609.33304"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:29Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.33304"}], "name": "Relevance Does Not Imply Applicability: Experience Activation for Personal GUI Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-09-27", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces ExpActivator, which matches the current GUI state to relevant historical states before supplying past actions as guidance. A separate activation rule decides whether a recurring user intent is warranted by the current time and situation or should be withheld.", "topics": ["memory", "planning", "training-free"], "updated_at": "2026-09-27", "url": "https://arxiv.org/abs/2609.33304"}
{"aliases": [], "authors": ["Keliang Li", "Heng Wang", "Chen Hu", "Daxin Jiang", "Hong Chang", "Shiguang Shan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "88daacf5fda62f59a08b504a633ad984ee1d03791bba24e89c223c0686adf34e", "submitted_at": "2026-09-27T15:11:30Z", "updated_at": "2026-09-27T15:11:30Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.33646v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:17Z", "role": "primary", "sha256": "5d7c9ff24833b254d33a41a7c4247e830e4d9c371881a1e16151bad5c71cc8b3", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2609.10297%2C2609.10451%2C2609.12394%2C2609.12404%2C2609.13287%2C2609.13491%2C2609.13548%2C2609.13873%2C2609.14239%2C2609.15134%2C2609.15307%2C2609.15364%2C2609.15457%2C2609.16251%2C2609.16635%2C2609.16732%2C2609.16900%2C2609.17653%2C2609.17885%2C2609.18411%2C2609.18690%2C2609.19125%2C2609.19523%2C2609.19843%2C2609.22083%2C2609.22724%2C2609.24277%2C2609.24890%2C2609.25769%2C2609.27273%2C2609.27307%2C2609.27353%2C2609.27461%2C2609.29892%2C2609.30186%2C2609.30604%2C2609.30887%2C2609.31957%2C2609.32036%2C2609.32130%2C2609.32166%2C2609.32351%2C2609.32385%2C2609.32600%2C2609.32679%2C2609.32750%2C2609.32915%2C2609.33304%2C2609.33646%2C2609.34113&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:17Z", "role": "primary", "sha256": "5d7c9ff24833b254d33a41a7c4247e830e4d9c371881a1e16151bad5c71cc8b3", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2609.10297%2C2609.10451%2C2609.12394%2C2609.12404%2C2609.13287%2C2609.13491%2C2609.13548%2C2609.13873%2C2609.14239%2C2609.15134%2C2609.15307%2C2609.15364%2C2609.15457%2C2609.16251%2C2609.16635%2C2609.16732%2C2609.16900%2C2609.17653%2C2609.17885%2C2609.18411%2C2609.18690%2C2609.19125%2C2609.19523%2C2609.19843%2C2609.22083%2C2609.22724%2C2609.24277%2C2609.24890%2C2609.25769%2C2609.27273%2C2609.27307%2C2609.27353%2C2609.27461%2C2609.29892%2C2609.30186%2C2609.30604%2C2609.30887%2C2609.31957%2C2609.32036%2C2609.32130%2C2609.32166%2C2609.32351%2C2609.32385%2C2609.32600%2C2609.32679%2C2609.32750%2C2609.32915%2C2609.33304%2C2609.33646%2C2609.34113&max_results=50"}], "id": "arxiv:2609.33646", "identifiers": {"arxiv": "2609.33646"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:17Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.33646"}], "name": "Probe to Act: Elevating Browser-Use Agent via Active Visual Probing", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-09-27", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces Probe to Act, which lets browser agents request visual evidence linking DOM handles and screen regions before committing an operation. It retains selected interaction evidence as memory and studies both prompted use and distillation into open models.", "topics": ["grounding", "memory", "web-navigation"], "updated_at": "2026-09-27", "url": "https://arxiv.org/abs/2609.33646"}
{"aliases": [], "authors": ["Ziqiang Wang", "Li Gu", "Zhixiang Chi", "Linlian Jiang", "Zihuan Jiang", "Linqiang Guo", "Siobhan Reid", "Zhi Liu", "Yang Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "7fbaeb0b4f891d64cc12a3d5663b6804df6c7db82e6a7ccfd54aa10e837010de", "submitted_at": "2026-09-28T05:00:45Z", "updated_at": "2026-09-28T05:00:45Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.34321v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2609.34321", "identifiers": {"arxiv": "2609.34321"}, "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/2609.34321"}], "name": "One Rollout Is All You Get: Fully Test-Time Adaptation for GUI Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-09-28", "relations": [], "review_status": "curated", "scope": "core", "summary": "Defines a test-time adaptation setting with one counted attempt per arriving GUI task and no ground-truth feedback. Its SOLO method filters or relabels completed episodes, then updates a small adapter from a short window of the agent’s own experience.", "topics": ["memory", "evaluation", "robustness"], "updated_at": "2026-09-28", "url": "https://arxiv.org/abs/2609.34321"}
{"aliases": [], "authors": ["Ziqi Zhang", "Shaohui Li", "Bing Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "7efd2eafa1984d7c8f4bf6ce79a94cd2ca4189da330bdd056b4be9c07f7f9dc1", "submitted_at": "2026-09-28T04:02:07Z", "updated_at": "2026-09-28T04:02:07Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.35904v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:20Z", "role": "primary", "sha256": "c420ee40cc87a574ae2554556843f27f2eda43af553f88e95affd6319b1dae33", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2609.34321%2C2609.34769%2C2609.34974%2C2609.35026%2C2609.35671%2C2609.35814%2C2609.35904%2C2609.36201%2C2609.36488%2C2609.36923%2C2609.36927%2C2609.36939%2C2609.37053%2C2609.37176%2C2609.37686%2C2609.38008%2C2609.38397%2C2609.38778%2C2609.38867%2C2609.39306%2C2609.39547%2C2609.39903%2C2609.40253%2C2609.40284%2C2610.00948%2C2610.01215%2C2610.01491%2C2610.01787%2C2610.02320%2C2610.02932%2C2610.03036%2C2610.03089%2C2610.03574%2C2610.03818%2C2610.04116%2C2610.04800%2C2610.04906%2C2610.05185%2C2610.05437%2C2610.05861%2C2610.06829%2C2610.07118%2C2610.07127%2C2610.07444%2C2610.07972%2C2610.08699%2C2610.08773%2C2610.09240%2C2610.09469%2C2610.11050&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:20Z", "role": "primary", "sha256": "c420ee40cc87a574ae2554556843f27f2eda43af553f88e95affd6319b1dae33", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2609.34321%2C2609.34769%2C2609.34974%2C2609.35026%2C2609.35671%2C2609.35814%2C2609.35904%2C2609.36201%2C2609.36488%2C2609.36923%2C2609.36927%2C2609.36939%2C2609.37053%2C2609.37176%2C2609.37686%2C2609.38008%2C2609.38397%2C2609.38778%2C2609.38867%2C2609.39306%2C2609.39547%2C2609.39903%2C2609.40253%2C2609.40284%2C2610.00948%2C2610.01215%2C2610.01491%2C2610.01787%2C2610.02320%2C2610.02932%2C2610.03036%2C2610.03089%2C2610.03574%2C2610.03818%2C2610.04116%2C2610.04800%2C2610.04906%2C2610.05185%2C2610.05437%2C2610.05861%2C2610.06829%2C2610.07118%2C2610.07127%2C2610.07444%2C2610.07972%2C2610.08699%2C2610.08773%2C2610.09240%2C2610.09469%2C2610.11050&max_results=50"}], "id": "arxiv:2609.35904", "identifiers": {"arxiv": "2609.35904"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:20Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.35904"}], "name": "Structured Interaction, Visual Localization, and Robust Execution for Complex Web Tasks: A Technical Report on the WebRetriever Challenge", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-09-28", "relations": [], "review_status": "curated", "scope": "core", "summary": "Describes a WebRetriever agent that combines semantic webpage operations with visual localization when structured representations are insufficient. Hierarchical context management and fault-aware execution support processing tasks across concurrent browsers.", "topics": ["web-navigation", "grounding", "robustness", "memory"], "updated_at": "2026-09-28", "url": "https://arxiv.org/abs/2609.35904"}
{"aliases": [], "authors": ["Dongchan Shin", "Xing Han Lù", "Jiaqi Deng", "Jay Gala", "Tomás Vergara Browne", "Jaewon Moon", "Fengyuan Liu", "Alexandre Drouin", "Siva Reddy", "Alexandre Lacoste"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2fe39365133228daa770445f19002f6abe34ca1151cf4952a225ba0001de5250", "submitted_at": "2026-09-29T01:46:48Z", "updated_at": "2026-09-29T01:46:48Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.36488v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2609.36488", "identifiers": {"arxiv": "2609.36488"}, "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/2609.36488"}], "name": "AdaptArena: Evaluating Test-Time Personalization of 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": "2026-09-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces AdaptArena to test whether web agents can infer unstated user preferences from relevant past trajectories and carry those preferences into new tasks. Its AdaptiveAgent baseline separates preference-retrieval challenges from failures in downstream web execution.", "topics": ["evaluation", "memory", "web-navigation"], "updated_at": "2026-09-29", "url": "https://arxiv.org/abs/2609.36488"}
{"aliases": [], "authors": ["Bin Kang", "Jiarui Ouyang", "Li Jiang", "Bin Chen", "Zhuotao Tian"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e3af30198735ca9b9d192e15acbb7f8a37fe222fa2c0c1d2ccff448ef5e79f55", "submitted_at": "2026-09-29T07:41:34Z", "updated_at": "2026-09-29T07:41:34Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.36923v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2609.36923", "identifiers": {"arxiv": "2609.36923"}, "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/2609.36923"}], "name": "PrecogUI: Proactive GUI Agents via Pre-cognitive Simulation and Experience Retrieval", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-09-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "Presents PrecogUI, which retrieves prior anomaly and success patterns and predicts symbolic UI transitions before choosing actions. It also introduces InterfereBench, generated for evaluating long GUI workflows with disruptive events and recovery requirements.", "topics": ["memory", "world-models", "robustness", "long-horizon"], "updated_at": "2026-09-29", "url": "https://arxiv.org/abs/2609.36923"}
{"aliases": [], "authors": ["Giulio Segalini", "Zhi Wen Soi", "Jérémie Decouchant", "Lydia Chen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "945184e8fc790979a1f7e466ef6b4e727ddd11c3f285be0371fa4dda42615f40", "submitted_at": "2026-09-29T10:01:52Z", "updated_at": "2026-09-29T10:01:52Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.37176v1"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2609.37176", "identifiers": {"arxiv": "2609.37176"}, "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/2609.37176"}], "name": "Absorbed in Inertia: Activation Analysis for Computer-Use Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-09-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "Studies repetitive, unproductive actions in desktop agents through their internal activation states. It proposes Reset, Reroute, Restore, a context intervention that temporarily resets an agent’s trajectory and later restores it to help escape repeated ineffective behavior.", "topics": ["robustness", "observability", "memory"], "updated_at": "2026-09-29", "url": "https://arxiv.org/abs/2609.37176"}
{"aliases": [], "authors": ["Hongbo Zhang", "Liuyang Song", "Quanquan Li", "Daqian Yang", "Yan Wen", "Zhengtao Yao"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "bb910e64ebb65bd49e8729e4373c1c81ce4b2d0a32647a5abaf15f52737e578d", "submitted_at": "2026-09-30T02:11:02Z", "updated_at": "2026-09-30T02:11:02Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.38778v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2609.38778", "identifiers": {"arxiv": "2609.38778"}, "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/2609.38778"}], "name": "Action Conditioned Bisimulation For GUI Agent Memory", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-09-30", "relations": [], "review_status": "curated", "scope": "core", "summary": "Defines a GUI-memory merge rule that treats two states as equivalent only when shared actions lead to matching outcomes and successor states. The training-free method replaces observation-similarity merging and is tested in closed-loop MiniWoB++ interaction.", "topics": ["memory", "web-navigation"], "updated_at": "2026-09-30", "url": "https://arxiv.org/abs/2609.38778"}
{"aliases": [], "authors": ["Bo Han", "Qianyi Wang", "Shuai Liu", "Xiong Zifan", "Changqiao Wu", "Yuanfa Li", "Pengzhi Gao", "Wei Liu", "Jian Luan", "Heng Qu", "Yunpeng Song", "Zhongmin Cai"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "3594620e97b41f3a7481871588b63ffa4aa8822e8beb42334b2cd888bcfcfaaf", "submitted_at": "2026-09-30T11:45:51Z", "updated_at": "2026-09-30T11:45:51Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.39547v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2609.39547", "identifiers": {"arxiv": "2609.39547"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:29Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.39547"}], "name": "Learning Reliable GUI Agents under Imperfect Priors", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-09-30", "relations": [], "review_status": "curated", "scope": "core", "summary": "Combines structured GUI exploration with training designed to tolerate imperfect retrieved experience. It builds state-transition graphs and synthetic task trajectories, then injects realistic forms of interface drift so agents learn when to rely on or reject prior guidance.", "topics": ["data-generation", "memory", "robustness"], "updated_at": "2026-09-30", "url": "https://arxiv.org/abs/2609.39547"}
{"aliases": [], "authors": ["Chengguang Gan", "Zimeng He", "Yoshihiro Tsujii", "Ken-ichiro Kobayashi", "Hiroki Itoh", "Kotaro Funakoshi"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "9a5b97264b3928c8dc74b955cc4baa7005cd34b0ce9e42cbc48a63bc6dd1f5a7", "submitted_at": "2026-10-01T11:31:37Z", "updated_at": "2026-10-01T11:31:37Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.01491v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2610.01491", "identifiers": {"arxiv": "2610.01491"}, "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/2610.01491"}], "name": "Auditing Web Agent Evaluation on WebArena-Lite: Human Review of Outcomes and Trajectories", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-10-01", "relations": [], "review_status": "curated", "scope": "core", "summary": "Audits WebArena-Lite outcomes and trajectories with human review to identify automatic-evaluator false negatives and the first consequential execution errors. It also studies explicit execution-state memory and procedural guidance, showing why final task scores can miss substantial differences in agent behavior.", "topics": ["evaluation", "observability", "memory"], "updated_at": "2026-10-01", "url": "https://arxiv.org/abs/2610.01491"}
{"aliases": [], "authors": ["Beining Wu", "Zihao Ding", "Jun Huang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "7e85a601c31c6d06831ed1c590deef29d1c365e5d4751a308e42e8496ef11b19", "submitted_at": "2026-10-01T14:34:32Z", "updated_at": "2026-10-01T14:34:32Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.01787v1"}}, "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:2610.01787", "identifiers": {"arxiv": "2610.01787"}, "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/2610.01787"}], "name": "Not All Experience Belongs in the Weights: Component Routing for Self-Improving GUI Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-10-01", "relations": [], "review_status": "curated", "scope": "core", "summary": "Studies how to route different parts of GUI experience into model weights or runtime context rather than treating whole trajectories as a single unit. It separates locators, procedures, state facts, and lessons, then predicts their destination from recurrence and dependence on the current state.", "topics": ["memory", "evaluation", "efficiency"], "updated_at": "2026-10-01", "url": "https://arxiv.org/abs/2610.01787"}
{"aliases": [], "authors": ["Justin Chih-Yao Chen", "Elias Stengel-Eskin", "Yan Chen", "Pol Llado", "Scott Counts", "Mohit Bansal", "Benjamin Van Durme", "Harsh Jhamtani", "Gaurav Verma"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "b9e40e7a3404f7aef992270e54986e655995ae88dd0c50db55ccd7f61cc90fb3", "submitted_at": "2026-10-04T18:19:33Z", "updated_at": "2026-10-06T03:17:00Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2610.05437v2"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:20Z", "role": "primary", "sha256": "c420ee40cc87a574ae2554556843f27f2eda43af553f88e95affd6319b1dae33", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2609.34321%2C2609.34769%2C2609.34974%2C2609.35026%2C2609.35671%2C2609.35814%2C2609.35904%2C2609.36201%2C2609.36488%2C2609.36923%2C2609.36927%2C2609.36939%2C2609.37053%2C2609.37176%2C2609.37686%2C2609.38008%2C2609.38397%2C2609.38778%2C2609.38867%2C2609.39306%2C2609.39547%2C2609.39903%2C2609.40253%2C2609.40284%2C2610.00948%2C2610.01215%2C2610.01491%2C2610.01787%2C2610.02320%2C2610.02932%2C2610.03036%2C2610.03089%2C2610.03574%2C2610.03818%2C2610.04116%2C2610.04800%2C2610.04906%2C2610.05185%2C2610.05437%2C2610.05861%2C2610.06829%2C2610.07118%2C2610.07127%2C2610.07444%2C2610.07972%2C2610.08699%2C2610.08773%2C2610.09240%2C2610.09469%2C2610.11050&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:20Z", "role": "primary", "sha256": "c420ee40cc87a574ae2554556843f27f2eda43af553f88e95affd6319b1dae33", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2609.34321%2C2609.34769%2C2609.34974%2C2609.35026%2C2609.35671%2C2609.35814%2C2609.35904%2C2609.36201%2C2609.36488%2C2609.36923%2C2609.36927%2C2609.36939%2C2609.37053%2C2609.37176%2C2609.37686%2C2609.38008%2C2609.38397%2C2609.38778%2C2609.38867%2C2609.39306%2C2609.39547%2C2609.39903%2C2609.40253%2C2609.40284%2C2610.00948%2C2610.01215%2C2610.01491%2C2610.01787%2C2610.02320%2C2610.02932%2C2610.03036%2C2610.03089%2C2610.03574%2C2610.03818%2C2610.04116%2C2610.04800%2C2610.04906%2C2610.05185%2C2610.05437%2C2610.05861%2C2610.06829%2C2610.07118%2C2610.07127%2C2610.07444%2C2610.07972%2C2610.08699%2C2610.08773%2C2610.09240%2C2610.09469%2C2610.11050&max_results=50"}], "id": "arxiv:2610.05437", "identifiers": {"arxiv": "2610.05437"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:20Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2610.05437"}], "name": "TeleTune: Evolving Agent Skills From Offline Telemetry", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-10-04", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces TeleTune, which edits a textual skill library using held-out action prediction on offline software-use logs with missing goals or interleaved tasks. The learned workflows also guide retrieval of demonstrations for subsequent web-agent execution.", "topics": ["memory", "workflow", "efficiency"], "updated_at": "2026-10-06", "url": "https://arxiv.org/abs/2610.05437"}
{"aliases": [], "authors": ["Chinmay Savadikar", "Zhaoyu Zhang", "Mingyu Zhao", "Shuang Xie", "Han Li", "Tianfu Wu", "Lingyun Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "4890856d5961395a9ac8c83c60b988da4adeaab1d99edd511e4046e4f99fd25d", "submitted_at": "2026-10-05T17:11:07Z", "updated_at": "2026-10-05T17:11:07Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.07118v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2610.07118", "identifiers": {"arxiv": "2610.07118"}, "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/2610.07118"}], "name": "AMBER: Training Long-Horizon Web Agents through Append-Only Memory", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-10-05", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces AMBER, an append-only memory framework in which a web agent learns to reason, act, and retain evidence using outcome-reward reinforcement learning. It studies whether preserving entries avoids the loss of facts and corrective feedback seen with rewritten memory.", "topics": ["memory", "long-horizon", "reinforcement-learning"], "updated_at": "2026-10-05", "url": "https://arxiv.org/abs/2610.07118"}
{"aliases": [], "authors": ["Yudong Bai", "Yihong Chen", "Quanming Yao", "Yaqing Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2627960a5698d4c2f245e91ca5033be8c09e8a1ddea02490fcc1ed9f3d42217d", "submitted_at": "2026-10-08T11:12:57Z", "updated_at": "2026-10-08T11:12:57Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.11707v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2610.11707", "identifiers": {"arxiv": "2610.11707"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:29Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2610.11707"}], "name": "DeltaReplay: Task-Relative Memory Reuse for Mobile GUI Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-10-08", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces DeltaReplay, which stores mobile GUI trajectories as transition graphs and decides at each step whether to reuse an action, substitute task-specific parameters, or defer to the base agent. This supports reuse of partially matching experiences without rewriting stored memory.", "topics": ["memory", "mobile-navigation", "planning"], "updated_at": "2026-10-08", "url": "https://arxiv.org/abs/2610.11707"}
{"aliases": [], "authors": ["Sunghwan Kim", "Deogyong Kim", "Sangam Lee", "Donguk Kwon", "Dongha Lee"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "cadcf2ab7be51b17d06975eba0e56573c333dca9ec034ade67b84e62d0096903", "submitted_at": "2026-10-09T11:23:40Z", "updated_at": "2026-10-09T11:23:40Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.13416v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:24Z", "role": "primary", "sha256": "9641755401e0e8ec9613f7c4509029d968ab4f10751ba4d43cab57e7d27c1da1", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2610.11655%2C2610.11707%2C2610.11942%2C2610.12611%2C2610.12823%2C2610.12866%2C2610.12984%2C2610.13268%2C2610.13416%2C2610.13703%2C2610.14157%2C2610.14228%2C2610.15138%2C2610.15765%2C2610.16234%2C2610.16319%2C2610.16720%2C2610.16778%2C2610.17457%2C2610.17710%2C2610.17792%2C2610.17895%2C2610.17928%2C2610.18508%2C2610.18784%2C2610.18868%2C2610.19100%2C2610.19128%2C2610.20584&max_results=29"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:24Z", "role": "primary", "sha256": "9641755401e0e8ec9613f7c4509029d968ab4f10751ba4d43cab57e7d27c1da1", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2610.11655%2C2610.11707%2C2610.11942%2C2610.12611%2C2610.12823%2C2610.12866%2C2610.12984%2C2610.13268%2C2610.13416%2C2610.13703%2C2610.14157%2C2610.14228%2C2610.15138%2C2610.15765%2C2610.16234%2C2610.16319%2C2610.16720%2C2610.16778%2C2610.17457%2C2610.17710%2C2610.17792%2C2610.17895%2C2610.17928%2C2610.18508%2C2610.18784%2C2610.18868%2C2610.19100%2C2610.19128%2C2610.20584&max_results=29"}], "id": "arxiv:2610.13416", "identifiers": {"arxiv": "2610.13416"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:24Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2610.13416"}], "name": "Web$hell: A Unified Interface for Agentic Web Access", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-10-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces Web$hell, an interface that compiles live pages into persistent information and action objects so agents can gather evidence, navigate websites, and reuse acquired content through shared operations.", "topics": ["web-navigation", "tool-integration", "memory", "efficiency"], "updated_at": "2026-10-09", "url": "https://arxiv.org/abs/2610.13416"}
{"aliases": [], "authors": ["Taeyun Roh", "Suhyeong Park", "Gijin Kim", "Wonjune Jang", "Jaewoo Kang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "551f198c0976c75693c6298400800cc66ce1b1b5ee304965155a6bb3ca9d68fe", "submitted_at": "2026-10-13T08:15:52Z", "updated_at": "2026-10-13T08:15:52Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.17928v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T01:46:05Z", "role": "primary", "sha256": "3fb5aabfb07df620955f0b74347f9d5926a1cd3d506b1098eccba9477627269f", "supports": ["environments", "notes"], "url": "https://raw.githubusercontent.com/dmis-lab/SwitchMem/HEAD/README.md"}], "id": "arxiv:2610.17928", "identifiers": {"arxiv": "2610.17928"}, "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/2610.17928"}], "name": "Learning How to Remember: Memory Modality Selection for GUI Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced. Android platform tagging is additionally supported by the primary project documentation.", "published_at": "2026-10-13", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces SwitchMem, which chooses a textual or visual representation of retrieved experience at each GUI interaction step instead of fixing one memory modality for a task.", "topics": ["memory", "mobile-navigation"], "updated_at": "2026-10-13", "url": "https://arxiv.org/abs/2610.17928"}
{"aliases": [], "authors": ["Yuchen Lu", "Julian McAuley", "Lei Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6f4a9b6fae506d4dd35798c1930d5df4ec7fc7a2c452ab8bd5a5714cf385d505", "submitted_at": "2026-10-13T17:35:27Z", "updated_at": "2026-10-14T04:46:42Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2610.18784v2"}}, "environments": ["desktop", "web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:24Z", "role": "primary", "sha256": "9641755401e0e8ec9613f7c4509029d968ab4f10751ba4d43cab57e7d27c1da1", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2610.11655%2C2610.11707%2C2610.11942%2C2610.12611%2C2610.12823%2C2610.12866%2C2610.12984%2C2610.13268%2C2610.13416%2C2610.13703%2C2610.14157%2C2610.14228%2C2610.15138%2C2610.15765%2C2610.16234%2C2610.16319%2C2610.16720%2C2610.16778%2C2610.17457%2C2610.17710%2C2610.17792%2C2610.17895%2C2610.17928%2C2610.18508%2C2610.18784%2C2610.18868%2C2610.19100%2C2610.19128%2C2610.20584&max_results=29"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:24Z", "role": "primary", "sha256": "9641755401e0e8ec9613f7c4509029d968ab4f10751ba4d43cab57e7d27c1da1", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2610.11655%2C2610.11707%2C2610.11942%2C2610.12611%2C2610.12823%2C2610.12866%2C2610.12984%2C2610.13268%2C2610.13416%2C2610.13703%2C2610.14157%2C2610.14228%2C2610.15138%2C2610.15765%2C2610.16234%2C2610.16319%2C2610.16720%2C2610.16778%2C2610.17457%2C2610.17710%2C2610.17792%2C2610.17895%2C2610.17928%2C2610.18508%2C2610.18784%2C2610.18868%2C2610.19100%2C2610.19128%2C2610.20584&max_results=29"}], "id": "arxiv:2610.18784", "identifiers": {"arxiv": "2610.18784"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:24Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2610.18784"}], "name": "Pack Light, CarryOn: Building Object-Oriented Context for Long-Horizon Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-10-13", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "Introduces CarryOn, which maintains evidence-linked task objects and remaining-work context across history compression, with evaluations on desktop tasks and web research.", "topics": ["memory", "long-horizon", "efficiency"], "updated_at": "2026-10-14", "url": "https://arxiv.org/abs/2610.18784"}
{"aliases": [], "authors": ["Minghao Guo", "Qingyue Jiao", "Xiujin Liu", "Jianhong Pang", "Ruimeng Yang", "Naihao Xu", "Liwei Che", "Haochen Xue", "Ziqi Wang", "Xinying Cai", "Jiaxuan Luo", "Jiaxin Deng", "Xuesong Ye", "Jintao Qu", "Wenyao Gao", "Jingwen Gu", "Chenyang Zhao", "Shichun Liu", "Guibin Zhang", "Zefang Liu", "Siyin Wang", "Wujiang Xu", "Danrui Li", "Zeru Shi", "Woongyeong Yeo", "Shuang Liu", "Wei-Chieh Huang", "Weizhi Zhang", "Tianyu Yang", "Kangsan Kim", "Yebowen Hu", "Bowen Wen", "Bowen Jiang", "Haohao Su", "Ming Yin", "Sheldon Yu", "Mian Zhang", "Yuyan Chen", "Hanrong Zhang", "Kai Mei", "Zihao Zhang", "Mingyu Jin", "Xi Zhu", "Hengyi Wang", "Tianxin Wei", "Linyong Nan", "Zhenwen Liang", "Zhendong Chu", "Ziqi Huang", "Yongye Su", "Sixun Dong", "Shujian Liu", "Kaiqu Liang", "Jiaojiao Han", "Zhaoye Fei", "Kun Wang", "Shihan Dou", "Zhichun Guo", "Xingjian Diao", "Zeqi Gu", "Heng Wang", "Tao Feng", "Cheng Qian", "Hamid Palangi", "Silei Xu", "Xi Yun", "Yi Zhu", "Shilong Liu", "Qingsong Wen", "Sung Ju Hwang", "Zuxuan Wu", "Yu-Gang Jiang", "Yiyu Shi", "Dimitris N. Metaxas", "Jiaxuan You", "Julian McAuley", "Mengdi Wang", "Jiawei Han", "Philip S. Yu", "Ziyi Ye"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "7f1a48aa4b89fb97e8e1a186ba7345826a8b9efc9aa7b82e0dd1f5ce17d42b73", "submitted_at": "2026-10-14T01:11:57Z", "updated_at": "2026-10-14T01:11:57Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.19366v1"}, "license_review": {"document": {"evidence_urls": ["https://arxiv.org/html/2610.19366v1"], "reason": "Paper-specific document declaration only; no code, data or weights permissions are inferred."}}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T12:31:39Z", "role": "primary", "sha256": "c52e4e283e0dae88ef8b3ce67b7424b1d4b48e8f41015d872f4da69f98782eaa", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2609.39924%2C2610.00346%2C2610.00583%2C2610.01939%2C2610.05099%2C2610.09127%2C2610.09484%2C2610.09633%2C2610.11251%2C2610.11922%2C2610.13098%2C2610.13166%2C2610.13361%2C2610.13623%2C2610.16117%2C2610.17953%2C2610.18067%2C2610.19366%2C2610.20424&max_results=19"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T19:02:22.864948+00:00", "role": "primary", "sha256": "308796868bf9b1cb042bc6008e1e6f4ba6e776a2d4e13d40ad5684b477861563", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "licenses.document"], "url": "https://arxiv.org/html/2610.19366v1"}], "id": "arxiv:2610.19366", "identifiers": {"arxiv": "2610.19366"}, "kind": "survey", "last_verified_at": "2026-10-20T19:02:22.864948+00:00", "licenses": {"code": null, "data": null, "document": {"basis": "The retained exact-v1 arXiv HTML infobox explicitly declares License: CC BY 4.0 for this document.", "declared": "CC-BY-4.0", "retrieved_at": "2026-10-20T19:02:22.864948+00:00", "url": "https://arxiv.org/html/2610.19366v1"}, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2610.19366"}], "name": "Seeing, Maintaining, and Learning: The Evolution of Multimodal Agent Memory", "notes": "A broad literature synthesis spanning images, video, audio, GUI and embodied settings. Its GUI section compares interface-state retention, action outcomes, cross-session reuse and visual evidence, but the paper does not provide a new GUI-agent experiment. Taxonomy corpus counts and benchmark classifications are authors' annotations, not independently audited field coverage. Updating stored content alone does not establish the survey's self-evolving-policy tier; endpoint task success does not isolate memory use. Lifecycle, resource-budget and provenance recommendations are proposed evaluation directions rather than demonstrated guarantees. Platform facets stay unspecified; cited systems and releases were not independently verified.", "published_at": "2026-10-14", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "Surveys multimodal memory using representation and management axes, distinguishing fixed-policy updates from policy evolution and comparing GUI memory evaluation with other modalities.", "topics": ["survey", "memory", "evaluation", "interaction-history"], "updated_at": "2026-10-14", "url": "https://arxiv.org/abs/2610.19366"}
{"aliases": [], "authors": ["Tengda Guo", "Jiawang Hu", "Yifei Tao", "Yin Wang", "Haotian Hu", "Fanyi Wang", "Liujuan Tang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "025c7018c86ae4de79a659242070b2f91c5913333e9ecd4bd7fd357d13d27a34", "submitted_at": "2026-10-14T14:39:41Z", "updated_at": "2026-10-14T14:39:41Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.22408v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-20T12:58:20.217517Z", "role": "primary", "sha256": "f275c47a8b6d349a70e39a9e55eea5a09d99d5f35cc896100f3bcfd5b0923e0b", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28abs%3A%22computer-use+agents%22+OR+abs%3A%22GUI+agents%22+OR+abs%3A%22browser+agents%22+OR+abs%3A%22mobile+GUI%22%29+AND+submittedDate%3A%5B202609200000+TO+202610201255%5D&start=0&max_results=200&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T13:12:54.814500Z", "role": "primary", "sha256": "0389d96a8f6c4975129f5c526e8df9ca4690f31364a2f4e66aaea4a75f0eabbc", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/html/2610.22408v1"}], "id": "arxiv:2610.22408", "identifiers": {"arxiv": "2610.22408"}, "kind": "paper", "last_verified_at": "2026-10-20T13:12:54.814500Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2610.22408"}], "name": "ECHOES: Seeing Beyond the Current Image via Recursive History Compression and Privileged On-Policy Distillation", "notes": "Privileged screenshot history is used during training, while deployment retains self-generated textual state. GUI results use action/step metrics and model-specific history settings; online Sokoban transfer is separate. GRPO rewards action correctness, valid output format and compression length. Diagnostic memory replay is distinct from the validation gap used to schedule training. No independent reproduction.", "published_at": "2026-10-14", "relations": [], "review_status": "curated", "scope": "core", "summary": "ECHOES learns recursive textual task memory through privileged screenshot-history distillation and GRPO, then predicts GUI actions from the current screenshot and memory.", "topics": ["memory", "interaction-history", "state-tracking", "long-horizon", "imitation-learning", "reinforcement-learning", "efficiency", "evaluation"], "updated_at": "2026-10-14", "url": "https://arxiv.org/abs/2610.22408"}
{"aliases": [], "authors": ["Adamo Orsini", "Seyed Mehdi Ayyoubzadeh", "Sadman Rakib Pinon", "Hannuo Zhang", "Koosha Gholipour Baradari", "Yuanhao Yu", "Xinxin Zuo", "Yang Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "caaf24ef4d8b86c8aaebc852235d0fe947a5ed859dc1ba8329ce07a2f7be9aa1", "submitted_at": "2026-10-15T21:40:34Z", "updated_at": "2026-10-15T21:40:34Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.22672v1"}}, "environments": ["android"], "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": "2026-10-20T12:56:56.464343+00:00", "role": "primary", "sha256": "5fd9d7926e99d068fa8fb7238a9e7811e3f8c55d00e6b89d5672754cb3e619bc", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/html/2610.22672v1"}], "id": "arxiv:2610.22672", "identifiers": {"arxiv": "2610.22672"}, "kind": "paper", "last_verified_at": "2026-10-20T12:56:56.464343+00:00", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2610.22672"}], "name": "Not All Graphs Are Maps: A Task-Agnostic Exploration Framework for Mobile GUI Navigation", "notes": "The controlled cohort has 84 tasks across six apps; the broader inventory spans 20 apps. Main results use human destination labels. DroidTask and SPA-Bench are converted to navigation-only protocols. Some comparison cells received another graph condition’s memory, with sensitivity analyses reported. Cross-family runs differ in app versions, state and denominators. A small stale-graph probe does not establish robustness. Models, code, artifacts and performance were not independently reproduced.", "published_at": "2026-10-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "TAG-Nav supplies Android GUI agents with destination and route memory distilled from offline app exploration. DestBench measures arrival at a reference pre-action screen, separately from downstream task execution.", "topics": ["mobile-navigation", "memory", "evaluation"], "updated_at": "2026-10-15", "url": "https://arxiv.org/abs/2610.22672"}
{"aliases": [], "authors": ["Zhicheng Zheng", "Cheng Zhang", "Shengping Yu", "Tianxiang Zhao"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "7dc2ceb9aadbf7506efc068df99326fea5b60fdc6feb340c490841b88aa19c30", "submitted_at": "2026-10-19T14:38:41Z", "updated_at": "2026-10-19T14:38:41Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.26503v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-20T05:52:00Z", "role": "primary", "sha256": "af310824eea97aac8209023ff53bad7bea77778816586b10bb94110148347af9", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agent%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B202610060000+TO+202610200000%5D&start=0&max_results=200&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T08:27:26.236797+00:00", "role": "primary", "sha256": "80bed80786f939973d0d13b4f892300fee02a6110e326247131fc83d257796de", "supports": ["name", "kind", "summary", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agent%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22+OR+ti%3A%22browser+agents%22+OR+ti%3A%22mobile+agent%22+OR+ti%3A%22mobile+agents%22+OR+ti%3A%22desktop+agent%22%29+AND+submittedDate%3A%5B202610130000+TO+202610192359%5D&start=0&max_results=200&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T08:29:02.225995+00:00", "role": "primary", "sha256": "0554aa531a5f94f51cba6c51bd1b13fb05b0f462497afa5d1dae7ec6fc6e375a", "supports": ["summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/html/2610.26503v1"}], "id": "arxiv:2610.26503", "identifiers": {"arxiv": "2610.26503"}, "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/2610.26503"}], "name": "ScopeMem: Latent Memory That Learns Where to Revise Frozen GUI Agents", "notes": "Reviewed the complete abstract and selected v1 controlled-study, method and limitations sections. Results are offline, step-level action matching; online task success is explicitly left untested. The frozen actor is queried twice per step. No independent reproduction or redistribution-license claim.", "published_at": "2026-10-19", "relations": [], "review_status": "curated", "scope": "core", "summary": "ScopeMem conditions latent-memory prefixes on a frozen GUI agent's initial action and trains them to correct errors while preserving correct decisions. Its evaluation separates action repairs from harmful revisions.", "topics": ["memory", "interaction-history", "imitation-learning", "robustness", "evaluation"], "updated_at": "2026-10-19", "url": "https://arxiv.org/abs/2610.26503"}
{"aliases": [], "authors": [], "details": {"component_path": "Mobile-Agent-E", "license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/LICENSE"], "reason": "Reviewed repository license declaration for code; third-party dependencies, data, weights and executable environment images are separate."}}, "repository_tree_sha": "11cea575561fb7800b5fb6b6cafa56f7a91de11f", "repository_url": "https://github.com/X-PLUG/MobileAgent"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:27:15Z", "role": "primary", "sha256": "f8515fbce73ea0ef3d1d558d4567bb949244e6384e61c956ba3ddbc27294f0ae", "supports": ["name", "kind", "environments", "scope", "notes", "relations"], "url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/Mobile-Agent-E/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T15:37:03Z", "role": "primary", "sha256": "d403251020124bebcd9f180f800c8ea9f18a3f117c82fcf04c3cbe9ffe0e35f9", "supports": ["identity"], "url": "https://api.github.com/repos/X-PLUG/MobileAgent"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:25:21Z", "role": "primary", "sha256": "7293efadde7c96b64ff8483b2157241e17d0b0fbfff4839de9c7b11485b07cbb", "supports": ["identity", "release_status", "details.repository_tree_sha", "relations"], "url": "https://api.github.com/repos/X-PLUG/MobileAgent/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["summary", "topics"], "url": "https://export.arxiv.org/api/query?id_list=2412.04454%2C2412.04626%2C2412.05467%2C2412.09362%2C2412.09605%2C2412.10342%2C2412.10742%2C2412.10840%2C2412.13103%2C2412.13194%2C2412.13501%2C2412.14161%2C2412.14470%2C2412.16256%2C2412.17589%2C2412.18116%2C2412.18426%2C2412.19723%2C2501.01149%2C2501.04575%2C2501.07572%2C2501.11733%2C2501.12326%2C2501.13896%2C2501.16150%2C2502.01600%2C2502.01977%2C2502.02982%2C2502.06395%2C2502.07949%2C2502.08047%2C2502.08226%2C2502.11357%2C2502.13053%2C2502.13923%2C2502.14282%2C2502.18906%2C2503.00401%2C2503.02268%2C2503.03196%2C2503.04730%2C2503.05143%2C2503.06470%2C2503.09263%2C2503.11069%2C2503.11170%2C2503.12687%2C2503.15661%2C2503.15937%2C2503.16465&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:42:03Z", "role": "primary", "sha256": "961819a89dacd1f4f161e1d2ebf616b538bf095a2eecfa3ef40b7ccec633fac4", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/LICENSE"}], "id": "component:x-plug/mobileagent/mobile-agent-e", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-18T04:27:15Z", "licenses": {"code": {"basis": "Reviewed repository license declaration for code; third-party dependencies, data, weights and executable environment images are separate.", "declared": "MIT", "retrieved_at": "2026-10-16T06:42:03Z", "url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/X-PLUG/MobileAgent/tree/main/Mobile-Agent-E"}], "name": "Mobile-Agent-E code component", "notes": "The inspected README explicitly supports Android only, with experiments on Samsung Galaxy A15; host-machine macOS/Linux instructions do not imply control of those operating systems. Agent code, launch scripts and benchmark definitions are present. The root MIT code license does not establish terms for the separately published task dataset or external models.", "published_at": null, "relations": [{"evidence_url": "https://api.github.com/repos/X-PLUG/MobileAgent/git/trees/main?recursive=1", "target_id": "github:x-plug/mobileagent", "type": "component_of"}, {"evidence_url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/Mobile-Agent-E/README.md", "target_id": "arxiv:2501.11733", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Android automation component with hierarchical planning, reflection and persistent tips/shortcuts learned across task sequences.", "topics": ["planning", "memory", "self-reflection", "mobile-navigation"], "updated_at": null, "url": "https://github.com/X-PLUG/MobileAgent/tree/main/Mobile-Agent-E"}
{"aliases": [], "authors": [], "details": {}, "environments": ["desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:02Z", "role": "primary", "sha256": "fe475d3e92c42807b792c542bc5986bf74caa7d166aea860d368fbe776342a82", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/BAAI-Agents/Cradle/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T06:42:04Z", "role": "primary", "sha256": "2964861ad96d977da95d891f62d8f5d60873f4a36da0b51f052d9e98ef97d02f", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/BAAI-Agents/Cradle/HEAD/LICENSE"}], "id": "github:baai-agents/cradle", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-16T06:36:02Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "MIT", "retrieved_at": "2026-10-16T06:42:04Z", "url": "https://raw.githubusercontent.com/BAAI-Agents/Cradle/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/BAAI-Agents/Cradle"}], "name": "Cradle", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/BAAI-Agents/Cradle/HEAD/README.md", "target_id": "arxiv:2403.03186", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Framework in which foundation models control games and software from screenshots alone, acting through keyboard and mouse, shown on RDR2, Stardew Valley, Cities: Skylines and Dealer's Life 2.", "topics": ["planning", "memory"], "updated_at": null, "url": "https://github.com/BAAI-Agents/Cradle"}
{"aliases": [], "authors": [], "details": {"availability": "Public inference/scoring source; benchmark data access remains gated.", "evaluation_caution": "Missing screenshots are marked AUTO_PASS and credited as correct according to the README.", "license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/CVC2233/AndroTMem/HEAD/LICENSE"], "reason": "Repository code license only; no data-license or gated-access permission is inferred."}}, "repository_revision": "0c2da7002baa5b0ee71f80dd6fbd90e33071e259"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:18:21Z", "role": "primary", "sha256": "6681ce1bd98c856fed4193e239e234a69eb2a74fa937e9c38db89f48db40f452", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations"], "url": "https://raw.githubusercontent.com/CVC2233/AndroTMem/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:18:20Z", "role": "primary", "sha256": "f2b7c762383746a56e62273f756ca7ed3223726535763a41fda3ac5e3467aa82", "supports": ["identity"], "url": "https://api.github.com/repos/CVC2233/AndroTMem"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:22:59Z", "role": "primary", "sha256": "302c804dc65ebb3e8558651e0c7cd0a79c1caf29fc9fb11868d26fba36afdf48", "supports": ["release_status", "details.repository_revision"], "url": "https://api.github.com/repos/CVC2233/AndroTMem/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:23:00Z", "role": "primary", "sha256": "6d572ccef64746baac78441c0b26488f525c70c2b7453f985bd4f19cbec00634", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/CVC2233/AndroTMem/HEAD/LICENSE"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:23:07Z", "role": "primary", "sha256": "e82623b7dc889eb7bff45c33f15a47c8c5caac3de0a3c56a7a9975ca9b2d6694", "supports": ["notes", "release_status"], "url": "https://huggingface.co/api/datasets/CVC2233/AndroTMem-Bench"}], "id": "github:cvc2233/androtmem", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-17T11:18:21Z", "licenses": {"code": {"basis": "Repository code license only; no data-license or gated-access permission is inferred.", "declared": "MIT", "retrieved_at": "2026-10-17T11:23:00Z", "url": "https://raw.githubusercontent.com/CVC2233/AndroTMem/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/CVC2233/AndroTMem"}], "name": "AndroTMem inference and evaluation code", "notes": "The source tree releases inference backends and evaluation scripts under evaluation/. The separately hosted benchmark is gated=auto and its raw card returned 401; no access request was made and it is not approved here as a directly accessible dataset. The README explicitly credits missing screenshots as AUTO_PASS/correct, so incomplete local image inventories can inflate scores. Its TCR is an offline anchor/dependency score and is not presented here as independently verified live task success.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/CVC2233/AndroTMem/HEAD/README.md", "target_id": "arxiv:2603.18429", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "A public Android GUI evaluation framework with action-history, summary and anchored-memory prompt modes, plus step-action and dependency-aware task scoring.", "topics": ["memory", "long-horizon", "evaluation"], "updated_at": null, "url": "https://github.com/CVC2233/AndroTMem"}
{"aliases": [], "authors": [], "details": {}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T01:46:05Z", "role": "primary", "sha256": "3fb5aabfb07df620955f0b74347f9d5926a1cd3d506b1098eccba9477627269f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/dmis-lab/SwitchMem/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T01:48:10Z", "role": "primary", "sha256": "cec8e4679c883f13983915e9f77d175a84a234d69386e78f9ea75157cefbdcb4", "supports": ["relations"], "url": "https://txxnrd.github.io/switchmem/"}], "id": "github:dmis-lab/switchmem", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-17T01:46:05Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/dmis-lab/SwitchMem"}], "name": "SwitchMem", "notes": "The implementation distinguishes step-level AITW/LearnGUI scores from terminal task success on MobileMiniWoB++. Benchmark outcomes and local smoke-test outputs are different claims; neither was independently reproduced here.", "published_at": null, "relations": [{"evidence_url": "https://txxnrd.github.io/switchmem/", "target_id": "arxiv:2610.17928", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Memory-routing implementation that chooses text or image demonstrations for a frozen GUI actor, with offline evaluation and Android transfer experiments.", "topics": ["memory", "mobile-navigation"], "updated_at": null, "url": "https://github.com/dmis-lab/SwitchMem"}
{"aliases": ["github:mobilegptsys/mobilegpt", "mobilegptsys/MobileGPT", "https://github.com/mobilegptsys/MobileGPT"], "authors": [], "details": {"availability": "Public Android client, Python server and manual offline-explorer source.", "license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/fclab-skku/MobileGPT/HEAD/LICENSE"], "reason": "Repository source license; linked benchmark data and external models retain separate terms."}}, "repository_revision": "7945048e6c05849aff1633744cbdf653c58bece2"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:32:07Z", "role": "primary", "sha256": "8683cff296d21c1a7a38bc372efaf004b8af644030fac77a4dc9d2b2e21d894b", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://raw.githubusercontent.com/fclab-skku/MobileGPT/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:30:55Z", "role": "primary", "sha256": "3d3ceb084d374b706842cff1f4ec986fde50c13cba141dc4619d5afe8b631d55", "supports": ["identity"], "url": "https://api.github.com/repos/mobilegptsys/MobileGPT"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:32:06Z", "role": "primary", "sha256": "2a8f30b3422a033b345c2054d55adff0836b7dac163e600583ba0db1b4423320", "supports": ["release_status", "details.repository_revision"], "url": "https://api.github.com/repos/fclab-skku/MobileGPT/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:24Z", "role": "primary", "sha256": "3aac6e9748769efff3c475ede0da192776daa03431d231b5f67e2aa50f8a3be9", "supports": ["relations"], "url": "https://export.arxiv.org/api/query?id_list=2308.03688%2C2308.15272%2C2309.08172%2C2309.11436%2C2310.04716%2C2310.04869%2C2310.05126%2C2310.06770%2C2310.08740%2C2310.10634%2C2310.11441%2C2310.16042%2C2310.18331%2C2311.01767%2C2311.07562%2C2311.09278%2C2311.12983%2C2311.18751%2C2312.03003%2C2312.06677%2C2312.08914%2C2312.11671%2C2312.13108%2C2312.13771%2C2312.15820%2C2312.16062%2C2401.01614%2C2401.04124%2C2401.07781%2C2401.10935%2C2401.13178%2C2401.13649%2C2401.13919%2C2401.14268%2C2401.16158%2C2402.01030%2C2402.04476%2C2402.04615%2C2402.05930%2C2402.06596%2C2402.07456%2C2402.07939%2C2402.07945%2C2402.08567%2C2402.11941%2C2402.14073%2C2402.15057%2C2402.16965%2C2402.17553%2C2403.02713&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:34:10Z", "role": "primary", "sha256": "c71d239df91726fc519c6eb72d318ec65820627232b2f796219e87dcf35d0ab4", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/fclab-skku/MobileGPT/HEAD/LICENSE"}, {"method": "github_repository_identity", "retrieved_at": "2026-10-17T14:14:04Z", "role": "primary", "sha256": "3d3ceb084d374b706842cff1f4ec986fde50c13cba141dc4619d5afe8b631d55", "supports": ["aliases"], "url": "https://api.github.com/repos/mobilegptsys/MobileGPT"}, {"method": "github_repository_identity", "retrieved_at": "2026-10-17T14:14:06Z", "role": "primary", "sha256": "3d3ceb084d374b706842cff1f4ec986fde50c13cba141dc4619d5afe8b631d55", "supports": ["aliases"], "url": "https://api.github.com/repos/fclab-skku/MobileGPT"}], "id": "github:fclab-skku/mobilegpt", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-17T12:32:07Z", "licenses": {"code": {"basis": "Repository source license; linked benchmark data and external models retain separate terms.", "declared": "Apache-2.0", "retrieved_at": "2026-10-17T12:34:10Z", "url": "https://raw.githubusercontent.com/fclab-skku/MobileGPT/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/fclab-skku/MobileGPT"}], "name": "MobileGPT", "notes": "The old mobilegptsys repository resolves to fclab-skku/MobileGPT. Public Android client and Python server sources require Android SDK 33+, Accessibility Service setup and external service credentials. The README explicitly omits the random explorer and human-in-the-loop memory repair from the open-source version. Its benchmark description differs from the retained arXiv snapshot, so no dataset counts or reproduction claim are merged into this code record. External benchmark data was not approved as a separate release.", "published_at": null, "relations": [{"evidence_url": "https://export.arxiv.org/api/query?id_list=2308.03688%2C2308.15272%2C2309.08172%2C2309.11436%2C2310.04716%2C2310.04869%2C2310.05126%2C2310.06770%2C2310.08740%2C2310.10634%2C2310.11441%2C2310.16042%2C2310.18331%2C2311.01767%2C2311.07562%2C2311.09278%2C2311.12983%2C2311.18751%2C2312.03003%2C2312.06677%2C2312.08914%2C2312.11671%2C2312.13108%2C2312.13771%2C2312.15820%2C2312.16062%2C2401.01614%2C2401.04124%2C2401.07781%2C2401.10935%2C2401.13178%2C2401.13649%2C2401.13919%2C2401.14268%2C2401.16158%2C2402.01030%2C2402.04476%2C2402.04615%2C2402.05930%2C2402.06596%2C2402.07456%2C2402.07939%2C2402.07945%2C2402.08567%2C2402.11941%2C2402.14073%2C2402.15057%2C2402.16965%2C2402.17553%2C2403.02713&max_results=50", "target_id": "arxiv:2312.03003", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "An Android client and Python-server agent that learns reusable app subtasks through exploration, selection, derivation and memory recall.", "topics": ["mobile-navigation", "memory", "planning", "workflow"], "updated_at": null, "url": "https://github.com/fclab-skku/MobileGPT"}
{"aliases": [], "authors": [], "details": {"license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/kwai/MemGUI-Agent/main/LICENSE", "https://raw.githubusercontent.com/kwai/MemGUI-Agent/main/README.md"], "reason": "The README declares Apache-2.0, but the inspected LICENSE is a short notice and link rather than the complete terms; the code license remains unconfirmed."}}, "repository_id": 1278022559}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-20T03:54:57.854138+00:00", "role": "primary", "sha256": "3f7545f466407ed5336f47d18b12e8354ec352ff83dcf973b1545f81d9a629ac", "supports": ["name", "kind", "summary", "topics", "scope", "relations", "notes", "licenses.code"], "url": "https://raw.githubusercontent.com/kwai/MemGUI-Agent/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T03:55:26.237759+00:00", "role": "primary", "sha256": "4c92c584b6885a3bde8ee89d04ccf48287e7f6119889f75cde0c034c90bd8217", "supports": ["summary", "environments", "topics", "scope", "notes"], "url": "https://raw.githubusercontent.com/kwai/MemGUI-Agent/main/evaluation/memgui3k_offline_eval/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T03:55:01.061844+00:00", "role": "primary", "sha256": "d4e344274155be1d0a192c8ae2940169abb0bd8546d045b96b45810593860636", "supports": ["licenses.code", "notes"], "url": "https://raw.githubusercontent.com/kwai/MemGUI-Agent/main/LICENSE"}, {"method": "github_repository_identity", "retrieved_at": "2026-10-19T19:10:26.954787+00:00", "role": "primary", "sha256": "0ae4477b885b0915e43874399c2a6b91f7d07954697370204550df931f7db82c", "supports": ["details.repository_id"], "url": "https://api.github.com/search/repositories?q=topic%3Agui-agent+is%3Apublic+fork%3Afalse+created%3A2008-01-01..2026-10-18&per_page=100&page=1&sort=updated&order=asc"}], "id": "github:kwai/memgui-agent", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-20T03:54:57.854138+00:00", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/kwai/MemGUI-Agent"}], "name": "MemGUI-Agent", "notes": "The inspected release documentation describes dataset utilities, a LoRA SFT template and gold-step offline evaluation; live Android rollouts are explicitly outside that evaluator and linked to separate benchmark repositories. Executable source files, external datasets/checkpoints and reported results were not independently verified in this bounded review.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/kwai/MemGUI-Agent/main/README.md", "target_id": "arxiv:2606.19926", "type": "introduced_by"}], "review_status": "curated", "scope": "core", "summary": "MemGUI-Agent training and offline evaluation tools provide trajectory conversion, LoRA SFT templates, and step-level checks of GUI actions, memory updates and context folding.", "topics": ["memory", "interaction-history", "long-horizon", "imitation-learning", "evaluation"], "updated_at": null, "url": "https://github.com/kwai/MemGUI-Agent"}
{"aliases": [], "authors": [], "details": {"benchmark_protocol": {"actions": ["touch actions", "MCP tools (optional)"], "counts": [{"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/ljang0/iOSWorld/HEAD/README.md", "population": "tasks.json benchmark", "qualifier": "exact", "unit": "tasks", "value": 133}, {"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/ljang0/iOSWorld/HEAD/README.md", "population": "bundled SwiftUI apps", "qualifier": "exact", "unit": "applications", "value": 26}], "evaluation": "Rubric satisfaction per task; full-task pass requires every criterion to be satisfied.", "evidence_urls": ["https://raw.githubusercontent.com/ljang0/iOSWorld/HEAD/README.md"], "focus": ["personalized phone tasks", "cross-app workflows"], "notes": "Vision-only, vision-plus-XML, and MCP modes expose different information and tools. The documented default evaluator is model-based; compare the same mode and judge.", "observations": ["screenshots", "accessibility XML (optional)"], "setting": "interactive"}, "evaluation_type": "rubric-based", "metrics": ["mean rubric score", "full-task pass rate"], "task_count": 133}, "environments": ["ios"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T10:42:12Z", "role": "primary", "sha256": "706f9d770b65a1238e56f74f7dd5089c7756cc88e0ac09d80c7b2b3085a09b27", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details.task_count", "details.metrics", "details.evaluation_type", "details.benchmark_protocol"], "url": "https://raw.githubusercontent.com/ljang0/iOSWorld/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T10:48:41Z", "role": "primary", "sha256": "fa9fe1e31ed8da21d0b51621d3a3669ced4f0e42353aaf4bd8be1615d200eaac", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/ljang0/iOSWorld/HEAD/LICENSE"}], "id": "github:ljang0/iosworld", "identifiers": {}, "kind": "benchmark", "last_verified_at": "2026-10-16T10:42:12Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "Apache-2.0", "retrieved_at": "2026-10-16T10:48:41Z", "url": "https://raw.githubusercontent.com/ljang0/iOSWorld/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/ljang0/iOSWorld"}], "name": "iOSWorld", "notes": "", "published_at": null, "relations": [], "review_status": "curated", "scope": "core", "summary": "Native iPhone-simulator benchmark with shared user data, application workflows, agent runners, and rubric-based scoring.", "topics": ["evaluation", "cross-app", "memory"], "updated_at": null, "url": "https://github.com/ljang0/iOSWorld"}
{"aliases": [], "authors": [], "details": {"availability": "Public memory, prompting, browser-experiment and fine-tuning source.", "license_conflict": "README MIT badge versus NTUITIVE agreement; the agreement contains inconsistent distribution provisions.", "license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/ltzheng/synapse/HEAD/README.md", "https://raw.githubusercontent.com/ltzheng/Synapse/HEAD/LICENSE"], "reason": "Unresolved conflict between README MIT badge and NTUITIVE non-commercial LICENSE, including inconsistent redistribution clauses. Keep permissions unknown rather than choose an interpretation."}}, "repository_revision": "08c3a252ff158822e50e9c109c1cd49dc2f6147d"}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:30:54Z", "role": "primary", "sha256": "ed5416d91665017fca4b354016b7c2c748acf32d7723483fecb5387a22a1fdff", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "licenses.code"], "url": "https://raw.githubusercontent.com/ltzheng/synapse/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:30:54Z", "role": "primary", "sha256": "79a238a6f486d2a8970baba199f0968d28f14d42922c86ce71753eb687e0e9c6", "supports": ["identity"], "url": "https://api.github.com/repos/ltzheng/synapse"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:32:05Z", "role": "primary", "sha256": "785a7163dac9f41b1d0ca97f8dd9311b6df6a09f3a5163438dd64db226a4ceaf", "supports": ["release_status", "details.repository_revision"], "url": "https://api.github.com/repos/ltzheng/Synapse/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:20Z", "role": "primary", "sha256": "cbe8bfcbce4170a94a6cf42de01fbe0a448e7b1d29fa75a0de877105a98ec08e", "supports": ["relations"], "url": "https://export.arxiv.org/api/query?id_list=1802.08802%2C1808.09132%2C2005.03776%2C2101.09465%2C2103.16057%2C2104.08560%2C2105.13231%2C2107.03374%2C2107.13731%2C2108.03353%2C2202.02312%2C2202.08137%2C2204.01691%2C2205.11029%2C2207.01206%2C2209.08199%2C2209.08655%2C2209.14927%2C2209.15099%2C2210.02663%2C2210.03629%2C2210.03945%2C2211.07615%2C2301.10165%2C2301.13280%2C2302.04761%2C2303.07280%2C2303.08774%2C2303.17491%2C2303.17580%2C2304.04370%2C2304.07061%2C2305.04091%2C2305.05662%2C2305.08144%2C2305.10601%2C2305.11854%2C2305.12487%2C2305.14257%2C2305.14325%2C2305.16504%2C2305.16653%2C2305.19308%2C2306.00245%2C2306.06070%2C2306.07863%2C2307.10088%2C2307.12856%2C2307.13854%2C2308.01990&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:34:09Z", "role": "primary", "sha256": "cc4bce2274289d2b924fcc3c185c68b8e7c539365c5f8c72be865883244088ec", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/ltzheng/Synapse/HEAD/LICENSE"}], "id": "github:ltzheng/synapse", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-17T12:30:54Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/ltzheng/Synapse"}], "name": "Synapse", "notes": "Training, memory-building and evaluation scripts for MiniWoB++ and Mind2Web are present. External benchmark data, model/API credentials and legacy dependencies are still required. License remains unknown: the README shows MIT, while LICENSE is an NTUITIVE non-commercial agreement that both permits some distribution and separately says distribution is prohibited. No interpretation resolves those contradictions, and no commercial or non-commercial permission is asserted. This admission confirms the released GUI implementation, not a licensing grant.", "published_at": null, "relations": [{"evidence_url": "https://export.arxiv.org/api/query?id_list=1802.08802%2C1808.09132%2C2005.03776%2C2101.09465%2C2103.16057%2C2104.08560%2C2105.13231%2C2107.03374%2C2107.13731%2C2108.03353%2C2202.02312%2C2202.08137%2C2204.01691%2C2205.11029%2C2207.01206%2C2209.08199%2C2209.08655%2C2209.14927%2C2209.15099%2C2210.02663%2C2210.03629%2C2210.03945%2C2211.07615%2C2301.10165%2C2301.13280%2C2302.04761%2C2303.07280%2C2303.08774%2C2303.17491%2C2303.17580%2C2304.04370%2C2304.07061%2C2305.04091%2C2305.05662%2C2305.08144%2C2305.10601%2C2305.11854%2C2305.12487%2C2305.14257%2C2305.14325%2C2305.16504%2C2305.16653%2C2305.19308%2C2306.00245%2C2306.06070%2C2306.07863%2C2307.10088%2C2307.12856%2C2307.13854%2C2308.01990&max_results=50", "target_id": "arxiv:2306.07863", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "A computer-control research implementation that retrieves example trajectories from memory, abstracts task-relevant state and prompts an agent to navigate browser tasks.", "topics": ["memory", "planning", "web-navigation", "imitation-learning"], "updated_at": null, "url": "https://github.com/ltzheng/Synapse"}
{"aliases": [], "authors": [], "details": {"availability": "Public conversational planning, candidate-generation and training/evaluation source.", "license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/magicgh/Self-MAP/HEAD/LICENSE"], "reason": "Project source LICENSE; separate dataset license is not folded into the code slot."}}, "repository_revision": "6cf779682b70dae597626f96dfa5130b202689bc"}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:30:55Z", "role": "primary", "sha256": "693197cb807e3ab4d5afda9de35dd51342b4feb18e68b8ccfcdaa370399dbeb3", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations"], "url": "https://raw.githubusercontent.com/magicgh/self-map/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:30:54Z", "role": "primary", "sha256": "82c425841c94bf129c4fbd1e70029c581aa1488d5f85cdd23aa0f292092146f0", "supports": ["identity"], "url": "https://api.github.com/repos/magicgh/self-map"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:32:05Z", "role": "primary", "sha256": "8d844f227c457321640231ab3a4953457e3c6cfeab8d8f4803f8aa01dc15a8f8", "supports": ["release_status", "details.repository_revision"], "url": "https://api.github.com/repos/magicgh/Self-MAP/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:34:10Z", "role": "primary", "sha256": "f271c479b19a620600d18fc2265224205bf71483b969672d3f6acd9bc552b3f5", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/magicgh/Self-MAP/HEAD/LICENSE"}], "id": "github:magicgh/self-map", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-17T12:30:55Z", "licenses": {"code": {"basis": "Project source LICENSE; separate dataset license is not folded into the code slot.", "declared": "MIT", "retrieved_at": "2026-10-17T12:34:10Z", "url": "https://raw.githubusercontent.com/magicgh/Self-MAP/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/magicgh/Self-MAP"}], "name": "Self-MAP", "notes": "The source tree provides candidate-generation and action-prediction training/evaluation scripts. Its documented workflow requires separately downloaded MT-Mind2Web data and configured model/API paths. The README cites the ACL publication by DOI; a retained original arXiv abstract explicitly names Self-MAP and MT-Mind2Web and supplies the catalog paper relation without a new bibliography entity. MIT is the code license; the separate data artifact declares CC-BY-4.0.", "published_at": null, "relations": [{"evidence_url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending", "target_id": "arxiv:2402.15057", "type": "documented_by"}, {"evidence_url": "https://raw.githubusercontent.com/magicgh/self-map/HEAD/README.md", "target_id": "url:bdfb50330deb97893ce4", "type": "uses"}], "review_status": "curated", "scope": "core", "summary": "A conversational web-agent implementation with candidate ranking and memory-assisted action planning, including optional memory simplification, refinement and self-reflection modules.", "topics": ["web-navigation", "memory", "self-reflection", "planning"], "updated_at": null, "url": "https://github.com/magicgh/Self-MAP"}
{"aliases": [], "authors": [], "details": {"license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/MobileLLM/AgentProg/HEAD/LICENSE"], "reason": "Reviewed repository license declaration for code; third-party dependencies, data, weights and executable environment images are separate."}}, "repository_tree_sha": "ef5e2998ac2defa86e0dc4c13e7e00d595e1e7cf", "repository_url": "https://github.com/MobileLLM/AgentProg"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:27:13Z", "role": "primary", "sha256": "becc4961d93f0170fab29120afa4b09b2999f2cd0e74b13ccb7f28cb9623761b", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations"], "url": "https://raw.githubusercontent.com/MobileLLM/AgentProg/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T15:37:01Z", "role": "primary", "sha256": "d14eecf7c041933cefecc03b62bbea45c1617a183fd8c4ec1334e22a644a7f67", "supports": ["identity"], "url": "https://api.github.com/repos/MobileLLM/AgentProg"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:25:22Z", "role": "primary", "sha256": "dd1cf9cd0aed31af63d7046ffa98d1be2c36258706951173ad8c28a64fcf9f7a", "supports": ["identity", "release_status", "details.repository_tree_sha"], "url": "https://api.github.com/repos/MobileLLM/AgentProg/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:35:19Z", "role": "primary", "sha256": "751123fccb2fd04b5c08b8c712a68a8bc69f56391593711efbfa533eaebbdb10", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/MobileLLM/AgentProg/HEAD/LICENSE"}], "id": "github:mobilellm/agentprog", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-18T04:27:13Z", "licenses": {"code": {"basis": "Reviewed repository license declaration for code; third-party dependencies, data, weights and executable environment images are separate.", "declared": "Apache-2.0", "retrieved_at": "2026-10-18T04:35:19Z", "url": "https://raw.githubusercontent.com/MobileLLM/AgentProg/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/MobileLLM/AgentProg"}], "name": "AgentProg", "notes": "Public tree contains the agentprog package, CLI, program/context-management code and AndroidWorld evaluation integration. README announces AW-Extend, but this record is the framework rather than a separately audited benchmark release. External model services and Android device/emulator setup are required. Apache-2.0 covers code, not external model weights, apps or VM images.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/MobileLLM/AgentProg/HEAD/README.md", "target_id": "arxiv:2512.10371", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Android GUI-agent framework that keeps long-running task context in structured programs, variables and control flow, with a CLI and AndroidWorld integration.", "topics": ["long-horizon", "memory", "planning", "state-tracking"], "updated_at": null, "url": "https://github.com/MobileLLM/AgentProg"}
{"aliases": [], "authors": [], "details": {"availability": "Public browser-agent source and npm release metadata.", "license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/omxyz/lumen/HEAD/LICENSE"], "reason": "Publisher project-source license; external model and browser-service terms are separate."}}, "package_version": "0.2.0", "repository_revision": "b1ad26a0784645ac3a97d402db99cd5d17f86334"}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:30:55Z", "role": "primary", "sha256": "7708f434142498959bb5ed5de35ac1b945f9ad6e070e436b4e73819b7db1df7f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://raw.githubusercontent.com/omxyz/lumen/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:30:55Z", "role": "primary", "sha256": "32b02ea43d19929d82e8b8c52bb8e448553aebdd96a8754d144927cb0f34c5b6", "supports": ["identity"], "url": "https://api.github.com/repos/omxyz/lumen"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:32:05Z", "role": "primary", "sha256": "9911953e501a245d657f3835b22c6822ad7d990b96f03e3cba93a15ebe4f70a6", "supports": ["release_status", "details.repository_revision"], "url": "https://api.github.com/repos/omxyz/lumen/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:34:10Z", "role": "primary", "sha256": "34a03b6dd8cd1960d5592a6569e2a1c1f6c1b2697da1386a58d0ca64157f059e", "supports": ["identity", "release_status", "details.package_version", "notes"], "url": "https://registry.npmjs.org/@omxyz%2Flumen/latest"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:34:10Z", "role": "primary", "sha256": "30dd5336a8a45b548b9051e751d45923a9864b3810a36c836df9b07447ca3c5c", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/omxyz/lumen/HEAD/LICENSE"}], "id": "github:omxyz/lumen", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-17T12:30:55Z", "licenses": {"code": {"basis": "Publisher project-source license; external model and browser-service terms are separate.", "declared": "MIT", "retrieved_at": "2026-10-17T12:34:10Z", "url": "https://raw.githubusercontent.com/omxyz/lumen/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/omxyz/lumen"}], "name": "Lumen browser agent", "notes": "Public TypeScript source and npm package metadata are available; the inspected package version is 0.2.0. The package declares Node ^20.19.0 or >=22.12.0 and requires a browser and model provider. Its README benchmark is explicitly preliminary and uses a small WebVoyager subset; this catalog makes no comparative performance or independent safety claim. Referenced research influences are not asserted as papers introducing Lumen.", "published_at": null, "relations": [], "review_status": "curated", "scope": "core", "summary": "A screenshot-driven browser agent over Chrome DevTools Protocol, with resumable sessions, context compression, action replay and configurable completion checks.", "topics": ["web-navigation", "memory", "automation", "observability"], "updated_at": null, "url": "https://github.com/omxyz/lumen"}
{"aliases": [], "authors": [], "details": {"availability": "Public Android executor, agents, prompts and setup documentation.", "license_review": {"code": {"evidence_urls": ["https://api.github.com/repos/OpenDFM/MobA/git/trees/main?recursive=1"], "reason": "No repository source-license declaration established from the current public tree."}}, "repository_revision": "05cd237d80fcd23ad0dd6096cdb144117ecb3471"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:30:53Z", "role": "primary", "sha256": "49c096e5ee6a87205497d693b5248817ea98cad0f7e5254793660007ef4292f7", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations"], "url": "https://raw.githubusercontent.com/OpenDFM/MobA/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:30:53Z", "role": "primary", "sha256": "0ecd1833a1b6840441769e1cc26a55268f2de7beb61bc3e783a851dd7cff45f1", "supports": ["identity"], "url": "https://api.github.com/repos/OpenDFM/MobA"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:32:04Z", "role": "primary", "sha256": "302f324fce2884c5948cf62c75f2f839cd40c301faff6ccf4b6eb5dd29f953a3", "supports": ["release_status", "details.repository_revision", "licenses.code"], "url": "https://api.github.com/repos/OpenDFM/MobA/git/trees/main?recursive=1"}], "id": "github:opendfm/moba", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-17T12:30:53Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/OpenDFM/MobA"}], "name": "MobA", "notes": "The repository releases a configurable Python executor and Android ADB integration; its setup documents a Windows host, but the controlled interface is Android. The source is marked under continuing development and requires external model credentials. The inspected repository has no license file; the separately hosted MobBench data declaration is not transferred to this code.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/OpenDFM/MobA/HEAD/README.md", "target_id": "arxiv:2410.13757", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "An Android agent that combines global task planning and memory with a local action executor, using reflection to revise mobile interaction steps.", "topics": ["mobile-navigation", "memory", "planning", "self-reflection"], "updated_at": null, "url": "https://github.com/OpenDFM/MobA"}
{"aliases": [], "authors": [], "details": {}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T13:14:29Z", "role": "primary", "sha256": "75f99d9388aee3e0b124ebdb5f33e08fbe8ec999ce1e9eba5cd71c21e60522e0", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/PathOnAI/LiteWebAgent/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T13:29:25Z", "role": "primary", "sha256": "ec6ef142e3900e9f5555009b825387e41da67a77a24ee72ef6963c99b065e6eb", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/PathOnAIOrg/LiteWebAgent/HEAD/LICENSE"}], "id": "github:pathonaiorg/litewebagent", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-16T13:14:29Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "Apache-2.0", "retrieved_at": "2026-10-16T13:29:25Z", "url": "https://raw.githubusercontent.com/PathOnAIOrg/LiteWebAgent/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/PathOnAIOrg/LiteWebAgent"}], "name": "LiteWebAgent", "notes": "The repository is maintained by the PathOnAI.org research community; its README explicitly disclaims affiliation with a for-profit company.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/PathOnAI/LiteWebAgent/HEAD/README.md", "target_id": "arxiv:2503.02950", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Modular web-agent framework with browser control, planning, workflow memory, and tree-search components, plus web and extension interfaces.", "topics": ["web-navigation", "planning", "memory", "tool-integration"], "updated_at": null, "url": "https://github.com/PathOnAIOrg/LiteWebAgent"}
{"aliases": [], "authors": [], "details": {}, "environments": ["desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:02Z", "role": "primary", "sha256": "9ea381fdb78eb326d988ea384b6f5074ef0ff4c0d5fab555a148eded385e88f1", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/simular-ai/Agent-S/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T06:42:03Z", "role": "primary", "sha256": "43070e2d4e532684de521b885f385d0841030efa2b1a20bafb76133a5e1379c1", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/simular-ai/Agent-S/HEAD/LICENSE"}], "id": "github:simular-ai/agent-s", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-16T06:36:02Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "Apache-2.0", "retrieved_at": "2026-10-16T06:42:03Z", "url": "https://raw.githubusercontent.com/simular-ai/Agent-S/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/simular-ai/Agent-S"}], "name": "Agent S", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/simular-ai/Agent-S/HEAD/README.md", "target_id": "arxiv:2410.08164", "type": "documented_by"}, {"evidence_url": "https://raw.githubusercontent.com/simular-ai/Agent-S/HEAD/README.md", "target_id": "arxiv:2504.00906", "type": "documented_by"}, {"evidence_url": "https://raw.githubusercontent.com/simular-ai/Agent-S/HEAD/README.md", "target_id": "organization:simular", "type": "published_by"}], "review_status": "curated", "scope": "core", "summary": "Simular's open-source framework for computer-use agents that click, type and scroll through desktop and web apps on macOS, Windows and Linux; its third generation, Agent S3, reports 72.6% on OSWorld.", "topics": ["planning", "memory"], "updated_at": null, "url": "https://github.com/simular-ai/Agent-S"}
{"aliases": ["github:mnotgod96/appagent", "mnotgod96/AppAgent", "https://github.com/mnotgod96/AppAgent"], "authors": [], "details": {}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:02Z", "role": "primary", "sha256": "ac94cca1f69807d2d31d30bd5521fbf1bda6155996881054df87a85522cd8f5b", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/TencentQQGYLab/AppAgent/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T06:42:03Z", "role": "primary", "sha256": "d9e2b107cbaa3599336510dbb6703cb30210f1b4345221e76cf75cba85c2d113", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/TencentQQGYLab/AppAgent/HEAD/LICENSE"}, {"method": "github_repository_identity", "retrieved_at": "2026-10-17T14:13:57Z", "role": "primary", "sha256": "cd15c021e5023f43c17cc6365a58177c08ae91ac7e854c64c0e531f4ed3c5d5f", "supports": ["aliases"], "url": "https://api.github.com/repos/mnotgod96/AppAgent"}, {"method": "github_repository_identity", "retrieved_at": "2026-10-17T14:13:58Z", "role": "primary", "sha256": "cd15c021e5023f43c17cc6365a58177c08ae91ac7e854c64c0e531f4ed3c5d5f", "supports": ["aliases"], "url": "https://api.github.com/repos/TencentQQGYLab/AppAgent"}], "id": "github:tencentqqgylab/appagent", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-16T06:36:02Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "MIT", "retrieved_at": "2026-10-16T06:42:03Z", "url": "https://raw.githubusercontent.com/TencentQQGYLab/AppAgent/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/TencentQQGYLab/AppAgent"}], "name": "AppAgent", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/TencentQQGYLab/AppAgent/HEAD/README.md", "target_id": "arxiv:2312.13771", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Multimodal agent that operates Android apps through taps and swipes without back-end access, building a reusable knowledge base from autonomous exploration or from watching human demonstrations.", "topics": ["mobile-navigation", "memory"], "updated_at": null, "url": "https://github.com/TencentQQGYLab/AppAgent"}
{"aliases": [], "authors": [], "details": {}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:02Z", "role": "primary", "sha256": "085f1a1810805bcade9555b1e5c374a7db1b8b1377afeeea285d927853b02f3d", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/Westlake-AGI-Lab/AppAgentX/HEAD/README.md"}], "id": "github:westlake-agi-lab/appagentx", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-16T06:36:02Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/Westlake-AGI-Lab/AppAgentX"}], "name": "AppAgentX", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/Westlake-AGI-Lab/AppAgentX/HEAD/README.md", "target_id": "arxiv:2503.02268", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Mobile agent research using accumulated interaction experience.", "topics": ["mobile-navigation", "memory"], "updated_at": null, "url": "https://github.com/Westlake-AGI-Lab/AppAgentX"}
{"aliases": [], "authors": [], "details": {"availability": "Public WebArena/Mind2Web pipeline, workflow induction and inference source.", "license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/zorazrw/agent-workflow-memory/HEAD/LICENSE"], "reason": "Project source LICENSE, without transferring rights to external benchmark data or model services."}}, "repository_revision": "8c0ff8cd11d648c8fceb99e4e42f37e3b75381b1"}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:30:56Z", "role": "primary", "sha256": "dd2cc4823b2a0b913f44fbaecf9f2cea98257871bdf974de6979459e49cb7b63", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations"], "url": "https://raw.githubusercontent.com/zorazrw/agent-workflow-memory/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:30:56Z", "role": "primary", "sha256": "f5c44d559932a22add3e94727b9b413d7c286f835db1396ae0fa1f05e20e773e", "supports": ["identity"], "url": "https://api.github.com/repos/zorazrw/agent-workflow-memory"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:32:06Z", "role": "primary", "sha256": "602e2238829ad29ef15c75ac7dc61f8bd3dee6a1a74adf0d8ce023067fde6da3", "supports": ["release_status", "details.repository_revision"], "url": "https://api.github.com/repos/zorazrw/agent-workflow-memory/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:34:10Z", "role": "primary", "sha256": "1e504b3bd326f7858b39e7f2bedf5ff55d69ffb126e11af364b4184fb980882e", "supports": ["notes", "release_status"], "url": "https://raw.githubusercontent.com/zorazrw/agent-workflow-memory/HEAD/webarena/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:34:11Z", "role": "primary", "sha256": "e047de93e1c73a39ab6bbf310f02e6023d15c2b06a2b1e80ad4a0875181a9533", "supports": ["notes", "release_status"], "url": "https://raw.githubusercontent.com/zorazrw/agent-workflow-memory/HEAD/mind2web/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:34:10Z", "role": "primary", "sha256": "c71d239df91726fc519c6eb72d318ec65820627232b2f796219e87dcf35d0ab4", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/zorazrw/agent-workflow-memory/HEAD/LICENSE"}], "id": "github:zorazrw/agent-workflow-memory", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-17T12:30:56Z", "licenses": {"code": {"basis": "Project source LICENSE, without transferring rights to external benchmark data or model services.", "declared": "Apache-2.0", "retrieved_at": "2026-10-17T12:34:10Z", "url": "https://raw.githubusercontent.com/zorazrw/agent-workflow-memory/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/zorazrw/agent-workflow-memory"}], "name": "Agent Workflow Memory implementation", "notes": "The repository releases separate WebArena and Mind2Web pipelines for online or offline workflow induction and use. WebArena execution requires separately configured browser environments, while Mind2Web workflows require the original dataset and rankings. Publisher performance claims were not reproduced. Apache-2.0 is recorded for this implementation; external benchmark, model and service terms remain separate.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/zorazrw/agent-workflow-memory/HEAD/README.md", "target_id": "arxiv:2409.07429", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Research code that extracts reusable browser-task routines from demonstrations or prior executions, stores them as workflow memory and reuses them during subsequent navigation.", "topics": ["workflow", "memory", "web-navigation"], "updated_at": null, "url": "https://github.com/zorazrw/agent-workflow-memory"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/AriaUI/Aria-UI/HEAD/README.md", "population": "context-aware episode training release", "qualifier": "approximately", "unit": "instruction-output pairs", "value": 992000}], "evidence_urls": ["https://raw.githubusercontent.com/AriaUI/Aria-UI/HEAD/README.md"], "formats": [], "modalities": ["screen observations", "instructions", "interaction history"], "notes": "The Hub card contains only licensing metadata. The project README supports the episode-training description and rounded count; archive-level quality and exact row counts remain unverified.", "splits": [], "supervision": ["context-aware grounding"]}, "disabled": false, "download_url": "https://huggingface.co/datasets/Aria-UI/Aria-UI_Context-aware_Data/tree/main", "gated": false, "hub_id": "Aria-UI/Aria-UI_Context-aware_Data", "languages": [], "modalities": [], "size_category_reported": null}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T11:11:53Z", "role": "primary", "sha256": "79f91aab71bb8a0ec3ad9859cbc42d70227e01db729ec80dc5690b466b259cad", "supports": ["name", "kind"], "url": "https://huggingface.co/api/datasets/Aria-UI/Aria-UI_Context-aware_Data"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:00Z", "role": "primary", "sha256": "a2a3c5ad115e4e2109b56686c9e0b073bcfec6d9f052a58bc92683af44933480", "supports": ["name", "summary", "environments", "topics", "scope", "details.dataset_profile", "relations"], "url": "https://raw.githubusercontent.com/AriaUI/Aria-UI/HEAD/README.md"}, {"method": "huggingface_api", "retrieved_at": "2026-10-16T11:11:53Z", "role": "primary", "sha256": "79f91aab71bb8a0ec3ad9859cbc42d70227e01db729ec80dc5690b466b259cad", "supports": ["details", "licenses.data"], "url": "https://huggingface.co/api/datasets/Aria-UI/Aria-UI_Context-aware_Data"}], "id": "url:07b65dc937756ddff4a1", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T11:11:53Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card", "declared": "apache-2.0", "retrieved_at": "2026-10-16T11:11:53Z", "url": "https://huggingface.co/datasets/Aria-UI/Aria-UI_Context-aware_Data"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/Aria-UI/Aria-UI_Context-aware_Data"}], "name": "Aria-UI Context-aware Data", "notes": "The publisher reports approximately 992K instruction-output pairs in its repository. The Hub card contains only a license declaration; schema quality and exact sample counts were not independently verified.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/AriaUI/Aria-UI/HEAD/README.md", "target_id": "github:ariaui/aria-ui", "type": "component_of"}], "review_status": "curated", "scope": "core", "summary": "Episode-based mobile GUI grounding data released with Aria-UI for incorporating interaction history into target localization.", "topics": ["grounding", "memory"], "updated_at": null, "url": "https://huggingface.co/datasets/Aria-UI/Aria-UI_Context-aware_Data"}
{"aliases": [], "authors": [], "details": {"availability": "Ungated JSON conversation shards and ranking-file manifest.", "dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://huggingface.co/datasets/magicgh/MT-Mind2Web/raw/main/README.md", "population": "train", "qualifier": "exact", "unit": "conversations", "value": 600}, {"basis": "reported", "evidence_url": "https://huggingface.co/datasets/magicgh/MT-Mind2Web/raw/main/README.md", "population": "test-task", "qualifier": "exact", "unit": "conversations", "value": 34}, {"basis": "reported", "evidence_url": "https://huggingface.co/datasets/magicgh/MT-Mind2Web/raw/main/README.md", "population": "test-website", "qualifier": "exact", "unit": "conversations", "value": 42}, {"basis": "reported", "evidence_url": "https://huggingface.co/datasets/magicgh/MT-Mind2Web/raw/main/README.md", "population": "test-subdomain", "qualifier": "exact", "unit": "conversations", "value": 44}, {"basis": "reported", "evidence_url": "https://huggingface.co/datasets/magicgh/MT-Mind2Web/raw/main/README.md", "population": "train", "qualifier": "exact", "unit": "dialogue turns", "value": 2896}, {"basis": "reported", "evidence_url": "https://huggingface.co/datasets/magicgh/MT-Mind2Web/raw/main/README.md", "population": "test-task", "qualifier": "exact", "unit": "dialogue turns", "value": 191}, {"basis": "reported", "evidence_url": "https://huggingface.co/datasets/magicgh/MT-Mind2Web/raw/main/README.md", "population": "test-website", "qualifier": "exact", "unit": "dialogue turns", "value": 218}, {"basis": "reported", "evidence_url": "https://huggingface.co/datasets/magicgh/MT-Mind2Web/raw/main/README.md", "population": "test-subdomain", "qualifier": "exact", "unit": "dialogue turns", "value": 216}], "evidence_urls": ["https://huggingface.co/datasets/magicgh/MT-Mind2Web/raw/main/README.md"], "formats": ["JSON", "auxiliary PKL"], "modalities": ["dialogue text", "HTML", "element candidates", "actions"], "notes": "Conversation, dialogue-turn and action-step units differ. The released file manifest is inspected; underlying examples were not downloaded.", "splits": ["train", "test_task", "test_website", "test_subdomain"], "supervision": ["multi-turn web instruction following", "element/action prediction"]}, "license_review": {"data": {"evidence_urls": ["https://huggingface.co/datasets/magicgh/MT-Mind2Web/raw/main/README.md"], "reason": "Publisher dataset-card declaration for the multi-turn annotation release."}}, "release_revision": "5ae9a3bedf431f54c2218fbb8ee474e5efbda3a8"}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:34:10Z", "role": "primary", "sha256": "8685e0a26a06b3339b5c078632a7b0ade1ead40e6b586ef1ff3676414699a4bc", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "details.dataset_profile", "licenses.data"], "url": "https://huggingface.co/datasets/magicgh/MT-Mind2Web/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T12:34:10Z", "role": "primary", "sha256": "4455e6daf1151694c03b2a1ebd4a7af40b5b079b0568706a1783a0222383bb1d", "supports": ["identity", "release_status", "details.release_revision"], "url": "https://huggingface.co/api/datasets/magicgh/MT-Mind2Web"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:58Z", "role": "primary", "sha256": "c87e97eb397da04806d53145cf9c13be739fc657bb54dca9318c8c4ed2a36249", "supports": ["relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "url:bdfb50330deb97893ce4", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-17T12:34:10Z", "licenses": {"code": null, "data": {"basis": "Publisher dataset-card declaration for the multi-turn annotation release.", "declared": "cc-by-4.0", "retrieved_at": "2026-10-17T12:34:10Z", "url": "https://huggingface.co/datasets/magicgh/MT-Mind2Web/raw/main/README.md"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/magicgh/MT-Mind2Web"}], "name": "MT-Mind2Web", "notes": "The ungated Hub manifest contains train, test-task, test-website and test-subdomain JSON shards plus a candidate-ranking pickle file. Counts distinguish conversation sessions from dialogue turns; neither is a screenshot or action-step count. No pickle or sample data was loaded. The card declares CC-BY-4.0 and identifies the original Mind2Web interactions as its source.", "published_at": null, "relations": [{"evidence_url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=800&max_results=100&sortBy=submittedDate&sortOrder=descending", "target_id": "arxiv:2402.15057", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Multi-turn web-navigation conversations constructed from Mind2Web interactions, with dialogue turns, HTML snapshots, element candidates and click/type/select action sequences.", "topics": ["web-navigation", "imitation-learning", "memory"], "updated_at": null, "url": "https://huggingface.co/datasets/magicgh/MT-Mind2Web"}