{"aliases": [], "authors": ["Mingzhe Xing", "Rongkai Zhang", "Hui Xue", "Qi Chen", "Fan Yang", "Zhen Xiao"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "0504cfffc5d46ecf447a4fdeac561aede9482bb8640fcfd9d7f02cf6eaf47597", "submitted_at": "2024-02-09T18:19:25Z", "updated_at": "2024-02-09T18:19:25Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2402.06596v1"}}, "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:2402.06596", "identifiers": {"arxiv": "2402.06596"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:24Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2402.06596"}], "name": "Understanding the Weakness of Large Language Model Agents within a Complex Android Environment", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-02-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces AndroidArena and analyzes agent limitations in cross-application tasks, user constraints, exploration, and reflection.", "topics": ["evaluation", "cross-app"], "updated_at": "2024-02-09", "url": "https://arxiv.org/abs/2402.06596"}
{"aliases": [], "authors": ["Chaoyun Zhang", "Liqun Li", "Shilin He", "Xu Zhang", "Bo Qiao", "Si Qin", "Minghua Ma", "Yu Kang", "Qingwei Lin", "Saravan Rajmohan", "Dongmei Zhang", "Qi Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "533acb1c9aa3f1c77adb2c67779da86e3d5de95264ba3ec1f81d803f67da7753", "submitted_at": "2024-02-08T15:40:35Z", "updated_at": "2024-05-23T05:18:58Z", "version": 5, "versioned_url": "https://arxiv.org/abs/2402.07939v5"}, "publication": {"authors": ["Chaoyun Zhang", "Liqun Li", "Shilin He", "Xu Zhang", "Bo Qiao", "Si Qin", "Minghua Ma", "Yu Kang", "Qingwei Lin", "Saravan Rajmohan", "Dongmei Zhang", "Qi Zhang"], "date_basis": "Explicit publisher citation_publication_date: 2025/4. Preserve supplied precision; do not substitute conference event dates.", "date_parts": [2025, 4], "doi": "10.18653/v1/2025.naacl-long.26", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2025.naacl-long.26/"], "identity_basis": "Same work: normalized title, ordered author correspondence and the specific contribution in the complete publisher abstract were reviewed against the retained arXiv identity.", "issue": null, "notes": "Publisher proceedings edition; retained separately from the arXiv snapshot.", "pages": "597-622", "pdf_url": "https://aclanthology.org/2025.naacl-long.26.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "UFO: A UI-Focused Agent for Windows OS Interaction", "type": "conference-paper", "url": "https://aclanthology.org/2025.naacl-long.26/", "venue": "Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)", "volume": null}}, "environments": ["windows", "desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:24Z", "role": "primary", "sha256": "3aac6e9748769efff3c475ede0da192776daa03431d231b5f67e2aa50f8a3be9", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2308.03688%2C2308.15272%2C2309.08172%2C2309.11436%2C2310.04716%2C2310.04869%2C2310.05126%2C2310.06770%2C2310.08740%2C2310.10634%2C2310.11441%2C2310.16042%2C2310.18331%2C2311.01767%2C2311.07562%2C2311.09278%2C2311.12983%2C2311.18751%2C2312.03003%2C2312.06677%2C2312.08914%2C2312.11671%2C2312.13108%2C2312.13771%2C2312.15820%2C2312.16062%2C2401.01614%2C2401.04124%2C2401.07781%2C2401.10935%2C2401.13178%2C2401.13649%2C2401.13919%2C2401.14268%2C2401.16158%2C2402.01030%2C2402.04476%2C2402.04615%2C2402.05930%2C2402.06596%2C2402.07456%2C2402.07939%2C2402.07945%2C2402.08567%2C2402.11941%2C2402.14073%2C2402.15057%2C2402.16965%2C2402.17553%2C2403.02713&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:09:48Z", "role": "primary", "sha256": "bbe15e312bcbc923811ef73e561fa7db6a7244ee3747bb9e3792e166b9327235", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.publisher", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue"], "url": "https://aclanthology.org/2025.naacl-long.26/"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:24Z", "role": "primary", "sha256": "3aac6e9748769efff3c475ede0da192776daa03431d231b5f67e2aa50f8a3be9", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2308.03688%2C2308.15272%2C2309.08172%2C2309.11436%2C2310.04716%2C2310.04869%2C2310.05126%2C2310.06770%2C2310.08740%2C2310.10634%2C2310.11441%2C2310.16042%2C2310.18331%2C2311.01767%2C2311.07562%2C2311.09278%2C2311.12983%2C2311.18751%2C2312.03003%2C2312.06677%2C2312.08914%2C2312.11671%2C2312.13108%2C2312.13771%2C2312.15820%2C2312.16062%2C2401.01614%2C2401.04124%2C2401.07781%2C2401.10935%2C2401.13178%2C2401.13649%2C2401.13919%2C2401.14268%2C2401.16158%2C2402.01030%2C2402.04476%2C2402.04615%2C2402.05930%2C2402.06596%2C2402.07456%2C2402.07939%2C2402.07945%2C2402.08567%2C2402.11941%2C2402.14073%2C2402.15057%2C2402.16965%2C2402.17553%2C2403.02713&max_results=50"}], "id": "arxiv:2402.07939", "identifiers": {"arxiv": "2402.07939"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:24Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2402.07939"}, {"rel": "publication", "url": "https://aclanthology.org/2025.naacl-long.26/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2025.naacl-long.26.pdf"}], "name": "UFO: A UI-Focused Agent for Windows OS Interaction", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-02-08", "relations": [], "review_status": "curated", "scope": "core", "summary": "UFO uses two agents to inspect Windows application interfaces and their control metadata, then grounds and executes user requests within and across applications. The study evaluates this UI-focused architecture on nine Windows applications.", "topics": ["cross-app", "planning", "grounding", "automation"], "updated_at": "2024-05-23", "url": "https://arxiv.org/abs/2402.07939"}
{"aliases": [], "authors": ["Tianbao Xie", "Danyang Zhang", "Jixuan Chen", "Xiaochuan Li", "Siheng Zhao", "Ruisheng Cao", "Toh Jing Hua", "Zhoujun Cheng", "Dongchan Shin", "Fangyu Lei", "Yitao Liu", "Yiheng Xu", "Shuyan Zhou", "Silvio Savarese", "Caiming Xiong", "Victor Zhong", "Tao Yu"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a347bfcb4de03a60410392b52d522e5b7ad5dbabb474d60fc1ec097e74c4008d", "submitted_at": "2024-04-11T17:56:05Z", "updated_at": "2024-05-30T08:55:12Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2404.07972v2"}, "publication": {"authors": ["Tianbao Xie", "Danyang Zhang", "Jixuan Chen", "Xiaochuan Li", "Siheng Zhao", "Ruisheng Cao", "Toh J. Hua", "Zhoujun Cheng", "Dongchan Shin", "Fangyu Lei", "Yitao Liu", "Yiheng Xu", "Shuyan Zhou", "Silvio Savarese", "Caiming Xiong", "Victor Zhong", "Tao Yu"], "date_basis": "Explicit publisher citation_publication_date: 2024-12-16. Preserve supplied precision; do not substitute conference event dates.", "date_parts": [2024, 12, 16], "doi": "10.52202/079017-1650", "event_dates": null, "evidence_urls": ["https://proceedings.neurips.cc/paper_files/paper/2024/hash/5d413e48f84dc61244b6be550f1cd8f5-Abstract-Datasets_and_Benchmarks_Track.html"], "identity_basis": "Same work: normalized title, ordered author correspondence and the specific contribution in the complete publisher abstract were reviewed against the retained arXiv identity.", "issue": null, "notes": "Publisher proceedings edition; retained separately from the arXiv snapshot. Publisher spelling, initials, name annotations or author lists differ from the retained preprint.", "pages": "52040-52094", "pdf_url": "https://proceedings.neurips.cc/paper_files/paper/2024/file/5d413e48f84dc61244b6be550f1cd8f5-Paper-Datasets_and_Benchmarks_Track.pdf", "publisher": null, "series": null, "title": "OSWorld: Benchmarking Multimodal Agents for Open-Ended Tasks in Real Computer Environments", "type": "conference-paper", "url": "https://proceedings.neurips.cc/paper_files/paper/2024/hash/5d413e48f84dc61244b6be550f1cd8f5-Abstract-Datasets_and_Benchmarks_Track.html", "venue": "Advances in Neural Information Processing Systems", "volume": "37"}}, "environments": ["desktop", "linux", "windows", "macos"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:28Z", "role": "primary", "sha256": "be244bfae7f89ef9449b69d85aac44295eb7b5f7e5e1bbedd76df312696676a7", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2403.03186%2C2403.03788%2C2403.07718%2C2403.10170%2C2403.11905%2C2403.16971%2C2403.17918%2C2403.20271%2C2404.01549%2C2404.01744%2C2404.02475%2C2404.03648%2C2404.05427%2C2404.05719%2C2404.05955%2C2404.06474%2C2404.07972%2C2404.08755%2C2404.08860%2C2404.08948%2C2404.09992%2C2404.10887%2C2404.11459%2C2404.16006%2C2404.16048%2C2404.16054%2C2404.16660%2C2404.18074%2C2404.19296%2C2405.00516%2C2405.11120%2C2405.12842%2C2405.14573%2C2405.14751%2C2405.15793%2C2405.16247%2C2405.20309%2C2406.01014%2C2406.01623%2C2406.03679%2C2406.06947%2C2406.07822%2C2406.08184%2C2406.08340%2C2406.08451%2C2406.10227%2C2406.10819%2C2406.11317%2C2406.11896%2C2406.12373&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:10:02Z", "role": "primary", "sha256": "f62cbd2e5c92d250cc894a05e46f9052cfd38f0cdbf14902af6a96b96bcc9e8d", "supports": ["details.publication.authors", "details.publication.date_basis", "details.publication.date_parts", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.title", "details.publication.type", "details.publication.url", "details.publication.venue", "details.publication.volume"], "url": "https://proceedings.neurips.cc/paper_files/paper/2024/hash/5d413e48f84dc61244b6be550f1cd8f5-Abstract-Datasets_and_Benchmarks_Track.html"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:28Z", "role": "primary", "sha256": "be244bfae7f89ef9449b69d85aac44295eb7b5f7e5e1bbedd76df312696676a7", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2403.03186%2C2403.03788%2C2403.07718%2C2403.10170%2C2403.11905%2C2403.16971%2C2403.17918%2C2403.20271%2C2404.01549%2C2404.01744%2C2404.02475%2C2404.03648%2C2404.05427%2C2404.05719%2C2404.05955%2C2404.06474%2C2404.07972%2C2404.08755%2C2404.08860%2C2404.08948%2C2404.09992%2C2404.10887%2C2404.11459%2C2404.16006%2C2404.16048%2C2404.16054%2C2404.16660%2C2404.18074%2C2404.19296%2C2405.00516%2C2405.11120%2C2405.12842%2C2405.14573%2C2405.14751%2C2405.15793%2C2405.16247%2C2405.20309%2C2406.01014%2C2406.01623%2C2406.03679%2C2406.06947%2C2406.07822%2C2406.08184%2C2406.08340%2C2406.08451%2C2406.10227%2C2406.10819%2C2406.11317%2C2406.11896%2C2406.12373&max_results=50"}], "id": "arxiv:2404.07972", "identifiers": {"arxiv": "2404.07972"}, "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/2404.07972"}, {"rel": "publication", "url": "https://proceedings.neurips.cc/paper_files/paper/2024/hash/5d413e48f84dc61244b6be550f1cd8f5-Abstract-Datasets_and_Benchmarks_Track.html"}, {"rel": "publication_pdf", "url": "https://proceedings.neurips.cc/paper_files/paper/2024/file/5d413e48f84dc61244b6be550f1cd8f5-Paper-Datasets_and_Benchmarks_Track.pdf"}], "name": "OSWorld: Benchmarking Multimodal Agents for Open-Ended Tasks in Real Computer Environments", "notes": "Editorial summary based on the primary paper abstract; experiments and released artifacts have not been independently reproduced.", "published_at": "2024-04-11", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces a real-computer evaluation environment with execution-based checks for tasks spanning desktop applications and operating-system state.", "topics": ["evaluation", "cross-app"], "updated_at": "2024-05-30", "url": "https://arxiv.org/abs/2404.07972"}
{"aliases": [], "authors": ["Shengcheng Yu", "Chunrong Fang", "Mingzhe Du", "Zimin Ding", "Zhenyu Chen", "Zhendong Su"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "0061e9004afe11bf188164ea37de1acd15e1b95ae2fcad22044264a89e4636b3", "submitted_at": "2024-06-12T15:48:39Z", "updated_at": "2024-06-12T15:48:39Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2406.08340v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:28Z", "role": "primary", "sha256": "be244bfae7f89ef9449b69d85aac44295eb7b5f7e5e1bbedd76df312696676a7", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2403.03186%2C2403.03788%2C2403.07718%2C2403.10170%2C2403.11905%2C2403.16971%2C2403.17918%2C2403.20271%2C2404.01549%2C2404.01744%2C2404.02475%2C2404.03648%2C2404.05427%2C2404.05719%2C2404.05955%2C2404.06474%2C2404.07972%2C2404.08755%2C2404.08860%2C2404.08948%2C2404.09992%2C2404.10887%2C2404.11459%2C2404.16006%2C2404.16048%2C2404.16054%2C2404.16660%2C2404.18074%2C2404.19296%2C2405.00516%2C2405.11120%2C2405.12842%2C2405.14573%2C2405.14751%2C2405.15793%2C2405.16247%2C2405.20309%2C2406.01014%2C2406.01623%2C2406.03679%2C2406.06947%2C2406.07822%2C2406.08184%2C2406.08340%2C2406.08451%2C2406.10227%2C2406.10819%2C2406.11317%2C2406.11896%2C2406.12373&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:28Z", "role": "primary", "sha256": "be244bfae7f89ef9449b69d85aac44295eb7b5f7e5e1bbedd76df312696676a7", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2403.03186%2C2403.03788%2C2403.07718%2C2403.10170%2C2403.11905%2C2403.16971%2C2403.17918%2C2403.20271%2C2404.01549%2C2404.01744%2C2404.02475%2C2404.03648%2C2404.05427%2C2404.05719%2C2404.05955%2C2404.06474%2C2404.07972%2C2404.08755%2C2404.08860%2C2404.08948%2C2404.09992%2C2404.10887%2C2404.11459%2C2404.16006%2C2404.16048%2C2404.16054%2C2404.16660%2C2404.18074%2C2404.19296%2C2405.00516%2C2405.11120%2C2405.12842%2C2405.14573%2C2405.14751%2C2405.15793%2C2405.16247%2C2405.20309%2C2406.01014%2C2406.01623%2C2406.03679%2C2406.06947%2C2406.07822%2C2406.08184%2C2406.08340%2C2406.08451%2C2406.10227%2C2406.10819%2C2406.11317%2C2406.11896%2C2406.12373&max_results=50"}], "id": "arxiv:2406.08340", "identifiers": {"arxiv": "2406.08340"}, "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/2406.08340"}], "name": "Practical, Automated Scenario-based Mobile App Testing", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-06-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "ScenTest builds scenario-specific event knowledge graphs from crowdsourced test reports containing GUI and action information. It uses the current app state and testing context to generate scenario tests across applications.", "topics": ["automation", "cross-app", "screen-understanding", "evaluation"], "updated_at": "2024-06-12", "url": "https://arxiv.org/abs/2406.08340"}
{"aliases": [], "authors": ["Quanfeng Lu", "Wenqi Shao", "Zitao Liu", "Lingxiao Du", "Fanqing Meng", "Boxuan Li", "Botong Chen", "Siyuan Huang", "Kaipeng Zhang", "Ping Luo"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "8b45039151353cc0fede71e2e9ba109ec53d0c1f495037a82d8fb742c9e97bbb", "submitted_at": "2024-06-12T17:44:26Z", "updated_at": "2025-08-01T03:37:57Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2406.08451v2"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:28Z", "role": "primary", "sha256": "be244bfae7f89ef9449b69d85aac44295eb7b5f7e5e1bbedd76df312696676a7", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2403.03186%2C2403.03788%2C2403.07718%2C2403.10170%2C2403.11905%2C2403.16971%2C2403.17918%2C2403.20271%2C2404.01549%2C2404.01744%2C2404.02475%2C2404.03648%2C2404.05427%2C2404.05719%2C2404.05955%2C2404.06474%2C2404.07972%2C2404.08755%2C2404.08860%2C2404.08948%2C2404.09992%2C2404.10887%2C2404.11459%2C2404.16006%2C2404.16048%2C2404.16054%2C2404.16660%2C2404.18074%2C2404.19296%2C2405.00516%2C2405.11120%2C2405.12842%2C2405.14573%2C2405.14751%2C2405.15793%2C2405.16247%2C2405.20309%2C2406.01014%2C2406.01623%2C2406.03679%2C2406.06947%2C2406.07822%2C2406.08184%2C2406.08340%2C2406.08451%2C2406.10227%2C2406.10819%2C2406.11317%2C2406.11896%2C2406.12373&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:28Z", "role": "primary", "sha256": "be244bfae7f89ef9449b69d85aac44295eb7b5f7e5e1bbedd76df312696676a7", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2403.03186%2C2403.03788%2C2403.07718%2C2403.10170%2C2403.11905%2C2403.16971%2C2403.17918%2C2403.20271%2C2404.01549%2C2404.01744%2C2404.02475%2C2404.03648%2C2404.05427%2C2404.05719%2C2404.05955%2C2404.06474%2C2404.07972%2C2404.08755%2C2404.08860%2C2404.08948%2C2404.09992%2C2404.10887%2C2404.11459%2C2404.16006%2C2404.16048%2C2404.16054%2C2404.16660%2C2404.18074%2C2404.19296%2C2405.00516%2C2405.11120%2C2405.12842%2C2405.14573%2C2405.14751%2C2405.15793%2C2405.16247%2C2405.20309%2C2406.01014%2C2406.01623%2C2406.03679%2C2406.06947%2C2406.07822%2C2406.08184%2C2406.08340%2C2406.08451%2C2406.10227%2C2406.10819%2C2406.11317%2C2406.11896%2C2406.12373&max_results=50"}], "id": "arxiv:2406.08451", "identifiers": {"arxiv": "2406.08451"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:28Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2406.08451"}], "name": "GUIOdyssey: A Comprehensive Dataset for Cross-App GUI Navigation on Mobile Devices", "notes": "Summary and topic tags reviewed against the primary abstract. Reported experiments, task counts, and artifact availability are not independently reproduced.", "published_at": "2024-06-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces annotated cross-application mobile trajectories and studies history-aware navigation over long task sequences.", "topics": ["cross-app", "data-generation"], "updated_at": "2025-08-01", "url": "https://arxiv.org/abs/2406.08451"}
{"aliases": [], "authors": ["Shihan Deng", "Weikai Xu", "Hongda Sun", "Wei Liu", "Tao Tan", "Jianfeng Liu", "Ang Li", "Jian Luan", "Bin Wang", "Rui Yan", "Shuo Shang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "90ce9900644c21004e0ba9e749e10d94238d2447d0e84bd7205ff55dcbd243f4", "submitted_at": "2024-07-01T06:10:01Z", "updated_at": "2024-07-01T06:10:01Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2407.00993v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:32Z", "role": "primary", "sha256": "b9a83994d8ac36ccdb606a5219511f8bb35ad3948ade4f9f3e28d386221aedfb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2406.12814%2C2406.13719%2C2406.14056%2C2406.14250%2C2406.14314%2C2406.18082%2C2406.19263%2C2407.00993%2C2407.01476%2C2407.01489%2C2407.01511%2C2407.01558%2C2407.03037%2C2407.03913%2C2407.04346%2C2407.05291%2C2407.07061%2C2407.08850%2C2407.09018%2C2407.09295%2C2407.10956%2C2407.13032%2C2407.13063%2C2407.15673%2C2407.15711%2C2407.16741%2C2407.17490%2C2407.18901%2C2407.19056%2C2407.20183%2C2408.00203%2C2408.02544%2C2408.06327%2C2408.07199%2C2408.11824%2C2408.15978%2C2409.02453%2C2409.04566%2C2409.07429%2C2409.08264%2C2409.09354%2C2409.10741%2C2409.11295%2C2409.13711%2C2409.14337%2C2409.14818%2C2409.15441%2C2409.15637%2C2409.17140%2C2409.20566&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:32Z", "role": "primary", "sha256": "b9a83994d8ac36ccdb606a5219511f8bb35ad3948ade4f9f3e28d386221aedfb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2406.12814%2C2406.13719%2C2406.14056%2C2406.14250%2C2406.14314%2C2406.18082%2C2406.19263%2C2407.00993%2C2407.01476%2C2407.01489%2C2407.01511%2C2407.01558%2C2407.03037%2C2407.03913%2C2407.04346%2C2407.05291%2C2407.07061%2C2407.08850%2C2407.09018%2C2407.09295%2C2407.10956%2C2407.13032%2C2407.13063%2C2407.15673%2C2407.15711%2C2407.16741%2C2407.17490%2C2407.18901%2C2407.19056%2C2407.20183%2C2408.00203%2C2408.02544%2C2408.06327%2C2408.07199%2C2408.11824%2C2408.15978%2C2409.02453%2C2409.04566%2C2409.07429%2C2409.08264%2C2409.09354%2C2409.10741%2C2409.11295%2C2409.13711%2C2409.14337%2C2409.14818%2C2409.15441%2C2409.15637%2C2409.17140%2C2409.20566&max_results=50"}], "id": "arxiv:2407.00993", "identifiers": {"arxiv": "2407.00993"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2407.00993"}], "name": "Mobile-Bench: An Evaluation Benchmark for LLM-based Mobile Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2024-07-01", "relations": [], "review_status": "curated", "scope": "core", "summary": "Evaluates mobile agents using combined UI and API actions, tasks with varying app scope, and checkpoints for intermediate progress.", "topics": ["evaluation", "cross-app", "tool-integration"], "updated_at": "2024-07-01", "url": "https://arxiv.org/abs/2407.00993"}
{"aliases": [], "authors": ["Tianqi Xu", "Linyao Chen", "Dai-Jie Wu", "Yanjun Chen", "Zecheng Zhang", "Xiang Yao", "Zhiqiang Xie", "Yongchao Chen", "Shilong Liu", "Bochen Qian", "Anjie Yang", "Zhaoxuan Jin", "Jianbo Deng", "Philip Torr", "Bernard Ghanem", "Guohao Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "12e5464ea0fc62bd566f19d301b887826b7ff317cfc27c86fa0cda8d3de271d9", "submitted_at": "2024-07-01T17:55:04Z", "updated_at": "2025-07-20T13:42:07Z", "version": 4, "versioned_url": "https://arxiv.org/abs/2407.01511v4"}, "publication": {"authors": ["Tianqi Xu", "Linyao Chen", "Dai-Jie Wu", "Yanjun Chen", "Zecheng Zhang", "Xiang Yao", "Zhiqiang Xie (谢志强)", "Yongchao Chen", "Shilong Liu", "Bochen Qian", "Anjie Yang", "Zhaoxuan Jin", "Jianbo Deng", "Philip Torr", "Bernard Ghanem", "Guohao 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.1113", "event_dates": null, "evidence_urls": ["https://aclanthology.org/2025.findings-acl.1113/"], "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": "21607-21647", "pdf_url": "https://aclanthology.org/2025.findings-acl.1113.pdf", "publisher": "Association for Computational Linguistics", "series": null, "title": "CRAB: Cross-environment Agent Benchmark for Multimodal Language Model Agents", "type": "conference-paper", "url": "https://aclanthology.org/2025.findings-acl.1113/", "venue": "Findings of the Association for Computational Linguistics: ACL 2025", "volume": null}}, "environments": ["desktop", "cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:32Z", "role": "primary", "sha256": "b9a83994d8ac36ccdb606a5219511f8bb35ad3948ade4f9f3e28d386221aedfb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2406.12814%2C2406.13719%2C2406.14056%2C2406.14250%2C2406.14314%2C2406.18082%2C2406.19263%2C2407.00993%2C2407.01476%2C2407.01489%2C2407.01511%2C2407.01558%2C2407.03037%2C2407.03913%2C2407.04346%2C2407.05291%2C2407.07061%2C2407.08850%2C2407.09018%2C2407.09295%2C2407.10956%2C2407.13032%2C2407.13063%2C2407.15673%2C2407.15711%2C2407.16741%2C2407.17490%2C2407.18901%2C2407.19056%2C2407.20183%2C2408.00203%2C2408.02544%2C2408.06327%2C2408.07199%2C2408.11824%2C2408.15978%2C2409.02453%2C2409.04566%2C2409.07429%2C2409.08264%2C2409.09354%2C2409.10741%2C2409.11295%2C2409.13711%2C2409.14337%2C2409.14818%2C2409.15441%2C2409.15637%2C2409.17140%2C2409.20566&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:09:51Z", "role": "primary", "sha256": "a8cf178f2faa088fd94b77d03785b8c6e2908db34b411287984066b4e5d1ef3f", "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.1113/"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:32Z", "role": "primary", "sha256": "b9a83994d8ac36ccdb606a5219511f8bb35ad3948ade4f9f3e28d386221aedfb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2406.12814%2C2406.13719%2C2406.14056%2C2406.14250%2C2406.14314%2C2406.18082%2C2406.19263%2C2407.00993%2C2407.01476%2C2407.01489%2C2407.01511%2C2407.01558%2C2407.03037%2C2407.03913%2C2407.04346%2C2407.05291%2C2407.07061%2C2407.08850%2C2407.09018%2C2407.09295%2C2407.10956%2C2407.13032%2C2407.13063%2C2407.15673%2C2407.15711%2C2407.16741%2C2407.17490%2C2407.18901%2C2407.19056%2C2407.20183%2C2408.00203%2C2408.02544%2C2408.06327%2C2408.07199%2C2408.11824%2C2408.15978%2C2409.02453%2C2409.04566%2C2409.07429%2C2409.08264%2C2409.09354%2C2409.10741%2C2409.11295%2C2409.13711%2C2409.14337%2C2409.14818%2C2409.15441%2C2409.15637%2C2409.17140%2C2409.20566&max_results=50"}], "id": "arxiv:2407.01511", "identifiers": {"arxiv": "2407.01511"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2407.01511"}, {"rel": "publication", "url": "https://aclanthology.org/2025.findings-acl.1113/"}, {"rel": "publication_pdf", "url": "https://aclanthology.org/2025.findings-acl.1113.pdf"}], "name": "CRAB: Cross-environment Agent Benchmark for Multimodal Language Model 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": "2024-07-01", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces CRAB, a framework for cross-environment agent tasks with graph-based evaluation and a benchmark spanning desktop and phone interactions.", "topics": ["evaluation", "cross-app"], "updated_at": "2025-07-20", "url": "https://arxiv.org/abs/2407.01511"}
{"aliases": [], "authors": ["Harsh Trivedi", "Tushar Khot", "Mareike Hartmann", "Ruskin Manku", "Vinty Dong", "Edward Li", "Shashank Gupta", "Ashish Sabharwal", "Niranjan Balasubramanian"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e59087ecf28d07baf81b60e4166a866d4bea4d8cd934bea06c2149dd61f38f1a", "submitted_at": "2024-07-26T17:55:45Z", "updated_at": "2024-07-26T17:55:45Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2407.18901v1"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:32Z", "role": "primary", "sha256": "b9a83994d8ac36ccdb606a5219511f8bb35ad3948ade4f9f3e28d386221aedfb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2406.12814%2C2406.13719%2C2406.14056%2C2406.14250%2C2406.14314%2C2406.18082%2C2406.19263%2C2407.00993%2C2407.01476%2C2407.01489%2C2407.01511%2C2407.01558%2C2407.03037%2C2407.03913%2C2407.04346%2C2407.05291%2C2407.07061%2C2407.08850%2C2407.09018%2C2407.09295%2C2407.10956%2C2407.13032%2C2407.13063%2C2407.15673%2C2407.15711%2C2407.16741%2C2407.17490%2C2407.18901%2C2407.19056%2C2407.20183%2C2408.00203%2C2408.02544%2C2408.06327%2C2408.07199%2C2408.11824%2C2408.15978%2C2409.02453%2C2409.04566%2C2409.07429%2C2409.08264%2C2409.09354%2C2409.10741%2C2409.11295%2C2409.13711%2C2409.14337%2C2409.14818%2C2409.15441%2C2409.15637%2C2409.17140%2C2409.20566&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:32Z", "role": "primary", "sha256": "b9a83994d8ac36ccdb606a5219511f8bb35ad3948ade4f9f3e28d386221aedfb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2406.12814%2C2406.13719%2C2406.14056%2C2406.14250%2C2406.14314%2C2406.18082%2C2406.19263%2C2407.00993%2C2407.01476%2C2407.01489%2C2407.01511%2C2407.01558%2C2407.03037%2C2407.03913%2C2407.04346%2C2407.05291%2C2407.07061%2C2407.08850%2C2407.09018%2C2407.09295%2C2407.10956%2C2407.13032%2C2407.13063%2C2407.15673%2C2407.15711%2C2407.16741%2C2407.17490%2C2407.18901%2C2407.19056%2C2407.20183%2C2408.00203%2C2408.02544%2C2408.06327%2C2408.07199%2C2408.11824%2C2408.15978%2C2409.02453%2C2409.04566%2C2409.07429%2C2409.08264%2C2409.09354%2C2409.10741%2C2409.11295%2C2409.13711%2C2409.14337%2C2409.14818%2C2409.15441%2C2409.15637%2C2409.17140%2C2409.20566&max_results=50"}], "id": "arxiv:2407.18901", "identifiers": {"arxiv": "2407.18901"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2407.18901"}], "name": "AppWorld: A Controllable World of Apps and People for Benchmarking Interactive Coding 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": "2024-07-26", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "Introduces simulated application APIs and state-based task evaluation for interactive coding agents performing everyday digital workflows.", "topics": ["evaluation", "tool-integration", "cross-app"], "updated_at": "2024-07-26", "url": "https://arxiv.org/abs/2407.18901"}
{"aliases": [], "authors": ["Zilong Wang", "Yuedong Cui", "Li Zhong", "Zimin Zhang", "Da Yin", "Bill Yuchen Lin", "Jingbo Shang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "c443e6a508c438332ac5990cf254a4441f5bdcc24354fd8aeca36bcbf4604a29", "submitted_at": "2024-07-26T19:27:17Z", "updated_at": "2024-07-26T19:27:17Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2407.19056v1"}}, "environments": ["desktop"], "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:2407.19056", "identifiers": {"arxiv": "2407.19056"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2407.19056"}], "name": "OfficeBench: Benchmarking Language Agents across Multiple Applications for Office 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": "2024-07-26", "relations": [], "review_status": "curated", "scope": "core", "summary": "Evaluates language agents on office workflows requiring information integration, long-term planning, and switching between applications.", "topics": ["evaluation", "cross-app", "long-horizon"], "updated_at": "2024-07-26", "url": "https://arxiv.org/abs/2407.19056"}
{"aliases": [], "authors": ["Rogerio Bonatti", "Dan Zhao", "Francesco Bonacci", "Dillon Dupont", "Sara Abdali", "Yinheng Li", "Yadong Lu", "Justin Wagle", "Kazuhito Koishida", "Arthur Bucker", "Lawrence Jang", "Zack Hui"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "fe77f7f8ffd7490192ab54b7f6d49c0cb5b10be6df0a645bfb181b3be724118a", "submitted_at": "2024-09-12T17:56:43Z", "updated_at": "2024-09-13T20:17:13Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2409.08264v2"}}, "environments": ["desktop", "windows"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:32Z", "role": "primary", "sha256": "b9a83994d8ac36ccdb606a5219511f8bb35ad3948ade4f9f3e28d386221aedfb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2406.12814%2C2406.13719%2C2406.14056%2C2406.14250%2C2406.14314%2C2406.18082%2C2406.19263%2C2407.00993%2C2407.01476%2C2407.01489%2C2407.01511%2C2407.01558%2C2407.03037%2C2407.03913%2C2407.04346%2C2407.05291%2C2407.07061%2C2407.08850%2C2407.09018%2C2407.09295%2C2407.10956%2C2407.13032%2C2407.13063%2C2407.15673%2C2407.15711%2C2407.16741%2C2407.17490%2C2407.18901%2C2407.19056%2C2407.20183%2C2408.00203%2C2408.02544%2C2408.06327%2C2408.07199%2C2408.11824%2C2408.15978%2C2409.02453%2C2409.04566%2C2409.07429%2C2409.08264%2C2409.09354%2C2409.10741%2C2409.11295%2C2409.13711%2C2409.14337%2C2409.14818%2C2409.15441%2C2409.15637%2C2409.17140%2C2409.20566&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:32Z", "role": "primary", "sha256": "b9a83994d8ac36ccdb606a5219511f8bb35ad3948ade4f9f3e28d386221aedfb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2406.12814%2C2406.13719%2C2406.14056%2C2406.14250%2C2406.14314%2C2406.18082%2C2406.19263%2C2407.00993%2C2407.01476%2C2407.01489%2C2407.01511%2C2407.01558%2C2407.03037%2C2407.03913%2C2407.04346%2C2407.05291%2C2407.07061%2C2407.08850%2C2407.09018%2C2407.09295%2C2407.10956%2C2407.13032%2C2407.13063%2C2407.15673%2C2407.15711%2C2407.16741%2C2407.17490%2C2407.18901%2C2407.19056%2C2407.20183%2C2408.00203%2C2408.02544%2C2408.06327%2C2408.07199%2C2408.11824%2C2408.15978%2C2409.02453%2C2409.04566%2C2409.07429%2C2409.08264%2C2409.09354%2C2409.10741%2C2409.11295%2C2409.13711%2C2409.14337%2C2409.14818%2C2409.15441%2C2409.15637%2C2409.17140%2C2409.20566&max_results=50"}], "id": "arxiv:2409.08264", "identifiers": {"arxiv": "2409.08264"}, "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/2409.08264"}], "name": "Windows Agent Arena: Evaluating Multi-Modal OS Agents at Scale", "notes": "Summary and topic tags reviewed against the primary abstract. Reported experiments, task counts, and artifact availability are not independently reproduced.", "published_at": "2024-09-12", "relations": [], "review_status": "curated", "scope": "core", "summary": "Adapts real-desktop evaluation to Windows applications and supports parallel execution of benchmark tasks.", "topics": ["evaluation", "cross-app"], "updated_at": "2024-09-13", "url": "https://arxiv.org/abs/2409.08264"}
{"aliases": [], "authors": ["Haowei Liu", "Xi Zhang", "Haiyang Xu", "Yuyang Wanyan", "Junyang Wang", "Ming Yan", "Ji Zhang", "Chunfeng Yuan", "Changsheng Xu", "Weiming Hu", "Fei Huang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d75b417f4248397b73934f8b8e45199006e4db6e2ee69b62e0846621411b2dd7", "submitted_at": "2025-02-20T05:41:55Z", "updated_at": "2025-02-21T02:54:09Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2502.14282v2"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2412.04454%2C2412.04626%2C2412.05467%2C2412.09362%2C2412.09605%2C2412.10342%2C2412.10742%2C2412.10840%2C2412.13103%2C2412.13194%2C2412.13501%2C2412.14161%2C2412.14470%2C2412.16256%2C2412.17589%2C2412.18116%2C2412.18426%2C2412.19723%2C2501.01149%2C2501.04575%2C2501.07572%2C2501.11733%2C2501.12326%2C2501.13896%2C2501.16150%2C2502.01600%2C2502.01977%2C2502.02982%2C2502.06395%2C2502.07949%2C2502.08047%2C2502.08226%2C2502.11357%2C2502.13053%2C2502.13923%2C2502.14282%2C2502.18906%2C2503.00401%2C2503.02268%2C2503.03196%2C2503.04730%2C2503.05143%2C2503.06470%2C2503.09263%2C2503.11069%2C2503.11170%2C2503.12687%2C2503.15661%2C2503.15937%2C2503.16465&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2412.04454%2C2412.04626%2C2412.05467%2C2412.09362%2C2412.09605%2C2412.10342%2C2412.10742%2C2412.10840%2C2412.13103%2C2412.13194%2C2412.13501%2C2412.14161%2C2412.14470%2C2412.16256%2C2412.17589%2C2412.18116%2C2412.18426%2C2412.19723%2C2501.01149%2C2501.04575%2C2501.07572%2C2501.11733%2C2501.12326%2C2501.13896%2C2501.16150%2C2502.01600%2C2502.01977%2C2502.02982%2C2502.06395%2C2502.07949%2C2502.08047%2C2502.08226%2C2502.11357%2C2502.13053%2C2502.13923%2C2502.14282%2C2502.18906%2C2503.00401%2C2503.02268%2C2503.03196%2C2503.04730%2C2503.05143%2C2503.06470%2C2503.09263%2C2503.11069%2C2503.11170%2C2503.12687%2C2503.15661%2C2503.15937%2C2503.16465&max_results=50"}], "id": "arxiv:2502.14282", "identifiers": {"arxiv": "2502.14282"}, "kind": "paper", "last_verified_at": "2026-10-16T06:31:38Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2502.14282"}], "name": "PC-Agent: A Hierarchical Multi-Agent Collaboration Framework for Complex Task Automation on PC", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-02-20", "relations": [], "review_status": "curated", "scope": "core", "summary": "Uses active perception and hierarchical agents for instruction decomposition, progress tracking, action selection, and reflection in PC workflows.", "topics": ["planning", "cross-app"], "updated_at": "2025-02-21", "url": "https://arxiv.org/abs/2502.14282"}
{"aliases": [], "authors": ["Xiaolei Li", "Jialun Cao", "Yepang Liu", "Shing-Chi Cheung", "Hailong Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "a646bb2136cdaf47a0d8b3f485d3e1973fc15ac4f844e1beddedf6021b56ffe4", "submitted_at": "2025-04-03T07:45:09Z", "updated_at": "2025-04-03T07:45:09Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2504.02357v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:45:04Z", "role": "primary", "sha256": "bab88dfed88ca440c231aa768530374a0f19498777a70921481a904fddd46bb8", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2303.17760%2C2310.03720%2C2312.11190%2C2403.02502%2C2406.12639%2C2409.18060%2C2410.00201%2C2501.01702%2C2501.11425%2C2502.02533%2C2502.02955%2C2502.12110%2C2502.15760%2C2502.17110%2C2502.19328%2C2503.02197%2C2503.02403%2C2503.02682%2C2503.06580%2C2503.07365%2C2503.09572%2C2503.14021%2C2503.24180%2C2504.02357%2C2504.11675%2C2504.13936&max_results=26"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:45:04Z", "role": "primary", "sha256": "bab88dfed88ca440c231aa768530374a0f19498777a70921481a904fddd46bb8", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "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"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:12:09Z", "role": "primary", "sha256": "054a5a305d86c7fb9107d467d194fc4a64aa4f5af23d214bea7f52c2f2c9fd35", "supports": ["environments"], "url": "https://arxiv.org/html/2504.02357"}], "id": "arxiv:2504.02357", "identifiers": {"arxiv": "2504.02357"}, "kind": "paper", "last_verified_at": "2026-10-16T06:45:04Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2504.02357"}], "name": "ReuseDroid: A VLM-empowered Android UI Test Migrator Boosted by Active Feedback", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced. The platform or GUI evaluation setting was additionally checked in the primary full text.", "published_at": "2025-04-03", "relations": [], "review_status": "curated", "scope": "core", "summary": "ReuseDroid migrates Android GUI tests between functionally similar apps using VLM agents that separate shared test intent from app-specific interaction flows. It introduces LinPro, a dataset of 578 migration tasks across 39 apps, to evaluate test transfer.", "topics": ["automation", "evaluation", "cross-app"], "updated_at": "2025-04-03", "url": "https://arxiv.org/abs/2504.02357"}
{"aliases": [], "authors": ["Chaoyun Zhang", "He Huang", "Chiming Ni", "Jian Mu", "Si Qin", "Shilin He", "Lu Wang", "Fangkai Yang", "Pu Zhao", "Chao Du", "Liqun Li", "Yu Kang", "Zhao Jiang", "Suzhen Zheng", "Rujia Wang", "Jiaxu Qian", "Minghua Ma", "Jian-Guang Lou", "Qingwei Lin", "Saravan Rajmohan", "Dongmei Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "dc47a80b48a6a5f126e36fa14639d56d8c8025a453fbcee2cbdb73eb67945561", "submitted_at": "2025-04-20T13:04:43Z", "updated_at": "2025-04-25T05:14:14Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2504.14603v2"}}, "environments": ["desktop", "windows"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T07:57:47Z", "role": "primary", "sha256": "27144160e8981bb9f36b60c391689774913ff3f026eeeffc2923154c404a6529", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2504.03206%2C2504.03601%2C2504.06821%2C2504.09723%2C2504.11301%2C2504.11543%2C2504.12609%2C2504.14233%2C2504.14348%2C2504.14603%2C2504.16073%2C2504.16419%2C2504.17892%2C2504.18580%2C2504.20965%2C2505.00019%2C2505.00693%2C2505.01578%2C2505.02094%2C2505.03570%2C2505.03574%2C2505.05026%2C2505.05849%2C2505.06507%2C2505.07003%2C2505.07473%2C2505.10468%2C2505.10887%2C2505.12053%2C2505.12490%2C2505.12728%2C2505.13291%2C2505.14030%2C2505.14454%2C2505.14646%2C2505.15801%2C2505.15816%2C2505.17020%2C2505.18746%2C2505.19095%2C2505.19713%2C2505.20023%2C2505.20732%2C2505.20795%2C2505.21055%2C2505.22304%2C2505.22429%2C2505.23450%2C2505.23559%2C2506.00641&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:20:07Z", "role": "primary", "sha256": "0ff3084c62a2ae0fbda31be5d409fb8ca3a6887da897a238beec93a8d7be7825", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/abs/2504.14603v2"}], "id": "arxiv:2504.14603", "identifiers": {"arxiv": "2504.14603"}, "kind": "paper", "last_verified_at": "2026-10-17T11:20:07Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2504.14603"}], "name": "UFO2: The Desktop AgentOS", "notes": "Reviewed against the exact-version official arXiv abstract page; its complete title, ordered author list, visible version, and normalized abstract SHA match the retained snapshot. The original API response body was not locally available. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2025-04-20", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces UFO2, a Windows desktop agent architecture combining a coordinating HostAgent, application-specialized agents, and a shared GUI/API action layer. It integrates UI Automation, visual parsing, speculative planning, and an isolated picture-in-picture desktop.", "topics": ["automation", "cross-app", "tool-integration"], "updated_at": "2025-04-25", "url": "https://arxiv.org/abs/2504.14603"}
{"aliases": [], "authors": ["Mariya Davydova", "Daniel Jeffries", "Patrick Barker", "Arturo Márquez Flores", "Sinéad Ryan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e8bb4c6f8080ed1ee91e1089233cae6f231ab3256b38ac7d6403ebea3ae56fe0", "submitted_at": "2025-05-06T14:29:47Z", "updated_at": "2025-05-06T14:29:47Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2505.03570v1"}}, "environments": ["desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-17T07:57:47Z", "role": "primary", "sha256": "27144160e8981bb9f36b60c391689774913ff3f026eeeffc2923154c404a6529", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2504.03206%2C2504.03601%2C2504.06821%2C2504.09723%2C2504.11301%2C2504.11543%2C2504.12609%2C2504.14233%2C2504.14348%2C2504.14603%2C2504.16073%2C2504.16419%2C2504.17892%2C2504.18580%2C2504.20965%2C2505.00019%2C2505.00693%2C2505.01578%2C2505.02094%2C2505.03570%2C2505.03574%2C2505.05026%2C2505.05849%2C2505.06507%2C2505.07003%2C2505.07473%2C2505.10468%2C2505.10887%2C2505.12053%2C2505.12490%2C2505.12728%2C2505.13291%2C2505.14030%2C2505.14454%2C2505.14646%2C2505.15801%2C2505.15816%2C2505.17020%2C2505.18746%2C2505.19095%2C2505.19713%2C2505.20023%2C2505.20732%2C2505.20795%2C2505.21055%2C2505.22304%2C2505.22429%2C2505.23450%2C2505.23559%2C2506.00641&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:20:17Z", "role": "primary", "sha256": "39de8f4a9aee74cc0eae64e24efc08ed070714eaaea4a30dbc20cb08397eb9bd", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/abs/2505.03570v1"}], "id": "arxiv:2505.03570", "identifiers": {"arxiv": "2505.03570"}, "kind": "paper", "last_verified_at": "2026-10-17T11:20:17Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2505.03570"}], "name": "OSUniverse: Benchmark for Multimodal GUI-navigation AI Agents", "notes": "Reviewed against the exact-version official arXiv abstract page; its complete title, ordered author list, visible version, and normalized abstract SHA match the retained snapshot. The original API response body was not locally available. Experimental results and artifact availability have not been independently reproduced.", "published_at": "2025-05-06", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces OSUniverse, a desktop GUI benchmark spanning precise clicks through multistep and multiapplication tasks. It provides both manual scoring and an automated validation mechanism.", "topics": ["evaluation", "cross-app", "long-horizon"], "updated_at": "2025-05-06", "url": "https://arxiv.org/abs/2505.03570"}
{"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": ["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": ["Musen Lin", "Minghao Liu", "Taoran Lu", "Lichen Yuan", "Yiwei Liu", "Haonan Xu", "Yu Miao", "Yuhao Chao", "Zhaojian Li"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "035048459d2b303721ef0738212829d1fa3314ccd0736b6649cc786c6e975ad4", "submitted_at": "2025-09-19T08:09:18Z", "updated_at": "2025-09-19T08:09:18Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2509.15738v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:50Z", "role": "primary", "sha256": "26fb4922c4c0bb5c9069935ed0aef0a9f56938cf8ae1af9724894fa1409955fb", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=600&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2509.15738", "identifiers": {"arxiv": "2509.15738"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:50Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2509.15738"}], "name": "GUI-ReWalk: Massive Data Generation for GUI Agent via Stochastic Exploration and Intent-Aware Reasoning", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-09-19", "relations": [], "review_status": "curated", "scope": "core", "summary": "GUI-ReWalk begins with stochastic interface exploration and then infers goals to generate coherent action trajectories. It supports long-horizon workflows across applications and studies training on the resulting data using grounding and navigation benchmarks.", "topics": ["data-generation", "cross-app", "long-horizon", "planning"], "updated_at": "2025-09-19", "url": "https://arxiv.org/abs/2509.15738"}
{"aliases": [], "authors": ["Jiaxin Ai", "Yukang Feng", "Fanrui Zhang", "Jianwen Sun", "Zizhen Li", "Chuanhao Li", "Yifan Chang", "Wenxiao Wu", "Ruoxi Wang", "Mingliang Zhai", "Kaipeng Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "afab3dcd2444bf8333f2a2fea887819f6fca55c41209d8c449f6211d1c62b819", "submitted_at": "2025-12-30T01:49:46Z", "updated_at": "2025-12-30T01:49:46Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2601.02399v1"}}, "environments": ["windows", "desktop"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:31:52Z", "role": "primary", "sha256": "8623b0ff7ef9d2e8ba484cd67359e7b8817e78e9ca22b8f27432c2533f5b90a0", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2510.01670%2C2510.04673%2C2510.10073%2C2510.12200%2C2510.14621%2C2510.16051%2C2510.17790%2C2510.18488%2C2510.18596%2C2510.19488%2C2510.19949%2C2510.20286%2C2510.22009%2C2510.24168%2C2510.24411%2C2510.24563%2C2511.00810%2C2511.04307%2C2511.07332%2C2511.09157%2C2511.12254%2C2511.13087%2C2511.15567%2C2511.16590%2C2511.19663%2C2511.20597%2C2511.20766%2C2512.08529%2C2512.10371%2C2512.10962%2C2512.12634%2C2512.14014%2C2512.15431%2C2512.16295%2C2512.16501%2C2512.19432%2C2512.22894%2C2512.24965%2C2601.02399%2C2601.06031%2C2601.07181%2C2601.07262%2C2601.07779%2C2601.08406%2C2601.09923%2C2601.11044%2C2601.11868%2C2601.12822%2C2601.13060%2C2601.15876&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:52Z", "role": "primary", "sha256": "8623b0ff7ef9d2e8ba484cd67359e7b8817e78e9ca22b8f27432c2533f5b90a0", "supports": ["name", "kind", "topics"], "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": "2026-10-17T04:06:36Z", "role": "primary", "sha256": "f9a44f8090369a296a584a9b94c4491662ee25bb8f7239ef75dfd8f775be8887", "supports": ["summary", "environments", "scope"], "url": "https://arxiv.org/html/2601.02399v1"}], "id": "arxiv:2601.02399", "identifiers": {"arxiv": "2601.02399"}, "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/2601.02399"}], "name": "ProSoftArena: Benchmarking Hierarchical Capabilities of Multimodal Agents in Professional Software Environments", "notes": "The paper abstract and official full-text sections were reviewed. The full text supports Windows 11 virtual-machine execution, screenshot/state observations, and mouse/keyboard actions. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2025-12-30", "relations": [], "review_status": "curated", "scope": "core", "summary": "ProSoftArena evaluates 436 professional-software tasks across six disciplines and 13 applications in a Windows 11 virtual-machine environment. Its capability hierarchy, execution-based checks, and human-in-the-loop evaluation cover precise GUI operations and multi-software workflows.", "topics": ["evaluation", "industry", "science", "cross-app"], "updated_at": "2025-12-30", "url": "https://arxiv.org/abs/2601.02399"}
{"aliases": [], "authors": ["Yi Qian", "Kunwei Qian", "Xingbang He", "Ligeng Chen", "Jikang Zhang", "Tiantai Zhang", "Haiyang Wei", "Linzhang Wang", "Hao Wu", "Bing Mao"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "21848dea9e000125550fff7fe3844a57bf100215bf9eb800cbdbb612675f7074", "submitted_at": "2026-01-18T10:54:54Z", "updated_at": "2026-07-15T15:32:11Z", "version": 3, "versioned_url": "https://arxiv.org/abs/2601.12349v3"}}, "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:2601.12349", "identifiers": {"arxiv": "2601.12349"}, "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/2601.12349"}], "name": "Mind the Gap: Action Rebinding Attacks against Android 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": "2026-01-18", "relations": [], "review_status": "curated", "scope": "core", "summary": "Studies action-rebinding attacks in which an Android GUI agent’s planned input is redirected to a different application during the observation–action delay. The evaluation examines cross-application execution and the consequences of treating an earlier screen as the current action target.", "topics": ["safety-security", "cross-app", "evaluation"], "updated_at": "2026-07-15", "url": "https://arxiv.org/abs/2601.12349"}
{"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": ["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": ["Jinchao Li", "Yunxin Li", "Chenrui Zhao", "Zhenran Xu", "Baotian Hu", "Min Zhang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "2120860b9c735466a1538ad26c7cda0bc436ab0c6204febdefa08ad66f0036b7", "submitted_at": "2026-04-30T12:13:27Z", "updated_at": "2026-04-30T12:13:27Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2604.27776v1"}}, "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:2604.27776", "identifiers": {"arxiv": "2604.27776"}, "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/2604.27776"}], "name": "WindowsWorld: A Process-Centric Benchmark of Autonomous GUI Agents in Professional Cross-Application Environments", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced. Desktop use is explicit; no additional operating-system claim is inferred solely from the benchmark name.", "published_at": "2026-04-30", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces WindowsWorld for professional workflows spanning multiple desktop applications, with generated tasks refined through human review. Intermediate subgoals support process-oriented evaluation of cross-application reasoning and execution efficiency.", "topics": ["evaluation", "cross-app", "workflow"], "updated_at": "2026-04-30", "url": "https://arxiv.org/abs/2604.27776"}
{"aliases": [], "authors": ["Kean Shi", "Zihang Li", "Tianyi Ma", "Zengji Tu", "Jialong Wu", "Xinbo Xu", "Qingyao Yang", "Ruoyu Wu", "Weichu Xie", "Ming Wu", "Jason Zeng", "Michael Heinrich", "Elvis Zhang", "Liang Chen", "Kuan Li", "Baobao Chang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "c4ee2439dd31a375b6f64cbc7f1d0ff75c87e1346b1627a53db145ae42682f8a", "submitted_at": "2026-05-15T09:35:15Z", "updated_at": "2026-05-24T16:35:56Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2605.15777v2"}}, "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", "topics"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%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": "2026-10-16T06:35:59Z", "role": "primary", "sha256": "8cfab6fca80090adb47c4ef634f89eedd49a6204dbdc2b5b8e0d6f04c32db289", "supports": ["summary", "environments", "scope", "notes"], "url": "https://raw.githubusercontent.com/UniPat-AI/SaaS-Bench/HEAD/README.md"}], "id": "arxiv:2605.15777", "identifiers": {"arxiv": "2605.15777"}, "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/2605.15777"}], "name": "SaaS-Bench: Can Computer-Use Agents Leverage Real-World SaaS to Solve Professional Workflows?", "notes": "The complete retained official abstract and cached official README were reviewed. The README establishes browser execution against running SaaS applications; no repository code was executed and benchmark results were not independently reproduced.", "published_at": "2026-05-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "SaaS-Bench evaluates browser agents on professional workflows across self-hosted applications. Weighted checkpoints measure task completion and partial progress, with application-state checks and configured content judges documented in the official harness.", "topics": ["web-navigation", "evaluation", "long-horizon", "workflow", "cross-app", "state-tracking"], "updated_at": "2026-05-24", "url": "https://arxiv.org/abs/2605.15777"}
{"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": ["Lawrence Keunho Jang", "Andrew Keunwoo Jang", "Jing Yu Koh", "Ruslan Salakhutdinov"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "acc529f96d6df52e3560f66df8e0630dabdae3ab30cadc8b59123f347eba187a", "submitted_at": "2026-06-15T14:08:09Z", "updated_at": "2026-06-15T14:08:09Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.16748v1"}}, "environments": ["desktop", "linux", "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:2606.16748", "identifiers": {"arxiv": "2606.16748"}, "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/2606.16748"}], "name": "MyPCBench: A Benchmark for Personally Intelligent Computer-Use Agents", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-06-15", "relations": [], "review_status": "curated", "scope": "core", "summary": "MyPCBench tests personal-assistant tasks on a Linux desktop with simulated web applications and data for a single fictional persona. A shared computer-and-shell interface enables evaluation of cross-application workflows that depend on personal context and account-like state.", "topics": ["evaluation", "cross-app", "workflow"], "updated_at": "2026-06-15", "url": "https://arxiv.org/abs/2606.16748"}
{"aliases": [], "authors": ["Li Gu", "Zihuan Jiang", "Linqiang Guo", "Zhixiang Chi", "Ziqiang Wang", "Huan Liu", "Yuanhao Yu", "Tse-Hsun Chen", "Yang Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "9bf8b07e08b275d90fcbfc63d867f54f5579509132cc9cf694d01447b7f91b78", "submitted_at": "2026-06-16T02:36:22Z", "updated_at": "2026-06-16T02:36:22Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.19388v1"}}, "environments": ["android"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:33:05Z", "role": "primary", "sha256": "2e5722527f8da4b2cf06372d1e2daccd33ef7fdc7151704b411a0bddd734beae", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2606.10423%2C2606.10522%2C2606.10742%2C2606.11078%2C2606.11520%2C2606.12485%2C2606.12666%2C2606.13239%2C2606.13385%2C2606.14106%2C2606.14832%2C2606.15017%2C2606.15034%2C2606.16659%2C2606.16748%2C2606.16802%2C2606.17321%2C2606.17645%2C2606.17929%2C2606.18448%2C2606.18671%2C2606.18890%2C2606.19388%2C2606.19926%2C2606.19930%2C2606.20363%2C2606.20636%2C2606.20717%2C2606.20724%2C2606.20785%2C2606.21654%2C2606.22864%2C2606.22948%2C2606.23449%2C2606.24515%2C2606.24525%2C2606.25705%2C2606.25760%2C2606.27330%2C2606.27595%2C2606.29445%2C2606.29537%2C2606.29705%2C2606.30697%2C2606.31154%2C2606.31270%2C2606.31410%2C2606.31612%2C2607.00502%2C2607.04425&max_results=50"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:33:05Z", "role": "primary", "sha256": "2e5722527f8da4b2cf06372d1e2daccd33ef7fdc7151704b411a0bddd734beae", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?id_list=2606.10423%2C2606.10522%2C2606.10742%2C2606.11078%2C2606.11520%2C2606.12485%2C2606.12666%2C2606.13239%2C2606.13385%2C2606.14106%2C2606.14832%2C2606.15017%2C2606.15034%2C2606.16659%2C2606.16748%2C2606.16802%2C2606.17321%2C2606.17645%2C2606.17929%2C2606.18448%2C2606.18671%2C2606.18890%2C2606.19388%2C2606.19926%2C2606.19930%2C2606.20363%2C2606.20636%2C2606.20717%2C2606.20724%2C2606.20785%2C2606.21654%2C2606.22864%2C2606.22948%2C2606.23449%2C2606.24515%2C2606.24525%2C2606.25705%2C2606.25760%2C2606.27330%2C2606.27595%2C2606.29445%2C2606.29537%2C2606.29705%2C2606.30697%2C2606.31154%2C2606.31270%2C2606.31410%2C2606.31612%2C2607.00502%2C2607.04425&max_results=50"}], "id": "arxiv:2606.19388", "identifiers": {"arxiv": "2606.19388"}, "kind": "paper", "last_verified_at": "2026-10-16T06:33:05Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2606.19388"}], "name": "Beyond the GUI Paradigm: Do Mobile Agents Need the Phone Screen?", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-06-16", "relations": [], "review_status": "curated", "scope": "adjacent", "summary": "This study compares general coding agents using mobile command-line access with reproducible GUI agents on AndroidWorld and MobileWorld. It also supplies oracle CLI analyses and a separate task suite for operations such as bulk changes, filtering and hidden device state.", "topics": ["evaluation", "automation", "cross-app", "tool-integration"], "updated_at": "2026-06-16", "url": "https://arxiv.org/abs/2606.19388"}
{"aliases": [], "authors": ["Yikun Fu", "Bowen Fu", "Zhenyu Wu", "Shuang Cheng", "Xiaowei Sun", "Bowen Yang", "Zehao Li", "Yibo Zhao", "Zichen Ding", "Zhoumianze Liu", "Shijie Wang", "Biqing Qi", "Bowen Zhou"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "6dc52f58bb2237d12041fa8b994b20bd1e41290aaab0c70d2c5dd79fcaca37b6", "submitted_at": "2026-06-21T15:32:57Z", "updated_at": "2026-06-21T15:32:57Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2606.22557v1"}}, "environments": ["desktop", "macos"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:32:03Z", "role": "primary", "sha256": "2adad07a9ec7189aea25e48c76fa28206eda97f8974d5745cadda9529103bb1a", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2605.06761%2C2605.07110%2C2605.07134%2C2605.09497%2C2605.11212%2C2605.12481%2C2605.12501%2C2605.12549%2C2605.13631%2C2605.14290%2C2605.14747%2C2605.15030%2C2605.15542%2C2605.15777%2C2605.16024%2C2605.16116%2C2605.16402%2C2605.17320%2C2605.18048%2C2605.18652%2C2605.18758%2C2605.19260%2C2605.19769%2C2605.21470%2C2605.25160%2C2605.25624%2C2605.25707%2C2605.26114%2C2605.27761%2C2605.28116%2C2605.28534%2C2605.28646%2C2605.28775%2C2605.29119%2C2605.29324%2C2605.29447%2C2605.29486%2C2605.29534%2C2605.29615%2C2606.06322%2C2606.09426%2C2606.11042%2C2606.22557%2C2606.23189%2C2606.24551%2C2608.17659%2C2609.22000%2C2609.34139&max_results=48"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:32:03Z", "role": "primary", "sha256": "2adad07a9ec7189aea25e48c76fa28206eda97f8974d5745cadda9529103bb1a", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?id_list=2605.06761%2C2605.07110%2C2605.07134%2C2605.09497%2C2605.11212%2C2605.12481%2C2605.12501%2C2605.12549%2C2605.13631%2C2605.14290%2C2605.14747%2C2605.15030%2C2605.15542%2C2605.15777%2C2605.16024%2C2605.16116%2C2605.16402%2C2605.17320%2C2605.18048%2C2605.18652%2C2605.18758%2C2605.19260%2C2605.19769%2C2605.21470%2C2605.25160%2C2605.25624%2C2605.25707%2C2605.26114%2C2605.27761%2C2605.28116%2C2605.28534%2C2605.28646%2C2605.28775%2C2605.29119%2C2605.29324%2C2605.29447%2C2605.29486%2C2605.29534%2C2605.29615%2C2606.06322%2C2606.09426%2C2606.11042%2C2606.22557%2C2606.23189%2C2606.24551%2C2608.17659%2C2609.22000%2C2609.34139&max_results=48"}], "id": "arxiv:2606.22557", "identifiers": {"arxiv": "2606.22557"}, "kind": "paper", "last_verified_at": "2026-10-16T06:32:03Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2606.22557"}], "name": "MacAgentBench: Benchmarking AI Agents on Real-World macOS Desktop", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-06-21", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces macOS tasks spanning GUI and command-line interaction, with deterministic checks and intermediate-progress scoring for application workflows.", "topics": ["evaluation", "cross-app", "tool-integration"], "updated_at": "2026-06-21", "url": "https://arxiv.org/abs/2606.22557"}
{"aliases": [], "authors": ["Anmol Goel", "Iryna Gurevych"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "da8190bc5afc0bae400fc8ab5e7d80a7f34488c3400dce4477ca61635b19d5b7", "submitted_at": "2026-06-22T11:36:58Z", "updated_at": "2026-09-10T18:49:54Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2606.23189v2"}}, "environments": ["unspecified"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:36Z", "role": "primary", "sha256": "aeacc1a41bb90f3d296823356174fe77d1547ffbe7cd56652cf9d98fca09206c", "supports": ["name", "kind", "summary", "environments", "topics", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=200&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2606.23189", "identifiers": {"arxiv": "2606.23189"}, "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/2606.23189"}], "name": "Capable but Careless: Do Computer-Use Agents Follow Contextual Integrity?", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-06-22", "relations": [], "review_status": "curated", "scope": "core", "summary": "AgentCIBench evaluates whether computer-use agents disclose information in ways that violate its original context. Deterministically scored scenarios test nearby sensitive UI content, ambiguous requests and recipient mismatches, including end-to-end agent execution.", "topics": ["evaluation", "safety-security", "cross-app"], "updated_at": "2026-09-10", "url": "https://arxiv.org/abs/2606.23189"}
{"aliases": [], "authors": ["Zejun Xu", "Taiyi Chen", "Jin Li", "Yongtong Gu", "Qi Cheng", "Aixuan Lv", "Shuai Zhu", "Pengfei Zhu", "Kaichen Yang", "Boyu Sun", "Yixian Yang", "Mulong Xie", "Xin Liu", "Dagang Li", "Xiaoteng Ma", "Hongru Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "d0e469cd7d84080e11df46d826460bea79eae71b8fa5786cacd96aac02b8b0a8", "submitted_at": "2026-08-09T01:08:19Z", "updated_at": "2026-08-09T01:08:19Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2608.08392v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:32Z", "role": "primary", "sha256": "13d3fb5811d203f2e6b80eb45d30ebc0b3cfefd88e2034ffa3b1b80687bef285", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=100&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2608.08392", "identifiers": {"arxiv": "2608.08392"}, "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/2608.08392"}], "name": "CAP: A Scalable Benchmark for Evaluating Cross-Site Browser Agents with Complex Actions and Perception", "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": "2026-08-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "CAP builds cross-site browser workflows by decomposing websites into operation and perception requirements and recomposing them into tasks. Its judge-based evaluation diagnoses complex interactions and visual-perception failures across real websites.", "topics": ["evaluation", "web-navigation", "cross-app", "screen-understanding"], "updated_at": "2026-08-09", "url": "https://arxiv.org/abs/2608.08392"}
{"aliases": [], "authors": ["Zixiang Chen", "Yuheng Lu", "Zihao Cheng", "Zeming Liu", "Jizeng Bai", "Ziye Huang", "Zhiyin Lin", "Zihan Li", "Yuhang Guo", "Yunhong Wang", "Haifeng Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "5ec17516bce87a8e349983acac14415a6ceeb12c167fce57aa37d6157cd3cd83", "submitted_at": "2026-09-09T16:59:59Z", "updated_at": "2026-09-09T16:59:59Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.10451v1"}}, "environments": ["desktop", "android", "windows", "linux"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2609.10451", "identifiers": {"arxiv": "2609.10451"}, "kind": "paper", "last_verified_at": "2026-10-16T06:36:29Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://arxiv.org/abs/2609.10451"}], "name": "JarvisGUI: Towards Cross-Device GUI Agents with Dynamic Task Composition", "notes": "The annotation and scope were reviewed against the complete retained original arXiv abstract. Experiments, reported performance and artifact availability have not been independently reproduced.", "published_at": "2026-09-09", "relations": [], "review_status": "curated", "scope": "core", "summary": "JarvisGUI composes typed input-output task transformations into dynamic workflows spanning Android, Windows and Ubuntu devices. Its virtual-environment benchmark tests transfer of intermediate results, shared state and cross-device dependencies.", "topics": ["evaluation", "cross-app", "planning", "long-horizon", "state-tracking"], "updated_at": "2026-09-09", "url": "https://arxiv.org/abs/2609.10451"}
{"aliases": [], "authors": ["Mei Wu", "Rui Xie", "Runyu Zhang", "Yuqiang Li", "Tianfan Fu", "Bo Chen", "Kai Yu", "Xin Chen", "Lu Chen"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "e032f67257a52dbc59310d921d802e4dc85af3af110d22bb6d005d202bbaa6b2", "submitted_at": "2026-09-29T08:58:51Z", "updated_at": "2026-09-29T08:58:51Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2609.37053v1"}}, "environments": ["desktop", "windows"], "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:2609.37053", "identifiers": {"arxiv": "2609.37053"}, "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/2609.37053"}], "name": "MatToolBench: Benchmarking Multimodal Agents in Real-World Materials Science Workflows", "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 MatToolBench, a Windows-based materials-science benchmark combining GUI operation, scientific-software scripting, and database queries. Expert-defined subcriteria support partial-credit scoring, while the evaluation examines domain knowledge gaps and cross-tool handoff failures.", "topics": ["science", "evaluation", "cross-app"], "updated_at": "2026-09-29", "url": "https://arxiv.org/abs/2609.37053"}
{"aliases": [], "authors": ["Hongcheng Gao", "Hailong Qu", "Yu Lei", "Henghui Sun", "Haoyang Li", "Yipeng Wei", "Naihao Xue", "Xiaohan Yu", "Zhuo Tao", "Yihe Zang", "Yajiao Wang", "Jingyi Tang", "Yi Li", "Jingjing Zhou", "Jie Luo", "Bohan Zeng", "Chengyu Shen", "Hao Jiang", "Chong Chen", "Bowen Qu", "Olive Huang", "Zeqiang Wang"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "aab7f586cb58a1f2d20267fd810ba34d6d4d166e3da715d2b3ff0b77e464473e", "submitted_at": "2026-09-29T14:32:35Z", "updated_at": "2026-10-09T08:15:21Z", "version": 2, "versioned_url": "https://arxiv.org/abs/2609.37686v2"}}, "environments": ["desktop"], "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:2609.37686", "identifiers": {"arxiv": "2609.37686"}, "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/2609.37686"}], "name": "EngiWorld: What Can Frontier Agents Deliver in Professional Engineering Environments?", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-09-29", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces professional engineering tasks across software domains, with GUI and CLI access and artifact checks for geometric, physical, and design validity.", "topics": ["evaluation", "science", "cross-app"], "updated_at": "2026-10-09", "url": "https://arxiv.org/abs/2609.37686"}
{"aliases": [], "authors": ["Xing Han Lù", "Dheeraj Vattikonda", "Sina Hajimiri", "Fatemeh Pesaran Zadeh", "Parishad BehnamGhader", "Ghazwa Darwiche", "Amirhossein Kazemnejad", "Christopher Pal", "Alexandre Drouin", "Siva Reddy"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "37413f11bf8263f1516a8edca8c594227fcab104c0c767a7daa8eead9645ec34", "submitted_at": "2026-10-08T01:03:48Z", "updated_at": "2026-10-08T01:03:48Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.11050v1"}}, "environments": ["cross-platform"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:29Z", "role": "primary", "sha256": "74e521e46c46489e48f9bad5d774625a85b5ccd32828969fd98e1a4c453b899f", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22%29+AND+submittedDate%3A%5B200001010000+TO+202610160636%5D&start=0&max_results=100&sortBy=submittedDate&sortOrder=descending"}], "id": "arxiv:2610.11050", "identifiers": {"arxiv": "2610.11050"}, "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/2610.11050"}], "name": "AgentHorizon: Evaluating Agentic Judges for Long-Horizon Computer-Use Tasks", "notes": "The annotation and scope were reviewed against the primary paper abstract. Experimental results and released artifacts have not been independently reproduced.", "published_at": "2026-10-08", "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces AgentHorizon, a benchmark for judging long computer-use trajectories using closely related instructions paired with recorded actions and screenshots. It tests whether judges can detect constraint violations and distinguish successful tasks from similar but incompatible requests.", "topics": ["evaluation", "long-horizon", "cross-app"], "updated_at": "2026-10-08", "url": "https://arxiv.org/abs/2610.11050"}
{"aliases": [], "authors": ["Songping Wang", "Borui Li", "Qinglong Liu", "Yueming Lyu", "Yi Huang", "Fengting Li", "Chao Feng", "Caifeng Shan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "ea753c1cb0666950d89f2781511f26872e3d5b1c2fd2434896248214160fa22c", "submitted_at": "2026-10-16T11:06:37Z", "updated_at": "2026-10-16T11:06:37Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.23303v1"}}, "environments": ["web"], "evidence": [{"method": "arxiv_api", "retrieved_at": "2026-10-20T05:52:01Z", "role": "primary", "sha256": "63e37cd26bfe69455a3fa835f3d6ba4cfcc5bb7dc91f90505920d37b1e08f0b6", "supports": ["name", "authors", "published_at", "updated_at", "identifiers.arxiv", "details.arxiv_snapshot"], "url": "https://export.arxiv.org/api/query?id_list=2610.19384%2C2610.22672%2C2610.22697%2C2610.23303%2C2610.23307%2C2610.24729&max_results=6"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T08:27:26.236797+00:00", "role": "primary", "sha256": "80bed80786f939973d0d13b4f892300fee02a6110e326247131fc83d257796de", "supports": ["name", "kind", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agent%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22+OR+ti%3A%22browser+agents%22+OR+ti%3A%22mobile+agent%22+OR+ti%3A%22mobile+agents%22+OR+ti%3A%22desktop+agent%22%29+AND+submittedDate%3A%5B202610130000+TO+202610192359%5D&start=0&max_results=200&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T10:42:09.967883+00:00", "role": "primary", "sha256": "ec9d647818be346ddbf7807be87b4789cd53e50c5d3d7a7904ad4f49bc71c38e", "supports": ["summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/html/2610.23303v1"}], "id": "arxiv:2610.23303", "identifiers": {"arxiv": "2610.23303"}, "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/2610.23303"}], "name": "MobileJailBench: Red Queen Jailbreak Benchmarking for Mobile Agents", "notes": "Reviewed the complete retained abstract and selected exact-v1 threat-model, construction and evaluation sections. Execution uses a browser-based Android-like simulator, not real Android isolation or live services. The attacker controls instructions and designated carriers, not privileged simulator state/APIs or system/model/harness internals. Rates depend on release, budget and adjudication coverage. SAF labels track observable expressions and actions, not latent reasoning; unresolved episodes are excluded from rate denominators. No independent reproduction.", "published_at": "2026-10-16", "relations": [], "review_status": "curated", "scope": "core", "summary": "MobileJailBench evaluates mobile-agent jailbreaks in a browser-based Android-like simulator, combining feedback-driven attack evolution, separately admitted task expansions and evidence-grounded outcome adjudication.", "topics": ["safety-security", "mobile-navigation", "evaluation", "observability", "cross-app"], "updated_at": "2026-10-16", "url": "https://arxiv.org/abs/2610.23303"}
{"aliases": [], "authors": ["Songping Wang", "Peiyang Li", "Yi Huang", "Fengting Li", "Xuhao Ran", "Yueming Lyu", "Chao Feng", "Caifeng Shan"], "details": {"arxiv_snapshot": {"abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1", "abstract_sha256": "3fc22fbeda19ac2947b00af73801f0674c550e0d7c78919f811cae754b06128c", "submitted_at": "2026-10-16T11:10:30Z", "updated_at": "2026-10-16T11:10:30Z", "version": 1, "versioned_url": "https://arxiv.org/abs/2610.23307v1"}}, "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": "2026-10-20T08:27:26.236797+00:00", "role": "primary", "sha256": "80bed80786f939973d0d13b4f892300fee02a6110e326247131fc83d257796de", "supports": ["name", "kind", "scope"], "url": "https://export.arxiv.org/api/query?search_query=%28ti%3A%22computer+use%22+OR+ti%3A%22computer-use%22+OR+ti%3A%22GUI+agent%22+OR+ti%3A%22GUI+agents%22+OR+ti%3A%22GUI+grounding%22+OR+ti%3A%22web+agent%22+OR+ti%3A%22web+agents%22+OR+ti%3A%22browser+agent%22+OR+ti%3A%22browser+agents%22+OR+ti%3A%22mobile+agent%22+OR+ti%3A%22mobile+agents%22+OR+ti%3A%22desktop+agent%22%29+AND+submittedDate%3A%5B202610130000+TO+202610192359%5D&start=0&max_results=200&sortBy=submittedDate&sortOrder=descending"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-20T09:17:36.223826+00:00", "role": "primary", "sha256": "3803984a185b6258276b87b87ac1991583878fb41dc22618ff743f6f69c7bc7e", "supports": ["summary", "environments", "topics", "scope", "notes"], "url": "https://arxiv.org/html/2610.23307v1"}], "id": "arxiv:2610.23307", "identifiers": {"arxiv": "2610.23307"}, "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/2610.23307"}], "name": "MobileInject: Evaluating Reasoning-to-Action Security of MLLM-Based Mobile Agents under Cross-Modal Consistency Injection", "notes": "Reviewed the complete abstract and selected v1 construction, threat-model and evaluation sections. The attacker modifies observation surfaces, not system instructions, model parameters or the OS. State oracles are distinct from emitted-rationale diagnoses, whose labels include outcome-dependent constraints. Those diagnostics are not independent evidence of hidden-reasoning causality. No reproduction or release verification.", "published_at": "2026-10-16", "relations": [], "review_status": "curated", "scope": "core", "summary": "MobileInject evaluates visual/textual observation injection in stateful Android workflows using resettable emulators, benign/adversarial state oracles and trace-based diagnosis of emitted rationales and actions.", "topics": ["safety-security", "mobile-navigation", "evaluation", "observability", "cross-app"], "updated_at": "2026-10-16", "url": "https://arxiv.org/abs/2610.23307"}
{"aliases": [], "authors": [], "details": {"component_path": "PC-Agent", "license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/LICENSE"], "reason": "Reviewed repository license declaration for code; third-party dependencies, data, weights and executable environment images are separate."}}, "repository_tree_sha": "11cea575561fb7800b5fb6b6cafa56f7a91de11f", "repository_url": "https://github.com/X-PLUG/MobileAgent"}, "environments": ["desktop", "windows", "macos"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:27:15Z", "role": "primary", "sha256": "ca5b488f10c229b79cc6f9e1f32007d7d84c34366492a10b90d0d66622e77273", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations"], "url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/PC-Agent/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T15:37:03Z", "role": "primary", "sha256": "d403251020124bebcd9f180f800c8ea9f18a3f117c82fcf04c3cbe9ffe0e35f9", "supports": ["identity"], "url": "https://api.github.com/repos/X-PLUG/MobileAgent"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:25:21Z", "role": "primary", "sha256": "7293efadde7c96b64ff8483b2157241e17d0b0fbfff4839de9c7b11485b07cbb", "supports": ["identity", "release_status", "details.repository_tree_sha", "relations"], "url": "https://api.github.com/repos/X-PLUG/MobileAgent/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:42:03Z", "role": "primary", "sha256": "961819a89dacd1f4f161e1d2ebf616b538bf095a2eecfa3ef40b7ccec633fac4", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/LICENSE"}], "id": "component:x-plug/mobileagent/pc-agent", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-18T04:27:15Z", "licenses": {"code": {"basis": "Reviewed repository license declaration for code; third-party dependencies, data, weights and executable environment images are separate.", "declared": "MIT", "retrieved_at": "2026-10-16T06:42:03Z", "url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/X-PLUG/MobileAgent/tree/main/PC-Agent"}], "name": "PC-Agent (X-PLUG component)", "notes": "Scoped to PC-Agent in the MobileAgent monorepo and distinct from GAIR-NLP/PC-Agent. The source tree contains current and v1 implementations; the README states a March 2025 code update. Root MIT covers repository code; the separately linked PC-Eval tasks and application software are not assigned that license by this record.", "published_at": null, "relations": [{"evidence_url": "https://api.github.com/repos/X-PLUG/MobileAgent/git/trees/main?recursive=1", "target_id": "github:x-plug/mobileagent", "type": "component_of"}, {"evidence_url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/PC-Agent/README.md", "target_id": "arxiv:2502.14282", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Multi-agent desktop automation component for Windows and macOS, combining active perception, task decomposition and reflection for workflows spanning applications.", "topics": ["planning", "cross-app", "self-reflection"], "updated_at": null, "url": "https://github.com/X-PLUG/MobileAgent/tree/main/PC-Agent"}
{"aliases": [], "authors": [], "details": {"availability": "Public desktop data-generation source; separate training-data and trained-model releases not verified.", "license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/bytedance/GUI-ReWalk/HEAD/LICENSE"], "reason": "Repository code license; no scope transfer to generated trajectories, pretrained detector weights, or GUI application content."}}, "repository_revision": "78a65d003d6f48802fac879508b62f1c89125892"}, "environments": ["desktop", "linux"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:12:15Z", "role": "primary", "sha256": "63708cb2500aefa36916f3fb171b595efaabf614f17b925cfe8b5c8cbf72be10", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations", "details"], "url": "https://raw.githubusercontent.com/bytedance/GUI-ReWalk/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:12:15Z", "role": "primary", "sha256": "34addf6eed41e1e29e51aab8819c5ac316b5699a8c11c4ce29a32c8e38557de8", "supports": ["identity"], "url": "https://api.github.com/repos/bytedance/GUI-ReWalk"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:14:07Z", "role": "primary", "sha256": "a218cb6f8a2c2a243e8207785c98255bf717c00f177521368705d8652e229d7d", "supports": ["release_status"], "url": "https://api.github.com/repos/bytedance/GUI-ReWalk/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:14:08Z", "role": "primary", "sha256": "43070e2d4e532684de521b885f385d0841030efa2b1a20bafb76133a5e1379c1", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/bytedance/GUI-ReWalk/HEAD/LICENSE"}], "id": "github:bytedance/gui-rewalk", "identifiers": {}, "kind": "framework", "last_verified_at": "2026-10-17T09:12:15Z", "licenses": {"code": {"basis": "Repository code license; no scope transfer to generated trajectories, pretrained detector weights, or GUI application content.", "declared": "Apache-2.0", "retrieved_at": "2026-10-17T09:14:08Z", "url": "https://raw.githubusercontent.com/bytedance/GUI-ReWalk/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/bytedance/GUI-ReWalk"}], "name": "GUI-ReWalk", "notes": "The inspected source tree contains the random walker, desktop environment integration and run_rewalk.sh. The documented runnable setup uses OSWorld and OmniParser. The README also discusses mobile data and a GUI-ReWalk-7B model, but no separately published corpus or model checkpoint was verified; those are not counted as released assets. Apache-2.0 is scoped to the implementation.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/bytedance/GUI-ReWalk/HEAD/README.md", "target_id": "github:xlang-ai/osworld", "type": "uses"}, {"evidence_url": "https://raw.githubusercontent.com/bytedance/GUI-ReWalk/HEAD/README.md", "target_id": "github:microsoft/omniparser", "type": "uses"}, {"evidence_url": "https://raw.githubusercontent.com/bytedance/GUI-ReWalk/HEAD/README.md", "target_id": "arxiv:2509.15738", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "A desktop trajectory-generation framework that moves from stochastic GUI exploration to inferred goals and multi-stage task completion, with retrospective reasoning annotations.", "topics": ["data-generation", "planning", "cross-app"], "updated_at": null, "url": "https://github.com/bytedance/GUI-ReWalk"}
{"aliases": [], "authors": [], "details": {"benchmark_protocol": {"actions": ["Python-defined environment actions"], "counts": [{"basis": "reported", "evidence_url": "https://export.arxiv.org/api/query?id_list=2406.12814%2C2406.13719%2C2406.14056%2C2406.14250%2C2406.14314%2C2406.18082%2C2406.19263%2C2407.00993%2C2407.01476%2C2407.01489%2C2407.01511%2C2407.01558%2C2407.03037%2C2407.03913%2C2407.04346%2C2407.05291%2C2407.07061%2C2407.08850%2C2407.09018%2C2407.09295%2C2407.10956%2C2407.13032%2C2407.13063%2C2407.15673%2C2407.15711%2C2407.16741%2C2407.17490%2C2407.18901%2C2407.19056%2C2407.20183%2C2408.00203%2C2408.02544%2C2408.06327%2C2408.07199%2C2408.11824%2C2408.15978%2C2409.02453%2C2409.04566%2C2409.07429%2C2409.08264%2C2409.09354%2C2409.10741%2C2409.11295%2C2409.13711%2C2409.14337%2C2409.14818%2C2409.15441%2C2409.15637%2C2409.17140%2C2409.20566&max_results=50", "population": "CRAB Benchmark-v0 (paper)", "qualifier": "exact", "reference_url": "https://arxiv.org/abs/2407.01511", "unit": "tasks", "value": 120}], "evaluation": "Graph-based evaluators measure progress through task requirements.", "evidence_urls": ["https://raw.githubusercontent.com/camel-ai/crab/HEAD/README.md", "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"], "focus": ["cross-environment tasks"], "notes": "CRAB is also an extensible framework. The 120-task count belongs to Benchmark-v0, not every environment that can be created with the framework.", "observations": [], "setting": "interactive"}}, "environments": ["desktop", "android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:35:58Z", "role": "primary", "sha256": "7b894c5474ba78294427cfd4ef0eea8cb22d76dfa0c2e8c7f01e31ad927064e9", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details.benchmark_protocol"], "url": "https://raw.githubusercontent.com/camel-ai/crab/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:32Z", "role": "primary", "sha256": "b9a83994d8ac36ccdb606a5219511f8bb35ad3948ade4f9f3e28d386221aedfb", "supports": ["details.benchmark_protocol"], "url": "https://export.arxiv.org/api/query?id_list=2406.12814%2C2406.13719%2C2406.14056%2C2406.14250%2C2406.14314%2C2406.18082%2C2406.19263%2C2407.00993%2C2407.01476%2C2407.01489%2C2407.01511%2C2407.01558%2C2407.03037%2C2407.03913%2C2407.04346%2C2407.05291%2C2407.07061%2C2407.08850%2C2407.09018%2C2407.09295%2C2407.10956%2C2407.13032%2C2407.13063%2C2407.15673%2C2407.15711%2C2407.16741%2C2407.17490%2C2407.18901%2C2407.19056%2C2407.20183%2C2408.00203%2C2408.02544%2C2408.06327%2C2408.07199%2C2408.11824%2C2408.15978%2C2409.02453%2C2409.04566%2C2409.07429%2C2409.08264%2C2409.09354%2C2409.10741%2C2409.11295%2C2409.13711%2C2409.14337%2C2409.14818%2C2409.15441%2C2409.15637%2C2409.17140%2C2409.20566&max_results=50"}], "id": "github:camel-ai/crab", "identifiers": {}, "kind": "benchmark", "last_verified_at": "2026-10-16T06:35:58Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/camel-ai/crab"}], "name": "CRAB", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/camel-ai/crab/HEAD/README.md", "target_id": "arxiv:2407.01511", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Cross-environment benchmark framework for multimodal agents.", "topics": ["evaluation", "cross-app"], "updated_at": null, "url": "https://github.com/camel-ai/crab"}
{"aliases": [], "authors": [], "details": {"benchmark_protocol": {"actions": ["click", "type", "scroll", "configured BrowserGym/AgentLab actions"], "counts": [], "evaluation": "Tasks define check_if_task_is_complete(initial_state,current_state); app-specific reward logic evaluates underlying state, with custom checks for code-editor file contents. Observation/history flags are configurable. No universal single fixed task total or published-run reproduction is implied.", "evidence_urls": ["https://raw.githubusercontent.com/facebookresearch/OpenApps/HEAD/README.md", "https://raw.githubusercontent.com/facebookresearch/OpenApps/HEAD/docs/tasks.md", "https://raw.githubusercontent.com/facebookresearch/OpenApps/HEAD/docs/agents.md", "https://api.github.com/repos/facebookresearch/OpenApps/git/trees/main?recursive=1"], "focus": ["UI-agent reliability under environment variation", "state-based browser task evaluation"], "notes": "Preserve the code/tag, app state, theme/layout/content choices, goal variant and action/observation configuration. Current main refactors the paper-era appearance surface, and the v1.0-paper tag includes post-publication fixes that may move rewards. The environments run as browser apps; phone/tablet viewports are not native Android/iOS environments.", "observations": ["screenshots", "accessibility trees", "set-of-marks and coordinates (configurable)"], "setting": "interactive"}, "license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/facebookresearch/OpenApps/HEAD/LICENSE"], "reason": "Reviewed repository license declaration for code; third-party dependencies, data, weights and executable environment images are separate."}}, "repository_tree_sha": "6e6782794433c60468d65ac75bc3f03c56007f8c", "repository_url": "https://github.com/facebookresearch/OpenApps"}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:35:18Z", "role": "primary", "sha256": "1df0f66584b37e7dba90811a7732d3b0c13c96cd7da4aec3970a18be36a6cec1", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations", "details.benchmark_protocol"], "url": "https://raw.githubusercontent.com/facebookresearch/OpenApps/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:33:19Z", "role": "primary", "sha256": "2c57b627cb344e360480c10e74dcdd1d1ed321d9a7a4e3891bc970355b15f4cd", "supports": ["identity"], "url": "https://api.github.com/repos/facebookresearch/OpenApps"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:35:18Z", "role": "primary", "sha256": "34aa76fbd6cca6ac2f3949792144662b90d85d7a18988afabcb570167ea5c1d6", "supports": ["identity", "release_status", "details.repository_tree_sha", "details.benchmark_protocol"], "url": "https://api.github.com/repos/facebookresearch/OpenApps/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:40:40Z", "role": "primary", "sha256": "af030b93a7604c063b08f003b0a0124a6e95e560649d0306bfe40dd87fed563d", "supports": ["details.benchmark_protocol"], "url": "https://raw.githubusercontent.com/facebookresearch/OpenApps/HEAD/docs/tasks.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:40:40Z", "role": "primary", "sha256": "aede4dfd965bd3484336f40ddb1bd6db48aed00ab45e27c048b76cc4601d71b3", "supports": ["details.benchmark_protocol"], "url": "https://raw.githubusercontent.com/facebookresearch/OpenApps/HEAD/docs/agents.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:35:20Z", "role": "primary", "sha256": "336255dc30193e8e15d689d9481bb05673d89055718f3a96923a7ffb99adbbaf", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/facebookresearch/OpenApps/HEAD/LICENSE"}], "id": "github:facebookresearch/openapps", "identifiers": {}, "kind": "benchmark", "last_verified_at": "2026-10-18T04:35:18Z", "licenses": {"code": {"basis": "Reviewed repository license declaration for code; third-party dependencies, data, weights and executable environment images are separate.", "declared": "CC-BY-NC-4.0", "retrieved_at": "2026-10-18T04:35:20Z", "url": "https://raw.githubusercontent.com/facebookresearch/OpenApps/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/facebookresearch/OpenApps"}], "name": "OpenApps", "notes": "Public code includes app/task/configuration and agent-launching implementations. README distinguishes moving main from v1.0-paper and explicitly says even that tag is not a byte-exact snapshot of the published runs. CC-BY-NC-4.0 is the repository’s declared code license; third-party WebShop, map and icon content retain separate terms. No commercial-use permission or fixed downloadable task total is inferred.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/facebookresearch/OpenApps/HEAD/README.md", "target_id": "arxiv:2511.20766", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Configurable browser apps and task rewards for testing computer-use agents under changes in content, layout and appearance.", "topics": ["evaluation", "robustness", "cross-app", "sandbox"], "updated_at": null, "url": "https://github.com/facebookresearch/OpenApps"}
{"aliases": [], "authors": [], "details": {}, "environments": ["desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:03Z", "role": "primary", "sha256": "178774927f003b5221d695fa78ded6fcc543aa8653c7b2dd8016418e6911afb7", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/GAIR-NLP/PC-Agent/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T06:42:04Z", "role": "primary", "sha256": "aee22091f60f10968afb7dc33dd8b728ea929e287c260ed82adffdec4866fcb5", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/GAIR-NLP/PC-Agent/HEAD/LICENSE"}], "id": "github:gair-nlp/pc-agent", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-16T06:36:03Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "MIT", "retrieved_at": "2026-10-16T06:42:04Z", "url": "https://raw.githubusercontent.com/GAIR-NLP/PC-Agent/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/GAIR-NLP/PC-Agent"}], "name": "PC-Agent", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/GAIR-NLP/PC-Agent/HEAD/README.md", "target_id": "arxiv:2412.17589", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Human-cognition-transfer framework for PC agents: PC Tracker logs human computer use, cognition completion turns logs into cognitive trajectories, and a planner works with a grounding agent.", "topics": ["planning", "cross-app"], "updated_at": null, "url": "https://github.com/GAIR-NLP/PC-Agent"}
{"aliases": [], "authors": [], "details": {"benchmark_protocol": {"actions": ["PyAutoGUI"], "counts": [{"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/GAIR-NLP/WindowsAgentArena-V2/HEAD/README.md", "population": "full WindowsAgentArena-V2 suite reported by README", "qualifier": "exact", "unit": "tasks", "value": 141}, {"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/GAIR-NLP/WindowsAgentArena-V2/HEAD/README.md", "population": "full suite coverage reported by README", "qualifier": "exact", "unit": "Windows applications", "value": 11}, {"basis": "computed", "evidence_url": "https://raw.githubusercontent.com/GAIR-NLP/WindowsAgentArena-V2/HEAD/src/win-arena-container/client/evaluation_examples_windows/test_custom.json", "population": "shipped test_custom.json selection", "qualifier": "exact", "unit": "task IDs", "value": 20}], "evaluation": "Runner exposes four observation choices and uses PyAutoGUI. count.py sums the last score in each results.json and divides by the number of traj.jsonl files, multiplying by 100; it separately reports result/trajectory/error-file coverage. This denominator is not automatically the full benchmark size.", "evidence_urls": ["https://raw.githubusercontent.com/GAIR-NLP/WindowsAgentArena-V2/HEAD/README.md", "https://raw.githubusercontent.com/GAIR-NLP/WindowsAgentArena-V2/HEAD/src/win-arena-container/client/run.py", "https://raw.githubusercontent.com/GAIR-NLP/WindowsAgentArena-V2/HEAD/src/win-arena-container/client/evaluation_examples_windows/test_custom.json", "https://raw.githubusercontent.com/GAIR-NLP/WindowsAgentArena-V2/HEAD/count.py", "https://api.github.com/repos/GAIR-NLP/WindowsAgentArena-V2/git/trees/main?recursive=1"], "focus": ["Windows desktop task execution", "revised task evaluators"], "notes": "README sample config uses max_steps=30. The shipped selected-task index has 20 IDs, not 141; preserve the selected task list, trial ID, score coverage and VM snapshot when comparing runs. Docker/Windows-11 setup is required. No VM or benchmark was executed.", "observations": ["screenshot", "accessibility tree", "screenshot plus accessibility tree", "set-of-marks"], "setting": "interactive"}, "license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/GAIR-NLP/WindowsAgentArena-V2/HEAD/LICENSE"], "reason": "Reviewed repository license declaration for code; third-party dependencies, data, weights and executable environment images are separate."}}, "repository_tree_sha": "2927fe55005d1be75d5a9188c0045f73ba28d192", "repository_url": "https://github.com/GAIR-NLP/WindowsAgentArena-V2"}, "environments": ["desktop", "windows"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:35:17Z", "role": "primary", "sha256": "1dd744ce565ec00b20ea5637eb7bc6112e97fb9044bd9a471f7cc4b7227ca344", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations", "details.benchmark_protocol"], "url": "https://raw.githubusercontent.com/GAIR-NLP/WindowsAgentArena-V2/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:33:20Z", "role": "primary", "sha256": "0b0af02d3c2c4916dd786b868b99d60b4681927d0273a23661eb2b6db7cf9404", "supports": ["identity"], "url": "https://api.github.com/repos/GAIR-NLP/WindowsAgentArena-V2"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:35:16Z", "role": "primary", "sha256": "638ef9ee593a3a560c1d0b3fb2bc4a9d31a3e904c0111971924a6d66e1763aa6", "supports": ["identity", "release_status", "details.repository_tree_sha", "details.benchmark_protocol"], "url": "https://api.github.com/repos/GAIR-NLP/WindowsAgentArena-V2/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:40:39Z", "role": "primary", "sha256": "494de9e74629ac2c55358629c0d9ba13703f36317057a3089388fbb6e80ca28f", "supports": ["details.benchmark_protocol"], "url": "https://raw.githubusercontent.com/GAIR-NLP/WindowsAgentArena-V2/HEAD/src/win-arena-container/client/run.py"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:40:39Z", "role": "primary", "sha256": "e45ba7fdc7555f46ad1b4cfaab79d00754909d053fc49d5bab651b0fb593439b", "supports": ["details.benchmark_protocol"], "url": "https://raw.githubusercontent.com/GAIR-NLP/WindowsAgentArena-V2/HEAD/src/win-arena-container/client/evaluation_examples_windows/test_custom.json"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:40:39Z", "role": "primary", "sha256": "f12147ac78c39c72eb75f9c37b293ba7fff3209fc5758fd51c69abbd96d80a97", "supports": ["details.benchmark_protocol"], "url": "https://raw.githubusercontent.com/GAIR-NLP/WindowsAgentArena-V2/HEAD/count.py"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:35:20Z", "role": "primary", "sha256": "2a2d1e46940b570d60f5e2198ed27fcec6316fc8489e9660da8777dbd75aaeec", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/GAIR-NLP/WindowsAgentArena-V2/HEAD/LICENSE"}], "id": "github:gair-nlp/windowsagentarena-v2", "identifiers": {}, "kind": "benchmark", "last_verified_at": "2026-10-18T04:35:17Z", "licenses": {"code": {"basis": "Reviewed repository license declaration for code; third-party dependencies, data, weights and executable environment images are separate.", "declared": "MIT", "retrieved_at": "2026-10-18T04:35:20Z", "url": "https://raw.githubusercontent.com/GAIR-NLP/WindowsAgentArena-V2/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/GAIR-NLP/WindowsAgentArena-V2"}], "name": "WindowsAgentArena-V2", "notes": "Repository contains executable evaluation code and task definitions. README links a separate Windows 11 VM snapshot; that image and its software licenses/access were not validated by downloading it. MIT is recorded for repository code only. Default task selection and the reported full suite have different scopes.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/GAIR-NLP/WindowsAgentArena-V2/HEAD/README.md", "target_id": "arxiv:2505.13909", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Windows benchmark revision with task/evaluator fixes and VM-initialization support, documented alongside PC-Agent-E.", "topics": ["evaluation", "cross-app", "robustness"], "updated_at": null, "url": "https://github.com/GAIR-NLP/WindowsAgentArena-V2"}
{"aliases": [], "authors": [], "details": {"benchmark_protocol": {"actions": ["PyAutoGUI", "UiPath Computer_13 action interface"], "counts": [{"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/HITsz-TMG/WindowsWorld/HEAD/README.md", "population": "WindowsWorld benchmark described in the retained README", "qualifier": "exact", "unit": "tasks", "value": 181}, {"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/HITsz-TMG/WindowsWorld/HEAD/README.md", "population": "WindowsWorld coverage described in the retained README", "qualifier": "exact", "unit": "desktop applications", "value": 17}], "evaluation": "Publisher distinguishes S_int (intermediate checkpoint score on L1–L3) and S_final (final completion on L1–L4). Documented step budgets are 15/25/40/20 for L1/L2/L3/L4. Preserve level, observation mode, agent interface and checkpoint/final metric when reporting a run.", "evidence_urls": ["https://raw.githubusercontent.com/HITsz-TMG/WindowsWorld/HEAD/README.md", "https://api.github.com/repos/HITsz-TMG/WindowsWorld/git/trees/main?recursive=1"], "focus": ["multi-application Windows workflows", "intermediate-state and final-goal completion"], "notes": "Interactive VMware/Windows environment. The public repository includes benchmark.json, task/evaluator code and monitoring support; the README refers VM import to Installation Guide.md, whose distribution links and payloads were not inspected for this profile. Checkpoint counts and application counts are not extra tasks. Apache-2.0 is recorded for repository code; Windows, apps and VM artifacts retain separate terms.", "observations": ["screenshot", "accessibility tree", "screenshot plus accessibility tree", "set-of-marks"], "setting": "interactive"}}, "environments": ["desktop", "windows"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:35:59Z", "role": "primary", "sha256": "9857bd8ae9f521f152761ae3bd85debe4e7471aad34c9eb47618332948bca6c9", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details.benchmark_protocol"], "url": "https://raw.githubusercontent.com/HITsz-TMG/WindowsWorld/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T14:14:05Z", "role": "primary", "sha256": "490c9257d56f0a5214e6424e02490379b0b00ca2f7d4626fd5d14e17239d05ed", "supports": ["details.benchmark_protocol"], "url": "https://api.github.com/repos/HITsz-TMG/WindowsWorld/git/trees/main?recursive=1"}, {"method": "license_header", "retrieved_at": "2026-10-16T06:41:59Z", "role": "primary", "sha256": "a8d1d88a0f9cce3cdbff73b03a80dc5e5d69e3ba2443946c268815a550c9e0ec", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/HITsz-TMG/WindowsWorld/HEAD/LICENSE"}], "id": "github:hitsz-tmg/windowsworld", "identifiers": {}, "kind": "benchmark", "last_verified_at": "2026-10-16T06:35:59Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "Apache-2.0", "retrieved_at": "2026-10-16T06:41:59Z", "url": "https://raw.githubusercontent.com/HITsz-TMG/WindowsWorld/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/HITsz-TMG/WindowsWorld"}], "name": "WindowsWorld", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/HITsz-TMG/WindowsWorld/HEAD/README.md", "target_id": "arxiv:2604.27776", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Cross-application Windows benchmark with intermediate task evaluation.", "topics": ["evaluation", "cross-app"], "updated_at": null, "url": "https://github.com/HITsz-TMG/WindowsWorld"}
{"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 graph/task manifests, evaluation source and ungated screenshot archive manifest.", "benchmark_protocol": {"actions": ["mobile GUI operations along graph transitions"], "counts": [{"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/MadeAgents/ColorBench/HEAD/README.md", "population": "reported ColorBench benchmark", "qualifier": "exact", "unit": "tasks", "value": 175}, {"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/MadeAgents/ColorBench/HEAD/README.md", "population": "reported ColorBench benchmark", "qualifier": "exact", "unit": "cross-app tasks", "value": 101}, {"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/MadeAgents/ColorBench/HEAD/README.md", "population": "reported ColorBench benchmark", "qualifier": "exact", "unit": "single-app tasks", "value": 74}, {"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/MadeAgents/ColorBench/HEAD/README.md", "population": "reported ColorBench benchmark", "qualifier": "exact", "unit": "mobile applications", "value": 21}], "evaluation": "The publisher reports success rate, completion rate and atomic-capability diagnostics using graph-based evaluation.", "evidence_urls": ["https://raw.githubusercontent.com/MadeAgents/ColorBench/HEAD/README.md"], "focus": ["long-horizon mobile task completion", "multi-path graph interaction"], "notes": "The interaction environment replays graph states rather than running real user accounts. The 101 and 74 subsets partition the reported 175 tasks; do not add the three counts.", "observations": ["mobile screenshots represented as graph states"], "setting": "interactive"}, "license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/MadeAgents/ColorBench/HEAD/LICENSE"], "reason": "Repository source LICENSE."}, "data": {"evidence_urls": ["https://huggingface.co/datasets/Norsheep/ColorBench/raw/main/README.md"], "reason": "Publisher metadata for the separately hosted screenshot archive; underlying third-party content retains applicable rights."}}, "repository_revision": "1f26a84230d26f99d84eab97d05032582b39018e"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:18:16Z", "role": "primary", "sha256": "5c0190fffb68b91c1855a408f6a5e252871d94000b95fa258504a3c100600cec", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations", "details.benchmark_protocol"], "url": "https://raw.githubusercontent.com/MadeAgents/ColorBench/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:18:16Z", "role": "primary", "sha256": "2f6940503cc0d6cb97bdf142d6b8c6ab116de82227f2fa6d97fd56d7e5aebec0", "supports": ["identity"], "url": "https://api.github.com/repos/MadeAgents/ColorBench"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:22:54Z", "role": "primary", "sha256": "6a00c093a6794ec4a172d30b70fdabc17400584a9e25a157a47a7661dbdd20fa", "supports": ["release_status", "details.repository_revision"], "url": "https://api.github.com/repos/MadeAgents/ColorBench/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:23:06Z", "role": "primary", "sha256": "c0c3286b5150e128e55034ba0393ef85f7fcf73e4b73f4276389a04c9bcec919", "supports": ["identity", "release_status"], "url": "https://huggingface.co/api/datasets/Norsheep/ColorBench"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:23:06Z", "role": "primary", "sha256": "4bcf87ecfbbb8e07a01b21415a970c8b53a5283bf6872b657040d3f45c9241f7", "supports": ["licenses.data"], "url": "https://huggingface.co/datasets/Norsheep/ColorBench/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T11:22:54Z", "role": "primary", "sha256": "130976c026bcafa9dcc479697fe6595dbefb73cdb375fc47b8478acc16e6ea73", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/MadeAgents/ColorBench/HEAD/LICENSE"}], "id": "github:madeagents/colorbench", "identifiers": {}, "kind": "benchmark", "last_verified_at": "2026-10-17T11:18:16Z", "licenses": {"code": {"basis": "Repository source LICENSE.", "declared": "Apache-2.0", "retrieved_at": "2026-10-17T11:22:54Z", "url": "https://raw.githubusercontent.com/MadeAgents/ColorBench/HEAD/LICENSE"}, "data": {"basis": "Publisher metadata for the separately hosted screenshot archive; underlying third-party content retains applicable rights.", "declared": "apache-2.0", "retrieved_at": "2026-10-17T11:23:06Z", "url": "https://huggingface.co/datasets/Norsheep/ColorBench/raw/main/README.md"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/MadeAgents/ColorBench"}], "name": "ColorBench", "notes": "The source tree contains graph and task JSON files plus construction/evaluation code; the linked ungated Hub provides the screenshot archive. The environment is static but interactive and graph-based, not a live-device deployment. Counts are publisher-reported task/app inventories. Apache-2.0 declarations for repository code and the Hub dataset are separately retained; third-party application visuals are not granted additional rights by this catalog.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/MadeAgents/ColorBench/HEAD/README.md", "target_id": "arxiv:2510.14621", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "A graph-based mobile GUI benchmark that evaluates long tasks through screenshot states and multiple valid or erroneous transitions, including reflection, replanning and backtracking.", "topics": ["evaluation", "mobile-navigation", "long-horizon", "cross-app"], "updated_at": null, "url": "https://github.com/MadeAgents/ColorBench"}
{"aliases": [], "authors": [], "details": {}, "environments": ["windows", "desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:02Z", "role": "primary", "sha256": "40d9a4010ab122ab50242a0cd7a5320e2fd16b3e53494d988eecdbd94fd8e6e8", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations"], "url": "https://raw.githubusercontent.com/microsoft/UFO/HEAD/README.md"}], "id": "github:microsoft/ufo", "identifiers": {}, "kind": "agent", "last_verified_at": "2026-10-16T06:36:02Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/microsoft/UFO"}], "name": "UFO", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/microsoft/UFO/HEAD/README.md", "target_id": "arxiv:2402.07939", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Windows agent framework for application and desktop automation.", "topics": ["planning", "cross-app"], "updated_at": null, "url": "https://github.com/microsoft/UFO"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://huggingface.co/datasets/hflqf88888/GUIOdyssey/raw/main/README.md", "population": "recommended hflqf88888/GUIOdyssey release", "qualifier": "exact", "unit": "episodes", "value": 8334}], "evidence_urls": ["https://raw.githubusercontent.com/OpenGVLab/GUI-Odyssey/HEAD/README.md", "https://huggingface.co/datasets/hflqf88888/GUIOdyssey/raw/main/README.md", "https://raw.githubusercontent.com/OpenGVLab/GUI-Odyssey/HEAD/introduction.md"], "formats": ["JSON annotations", "PNG screenshots"], "modalities": ["screenshots", "instructions", "actions", "bounding boxes", "step context"], "notes": "The repository gives conflicting episode totals (8,334 and 8,834). This profile uses 8,334 from the specifically recommended current dataset card. No independent archive recount was performed. Coordinates are normalized to 0–1000.", "splits": ["random", "task", "device", "app"], "supervision": ["cross-app trajectories", "action grounding"]}}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:01Z", "role": "primary", "sha256": "4585b765d7c8b3e9bb6187d36d576edde454d4c0158a6aa8b3fdbedf34c89a15", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details.dataset_profile"], "url": "https://raw.githubusercontent.com/OpenGVLab/GUI-Odyssey/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T11:39:52Z", "role": "primary", "sha256": "fd627ee0a6ee0596e0ee8147996b9a7a3436f754fc95ed63cf5f408d778b2831", "supports": ["details.dataset_profile"], "url": "https://huggingface.co/datasets/hflqf88888/GUIOdyssey/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T11:39:52Z", "role": "primary", "sha256": "78c19ee712487af60df1913f11572958564e59a1a2cc663b1b30d971855d7a28", "supports": ["details.dataset_profile"], "url": "https://raw.githubusercontent.com/OpenGVLab/GUI-Odyssey/HEAD/introduction.md"}], "id": "github:opengvlab/gui-odyssey", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T06:36:01Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/OpenGVLab/GUI-Odyssey"}], "name": "GUI-Odyssey", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/OpenGVLab/GUI-Odyssey/HEAD/README.md", "target_id": "arxiv:2406.08451", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Human-demonstrated Android episodes for cross-app navigation, spanning six task types, 212 apps and about 1,400 app combinations, for training and evaluating agents.", "topics": ["cross-app", "data-generation"], "updated_at": null, "url": "https://github.com/OpenGVLab/GUI-Odyssey"}
{"aliases": [], "authors": [], "details": {"benchmark_protocol": {"actions": ["GUI gestures", "foreground-app tool calls in hybrid mode"], "counts": [{"basis": "computed", "evidence_url": "https://raw.githubusercontent.com/OpenMobile2/MobileGym-plusplus/HEAD/bench_env/benchmark/official.json", "population": "official.json", "qualifier": "exact", "unit": "task definitions", "value": 215}, {"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/OpenMobile2/MobileGym-plusplus/HEAD/README.md", "population": "complete simulator installation", "qualifier": "exact", "unit": "simulated clients", "value": 63}, {"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/OpenMobile2/MobileGym-plusplus/HEAD/README.md", "population": "new clients in this release", "qualifier": "exact", "unit": "simulated clients", "value": 35}], "evaluation": "Programmatic checkers inspect changes in application state; both modes share task definitions and initial states.", "evidence_urls": ["https://raw.githubusercontent.com/OpenMobile2/MobileGym-plusplus/HEAD/README.md", "https://raw.githubusercontent.com/OpenMobile2/MobileGym-plusplus/HEAD/bench_env/benchmark/official.json"], "focus": ["task completion", "cross-application workflows", "GUI and app-native tool comparison"], "notes": "The task count was checked against unique task_id values in the published manifest. The 35 new clients are a subset of 63 total clients. This simulator runs in a browser; Android-emulator apps are a separate part of OpenMobile-2.", "observations": ["screenshots"], "setting": "interactive"}}, "environments": ["web"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T01:48:10Z", "role": "primary", "sha256": "d9f850ec196e715650e679cb81515adaa64ddc5a60a882bfe97b0f6a098ddc83", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://raw.githubusercontent.com/OpenMobile2/MobileGym-plusplus/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T01:46:06Z", "role": "primary", "sha256": "9b385e9f6aa2f7169f616a76887df3b76df605cb2bea071d99ee7ca3f9b659cc", "supports": ["relations"], "url": "https://os-copilot.github.io/OpenMobile2-Home/"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T01:50:11Z", "role": "primary", "sha256": "35c1641b349201c68917fbe4c4036814487678e9588d9b1c903af26917800316", "supports": ["details.benchmark_protocol"], "url": "https://raw.githubusercontent.com/OpenMobile2/MobileGym-plusplus/HEAD/bench_env/benchmark/official.json"}, {"method": "license_header", "retrieved_at": "2026-10-17T01:52:51Z", "role": "primary", "sha256": "7618cd7de039c264759ca8ca0b9130c6de63a4536dcf436baedd4cfbf639073e", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/OpenMobile2/MobileGym-plusplus/HEAD/LICENSE"}], "id": "github:openmobile2/mobilegym-plusplus", "identifiers": {}, "kind": "benchmark", "last_verified_at": "2026-10-17T01:48:10Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "Apache-2.0", "retrieved_at": "2026-10-17T01:52:51Z", "url": "https://raw.githubusercontent.com/OpenMobile2/MobileGym-plusplus/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/OpenMobile2/MobileGym-plusplus"}], "name": "MobileGym++ / MobileGym++ Bench", "notes": "These clients simulate named commercial applications and are not those commercial products. The code and simulator data have distinct license files. No benchmark run was performed. The README describes a phone that runs in the browser, and the Android-emulator playground is a separate OpenMobile2 release, so the platform facet is web rather than a mobile operating system.", "published_at": null, "relations": [{"evidence_url": "https://os-copilot.github.io/OpenMobile2-Home/", "target_id": "arxiv:2610.14157", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Browser-rendered mobile simulation with resettable app state and a shared task suite for GUI-only and GUI-plus-tool agents.", "topics": ["mobile-navigation", "evaluation", "cross-app", "tool-integration"], "updated_at": null, "url": "https://github.com/OpenMobile2/MobileGym-plusplus"}
{"aliases": [], "authors": [], "details": {"availability": "Public task definitions, execution harness and an ungated VM-file manifest.", "benchmark_protocol": {"actions": ["actions executed by the OSWorld-derived DesktopEnv"], "counts": [{"basis": "computed", "evidence_url": "https://raw.githubusercontent.com/OS-Copilot/OS-Map/HEAD/evaluation_examples/all.json", "population": "evaluation_examples/all.json", "qualifier": "exact", "unit": "category/task references", "value": 416}], "evaluation": "Task configurations provide evaluators for application/system state, using an OSWorld-derived runtime.", "evidence_urls": ["https://raw.githubusercontent.com/OS-Copilot/OS-Map/HEAD/README.md", "https://raw.githubusercontent.com/OS-Copilot/OS-Map/HEAD/evaluation_examples/all.json", "https://raw.githubusercontent.com/OS-Copilot/OS-Map/HEAD/evaluation_examples/README.md"], "focus": ["breadth of desktop task coverage", "depth of desktop task execution"], "notes": "416 category/task pairs are unique, but only 412 bare ID strings are unique across categories. This is a manifest-reference count, not a claim of 416 semantically distinct tasks. Main runtime documentation specifies Ubuntu; the subdirectory README retains older Windows-only wording.", "observations": ["desktop observations from an Ubuntu virtual machine"], "setting": "interactive"}, "license_review": {"code": {"evidence_urls": ["https://raw.githubusercontent.com/OS-Copilot/OS-Map/HEAD/LICENSE"], "reason": "Repository code LICENSE; no license is inferred for the VM image or bundled third-party applications."}}, "repository_revision": "191efd0e6766acab444a31685b437565e1a94ac2"}, "environments": ["desktop", "linux"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:12:14Z", "role": "primary", "sha256": "fc9f3c8d9fffdff8b51e312b013be0d9fdfc10e57fdc2ec4a8a8d012c4b42475", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "notes", "relations", "details"], "url": "https://raw.githubusercontent.com/OS-Copilot/OS-Map/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:12:14Z", "role": "primary", "sha256": "d6c91b4910e92f40cc694b4cb6467c1a20f42c0c14047b95a1e777aca0892677", "supports": ["identity"], "url": "https://api.github.com/repos/OS-Copilot/OS-Map"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:14:07Z", "role": "primary", "sha256": "41342157c53f28ad2729137491930225ac11da922e7b16762e84ac958a2e9d28", "supports": ["release_status"], "url": "https://api.github.com/repos/OS-Copilot/OS-Map/git/trees/main?recursive=1"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:14:07Z", "role": "primary", "sha256": "1dd924135cbf0a85ce72d75025259adb7b849970d400af4ed854327a34e7ee41", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/OS-Copilot/OS-Map/HEAD/LICENSE"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:17:22Z", "role": "primary", "sha256": "4199af4c8bc60451079d8a2e7598ab7386bd82b0d5c9ad7e6927d6cb8e03ad6f", "supports": ["details.benchmark_protocol"], "url": "https://raw.githubusercontent.com/OS-Copilot/OS-Map/HEAD/evaluation_examples/all.json"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:17:22Z", "role": "primary", "sha256": "a6b60d5721cfdd26416e7d0ae3192360896427a97ccb3793812cc150e9cd7126", "supports": ["details.benchmark_protocol", "notes"], "url": "https://raw.githubusercontent.com/OS-Copilot/OS-Map/HEAD/evaluation_examples/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:14:16Z", "role": "primary", "sha256": "d2da837872b123f2927a069a020aeb14fb4b22148599c9f2e8b6689ee444ec88", "supports": ["release_status", "details"], "url": "https://huggingface.co/api/datasets/os-map/OSMAP"}], "id": "github:os-copilot/os-map", "identifiers": {}, "kind": "benchmark", "last_verified_at": "2026-10-17T09:12:14Z", "licenses": {"code": {"basis": "Repository code LICENSE; no license is inferred for the VM image or bundled third-party applications.", "declared": "MIT", "retrieved_at": "2026-10-17T09:14:07Z", "url": "https://raw.githubusercontent.com/OS-Copilot/OS-Map/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/OS-Copilot/OS-Map"}], "name": "OS-Map", "notes": "Public task manifests, agent runner and evaluator code are present, and the linked ungated Hub manifest lists Ubuntu0.zip and task files. The main setup supports x86 VMware Workstation and says ARM/Docker alternatives are still in development. An inherited evaluation-examples README still mentions Windows 10 and under-construction status; those statements are retained as a documentation inconsistency, not interpreted as verified Windows support. The manifest has 416 category/task references but 412 globally distinct ID strings.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/OS-Copilot/OS-Map/HEAD/README.md", "target_id": "github:xlang-ai/osworld", "type": "uses"}, {"evidence_url": "https://raw.githubusercontent.com/OS-Copilot/OS-Map/HEAD/README.md", "target_id": "arxiv:2507.19132", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "An OSWorld-derived desktop benchmark with an Ubuntu virtual-machine environment and expanded task configurations for measuring breadth and depth of computer interaction.", "topics": ["evaluation", "long-horizon", "cross-app"], "updated_at": null, "url": "https://github.com/OS-Copilot/OS-Map"}
{"aliases": [], "authors": [], "details": {"benchmark_protocol": {"actions": ["Python code", "application API calls"], "counts": [{"basis": "reported", "evidence_url": "https://export.arxiv.org/api/query?id_list=2406.12814%2C2406.13719%2C2406.14056%2C2406.14250%2C2406.14314%2C2406.18082%2C2406.19263%2C2407.00993%2C2407.01476%2C2407.01489%2C2407.01511%2C2407.01558%2C2407.03037%2C2407.03913%2C2407.04346%2C2407.05291%2C2407.07061%2C2407.08850%2C2407.09018%2C2407.09295%2C2407.10956%2C2407.13032%2C2407.13063%2C2407.15673%2C2407.15711%2C2407.16741%2C2407.17490%2C2407.18901%2C2407.19056%2C2407.20183%2C2408.00203%2C2408.02544%2C2408.06327%2C2408.07199%2C2408.11824%2C2408.15978%2C2409.02453%2C2409.04566%2C2409.07429%2C2409.08264%2C2409.09354%2C2409.10741%2C2409.11295%2C2409.13711%2C2409.14337%2C2409.14818%2C2409.15441%2C2409.15637%2C2409.17140%2C2409.20566&max_results=50", "population": "original AppWorld benchmark (2024 paper)", "qualifier": "exact", "reference_url": "https://arxiv.org/abs/2407.18901", "unit": "tasks", "value": 750}, {"basis": "reported", "evidence_url": "https://export.arxiv.org/api/query?id_list=2406.12814%2C2406.13719%2C2406.14056%2C2406.14250%2C2406.14314%2C2406.18082%2C2406.19263%2C2407.00993%2C2407.01476%2C2407.01489%2C2407.01511%2C2407.01558%2C2407.03037%2C2407.03913%2C2407.04346%2C2407.05291%2C2407.07061%2C2407.08850%2C2407.09018%2C2407.09295%2C2407.10956%2C2407.13032%2C2407.13063%2C2407.15673%2C2407.15711%2C2407.16741%2C2407.17490%2C2407.18901%2C2407.19056%2C2407.20183%2C2408.00203%2C2408.02544%2C2408.06327%2C2408.07199%2C2408.11824%2C2408.15978%2C2409.02453%2C2409.04566%2C2409.07429%2C2409.08264%2C2409.09354%2C2409.10741%2C2409.11295%2C2409.13711%2C2409.14337%2C2409.14818%2C2409.15441%2C2409.15637%2C2409.17140%2C2409.20566&max_results=50", "population": "original AppWorld environment (2024 paper)", "qualifier": "exact", "reference_url": "https://arxiv.org/abs/2407.18901", "unit": "applications", "value": 9}], "evaluation": "State-based unit tests check task outcomes and unintended changes.", "evidence_urls": ["https://raw.githubusercontent.com/stonybrooknlp/appworld/HEAD/README.md", "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"], "focus": ["interactive coding", "cross-app tool use"], "notes": "This is an adjacent API/coding benchmark, not a screenshot-only task suite. Preserve the named release; task counts here refer to the original paper.", "observations": ["API documentation", "execution feedback"], "setting": "interactive"}, "leaderboard_url": "https://appworld.dev/appworld/leaderboard"}, "environments": ["unspecified"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T10:45:03Z", "role": "primary", "sha256": "f4049141b6e520217395b7ccd9f27048c8e0f90331690dcbef300d2bc4ff7bdd", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details.benchmark_protocol", "details.leaderboard_url"], "url": "https://raw.githubusercontent.com/stonybrooknlp/appworld/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:32Z", "role": "primary", "sha256": "b9a83994d8ac36ccdb606a5219511f8bb35ad3948ade4f9f3e28d386221aedfb", "supports": ["details.benchmark_protocol"], "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": "license_header", "retrieved_at": "2026-10-16T10:48:42Z", "role": "primary", "sha256": "c71d239df91726fc519c6eb72d318ec65820627232b2f796219e87dcf35d0ab4", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/StonyBrookNLP/appworld/HEAD/LICENSE"}], "id": "github:stonybrooknlp/appworld", "identifiers": {}, "kind": "benchmark", "last_verified_at": "2026-10-16T10:45:03Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "Apache-2.0", "retrieved_at": "2026-10-16T10:48:42Z", "url": "https://raw.githubusercontent.com/StonyBrookNLP/appworld/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/StonyBrookNLP/appworld"}], "name": "AppWorld", "notes": "Primarily API and coding interaction; included as an adjacent comparison for GUI and hybrid agents.", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/stonybrooknlp/appworld/HEAD/README.md", "target_id": "arxiv:2407.18901", "type": "documented_by"}], "review_status": "curated", "scope": "adjacent", "summary": "Simulated applications, user data, and state-based task checks for evaluating agents that interact through APIs and code.", "topics": ["evaluation", "tool-integration", "cross-app"], "updated_at": null, "url": "https://github.com/StonyBrookNLP/appworld"}
{"aliases": [], "authors": [], "details": {"benchmark_protocol": {"actions": ["web browsing", "code execution", "coworker communication"], "counts": [{"basis": "reported", "evidence_url": "https://raw.githubusercontent.com/TheAgentCompany/TheAgentCompany/HEAD/README.md", "population": "workspace images documented by the README", "qualifier": "exact", "unit": "task images", "value": 175}], "evaluation": "Task-specific evaluator scripts run within the prepared workplace environment.", "evidence_urls": ["https://raw.githubusercontent.com/TheAgentCompany/TheAgentCompany/HEAD/README.md"], "focus": ["professional workflows", "multi-tool tasks"], "notes": "Each task is packaged as a Docker image. Record task-image and server versions; the environment includes simulated coworkers and external application state.", "observations": [], "setting": "interactive"}, "leaderboard_url": "https://the-agent-company.com/#/leaderboard"}, "environments": ["web", "desktop"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T10:42:34Z", "role": "primary", "sha256": "d1646f67d49d38ef1e81bf7b5057dfc3d6f172b4c8a7413a2cc78741ccb08723", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details.benchmark_protocol", "details.leaderboard_url"], "url": "https://raw.githubusercontent.com/TheAgentCompany/TheAgentCompany/HEAD/README.md"}, {"method": "license_header", "retrieved_at": "2026-10-16T10:48:42Z", "role": "primary", "sha256": "2d6976a9b482bb14d9cd4f0d97ce6e14305496f1649b8667cb91d027c670930f", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/TheAgentCompany/TheAgentCompany/HEAD/LICENSE"}], "id": "github:theagentcompany/theagentcompany", "identifiers": {}, "kind": "benchmark", "last_verified_at": "2026-10-16T10:42:34Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "MIT", "retrieved_at": "2026-10-16T10:48:42Z", "url": "https://raw.githubusercontent.com/TheAgentCompany/TheAgentCompany/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/TheAgentCompany/TheAgentCompany"}], "name": "TheAgentCompany", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/TheAgentCompany/TheAgentCompany/HEAD/README.md", "target_id": "arxiv:2412.14161", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Self-contained workplace environment for evaluating agents on professional tasks involving websites, files, code, and simulated coworkers.", "topics": ["evaluation", "workflow", "cross-app"], "updated_at": null, "url": "https://github.com/TheAgentCompany/TheAgentCompany"}
{"aliases": [], "authors": [], "details": {"benchmark_protocol": {"actions": ["mouse and keyboard via PyAutoGUI"], "counts": [{"basis": "computed", "evidence_url": "https://raw.githubusercontent.com/xlang-ai/OSWorld/HEAD/evaluation_examples/test_all.json", "population": "test_all.json", "qualifier": "exact", "unit": "task definitions", "value": 369}, {"basis": "computed", "evidence_url": "https://raw.githubusercontent.com/xlang-ai/OSWorld/HEAD/evaluation_examples/test_nogdrive.json", "population": "test_nogdrive.json", "qualifier": "exact", "unit": "task definitions", "value": 361}], "evaluation": "Task-specific execution-based evaluators inspect application or system state.", "evidence_urls": ["https://raw.githubusercontent.com/xlang-ai/OSWorld/HEAD/README.md", "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", "https://raw.githubusercontent.com/xlang-ai/OSWorld/HEAD/evaluation_examples/test_all.json", "https://raw.githubusercontent.com/xlang-ai/OSWorld/HEAD/evaluation_examples/test_nogdrive.json"], "focus": ["task completion", "cross-application workflows"], "notes": "Counts were computed from unique task identifiers in the two inspected manifests. The subsets differ; record the manifest, repository revision, agent interface, and step budget with a score.", "observations": ["screenshots"], "setting": "interactive"}, "leaderboard_url": "https://os-world.github.io/"}, "environments": ["desktop", "linux", "windows", "macos"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:35:57Z", "role": "primary", "sha256": "83b8684c2f4ac4f31996b8b4752195fcb6cc5429f89d92398177b9b721d787fc", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details.benchmark_protocol", "details.leaderboard_url"], "url": "https://raw.githubusercontent.com/xlang-ai/OSWorld/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:28Z", "role": "primary", "sha256": "be244bfae7f89ef9449b69d85aac44295eb7b5f7e5e1bbedd76df312696676a7", "supports": ["details.benchmark_protocol"], "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": "2026-10-16T11:34:59Z", "role": "primary", "sha256": "9ebc5187cbd727ef26c24626820076b102fff812863c640a67c467fea9542ab5", "supports": ["details.benchmark_protocol"], "url": "https://raw.githubusercontent.com/xlang-ai/OSWorld/HEAD/evaluation_examples/test_all.json"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T11:34:59Z", "role": "primary", "sha256": "fcb9497e93a8986407345d3012c872c9c2fed253420730fbea64ccfcace67dbb", "supports": ["details.benchmark_protocol"], "url": "https://raw.githubusercontent.com/xlang-ai/OSWorld/HEAD/evaluation_examples/test_nogdrive.json"}, {"method": "license_header", "retrieved_at": "2026-10-16T06:41:57Z", "role": "primary", "sha256": "a8d1d88a0f9cce3cdbff73b03a80dc5e5d69e3ba2443946c268815a550c9e0ec", "supports": ["licenses.code"], "url": "https://raw.githubusercontent.com/xlang-ai/OSWorld/HEAD/LICENSE"}], "id": "github:xlang-ai/osworld", "identifiers": {}, "kind": "benchmark", "last_verified_at": "2026-10-16T06:35:57Z", "licenses": {"code": {"basis": "repository LICENSE header; not a weights/data license", "declared": "Apache-2.0", "retrieved_at": "2026-10-16T06:41:57Z", "url": "https://raw.githubusercontent.com/xlang-ai/OSWorld/HEAD/LICENSE"}, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://github.com/xlang-ai/OSWorld"}], "name": "OSWorld", "notes": "", "published_at": null, "relations": [{"evidence_url": "https://raw.githubusercontent.com/xlang-ai/OSWorld/HEAD/README.md", "target_id": "arxiv:2404.07972", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Execution-based evaluation of agents performing tasks in real desktop environments.", "topics": ["evaluation", "cross-app"], "updated_at": null, "url": "https://github.com/xlang-ai/OSWorld"}
{"aliases": [], "authors": [], "details": {"benchmark_protocol": {"actions": [], "counts": [{"basis": "reported", "evidence_url": "https://huggingface.co/datasets/mikewang/mobile_eval_e/raw/b8eca82988ed8d4f9dec01eaa9d51abf4b35aa83/README.md", "population": "Hub test split at revision b8eca829", "qualifier": "exact", "unit": "task-definition records", "value": 25}], "evaluation": "The published task schema contains rubrics and human_reference_operations. No automated scoring protocol or single numeric agent metric is established by the inspected card; execution uses separately configured Android tooling.", "evidence_urls": ["https://huggingface.co/datasets/mikewang/mobile_eval_e/raw/b8eca82988ed8d4f9dec01eaa9d51abf4b35aa83/README.md", "https://huggingface.co/api/datasets/mikewang/mobile_eval_e", "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/Mobile-Agent-E/README.md", "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"], "focus": ["long-horizon multi-app Android task execution"], "notes": "Features are task_id, instruction, type, apps, scenario, rubrics and human-reference operation strings. A task record is not an action count or recorded screenshot trajectory. Public Parquet inventory was checked but no sample rows were downloaded; companion README limits the tested agent platform to Android.", "observations": ["task instructions", "app/scenario metadata"], "setting": "interactive"}, "disabled": false, "download_url": "https://huggingface.co/datasets/mikewang/mobile_eval_e/tree/b8eca82988ed8d4f9dec01eaa9d51abf4b35aa83", "gated": false, "hub_id": "mikewang/mobile_eval_e", "hub_revision": "b8eca82988ed8d4f9dec01eaa9d51abf4b35aa83", "license_review": {"data": {"evidence_urls": ["https://huggingface.co/datasets/mikewang/mobile_eval_e/raw/b8eca82988ed8d4f9dec01eaa9d51abf4b35aa83/README.md"], "reason": "No data-license declaration is present in the inspected task card; companion code MIT does not establish a dataset grant."}}, "repository_url": "https://github.com/X-PLUG/MobileAgent/tree/main/Mobile-Agent-E"}, "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", "details.benchmark_protocol"], "url": "https://raw.githubusercontent.com/X-PLUG/MobileAgent/HEAD/Mobile-Agent-E/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:33:29Z", "role": "primary", "sha256": "f9b10eada9e0a64e26f1a7d477768f218b807103199855d893b3ab07feddc755", "supports": ["identity", "release_status", "details.hub_id", "details.hub_revision", "details.gated", "details.disabled", "details.download_url", "details.benchmark_protocol"], "url": "https://huggingface.co/api/datasets/mikewang/mobile_eval_e"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-18T04:35:15Z", "role": "primary", "sha256": "9fdf43753fbfac9868d8bec57fa5601db3e03ed5c30fb981f9da88f6b75ffcc5", "supports": ["summary", "details.benchmark_protocol", "licenses.data"], "url": "https://huggingface.co/datasets/mikewang/mobile_eval_e/raw/b8eca82988ed8d4f9dec01eaa9d51abf4b35aa83/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:31:38Z", "role": "primary", "sha256": "e96b86015c763c780e0430a8d0867b96cae947f50e888b3441c85abcaa42de78", "supports": ["summary", "topics", "details.benchmark_protocol"], "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": "url:342b081c3668a8bbe155", "identifiers": {}, "kind": "benchmark", "last_verified_at": "2026-10-18T04:27:15Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/mikewang/mobile_eval_e"}], "name": "Mobile-Eval-E task definitions", "notes": "The public Hub manifest contains one test Parquet file. This resource is the benchmark’s task-definition package, not a bundled phone environment or agent. Card metadata supplies rubrics/reference-operation fields but no data-license declaration or verified automated scorer; the code repository’s MIT license is not transferred to these records.", "published_at": null, "relations": [{"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": "Task definitions for Mobile-Agent-E’s long-horizon, multi-app Android evaluation, packaged as a 25-record test split with rubrics and human-reference operation strings.", "topics": ["evaluation", "long-horizon", "cross-app"], "updated_at": null, "url": "https://huggingface.co/datasets/mikewang/mobile_eval_e"}
{"aliases": [], "authors": ["Toby Jia-Jun Li", "Amos Azaria", "Brad A. Myers"], "details": {"publication": {"authors": ["Toby Jia-Jun Li", "Amos Azaria", "Brad A. Myers"], "date_basis": "Supplied year precision: 2017. Publications: SUGILITE, CHI 2017 Only publication year is asserted; event dates are kept separate.", "date_parts": [2017], "doi": "10.1145/3025453.3025483", "event_dates": null, "evidence_urls": ["https://www.cs.cmu.edu/~NatProg/sugilite.html", "https://toby.li/files/TobyLi-CHI2017-Sugilite.pdf"], "identity_basis": "No same-work record located in the full active catalog. Similarity screening was not used as automatic identity approval. The raw archive and exclusions are additionally checked during normalization.", "issue": null, "notes": "Related to later APPINITE and PUMICE work but distinguished here by its 2017 three-author publication and base demonstration-automation contribution. Publication month/day are not independently identified by the retained sources.", "pages": "6038-6049", "pdf_url": "https://toby.li/files/TobyLi-CHI2017-Sugilite.pdf", "publisher": null, "series": null, "title": "SUGILITE: Creating Multimodal Smartphone Automation by Demonstration", "type": "conference-paper", "url": "https://doi.org/10.1145/3025453.3025483", "venue": "Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems", "volume": null}}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:53:22Z", "role": "primary", "sha256": "32ee05ff1a91ed4a22d4bc156a462872aa4e28f5c79abfbce5dbf3655f412033", "supports": ["details.publication.date_basis", "details.publication.date_parts", "details.publication.notes", "details.publication.type", "details.publication.venue"], "url": "https://www.cs.cmu.edu/~NatProg/sugilite.html"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:54:39Z", "role": "primary", "sha256": "cc23c19284f0a558c4b99b8b051abca8ed61e274e9e1f1fed1ae91487f4368ad", "supports": ["details.publication.authors", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.title", "details.publication.url"], "url": "https://toby.li/files/TobyLi-CHI2017-Sugilite.pdf"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:54:39Z", "role": "primary", "sha256": "cc23c19284f0a558c4b99b8b051abca8ed61e274e9e1f1fed1ae91487f4368ad", "supports": ["details.publication.authors", "details.publication.doi", "details.publication.identity_basis", "details.publication.notes", "details.publication.pages", "details.publication.pdf_url", "details.publication.title", "details.publication.url", "environments", "name", "scope", "summary", "topics"], "url": "https://toby.li/files/TobyLi-CHI2017-Sugilite.pdf"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T09:53:22Z", "role": "primary", "sha256": "32ee05ff1a91ed4a22d4bc156a462872aa4e28f5c79abfbce5dbf3655f412033", "supports": ["details.publication.date_basis", "details.publication.date_parts", "details.publication.notes", "details.publication.type", "details.publication.venue", "kind"], "url": "https://www.cs.cmu.edu/~NatProg/sugilite.html"}], "id": "url:392a602c660ffa91be46", "identifiers": {"doi": "10.1145/3025453.3025483"}, "kind": "paper", "last_verified_at": "2026-10-17T09:54:39Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://doi.org/10.1145/3025453.3025483"}, {"rel": "publication_pdf", "url": "https://toby.li/files/TobyLi-CHI2017-Sugilite.pdf"}], "name": "SUGILITE: Creating Multimodal Smartphone Automation by Demonstration", "notes": "The annotation was reviewed against retained primary publication sources. Experiments have not been independently reproduced. Related to later APPINITE and PUMICE work but distinguished here by its 2017 three-author publication and base demonstration-automation contribution.", "published_at": null, "relations": [], "review_status": "curated", "scope": "core", "summary": "Introduces SUGILITE, which combines spoken instructions, demonstrated Android actions, and UI hierarchies to generate reusable automation scripts. Its interface includes context checking and repairs for unfamiliar situations and changing app interfaces.", "topics": ["mobile-navigation", "recording", "human-in-the-loop", "cross-app"], "updated_at": null, "url": "https://doi.org/10.1145/3025453.3025483"}
{"aliases": [], "authors": [], "details": {"dataset_profile": {"counts": [{"basis": "reported", "evidence_url": "https://huggingface.co/datasets/hflqf88888/GUIOdyssey/raw/main/README.md", "population": "recommended hflqf88888/GUIOdyssey release", "qualifier": "exact", "unit": "episodes", "value": 8334}], "evidence_urls": ["https://raw.githubusercontent.com/OpenGVLab/GUI-Odyssey/HEAD/README.md", "https://huggingface.co/datasets/hflqf88888/GUIOdyssey/raw/main/README.md", "https://raw.githubusercontent.com/OpenGVLab/GUI-Odyssey/HEAD/introduction.md"], "formats": ["JSON annotations", "PNG screenshots"], "modalities": ["screenshots", "instructions", "actions", "bounding boxes", "step context"], "notes": "The repository gives conflicting episode totals (8,334 and 8,834). This profile uses 8,334 from the specifically recommended current dataset card. No independent archive recount was performed. Coordinates are normalized to 0–1000.", "splits": ["random", "task", "device", "app"], "supervision": ["cross-app trajectories", "action grounding"]}, "download_url": "https://huggingface.co/datasets/hflqf88888/GUIOdyssey/tree/main", "sample_count": 8334, "sample_unit": "episodes reported by the current release card"}, "environments": ["android"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-16T11:39:52Z", "role": "primary", "sha256": "fd627ee0a6ee0596e0ee8147996b9a7a3436f754fc95ed63cf5f408d778b2831", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/datasets/hflqf88888/GUIOdyssey/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T06:36:01Z", "role": "primary", "sha256": "4585b765d7c8b3e9bb6187d36d576edde454d4c0158a6aa8b3fdbedf34c89a15", "supports": ["details.dataset_profile"], "url": "https://raw.githubusercontent.com/OpenGVLab/GUI-Odyssey/HEAD/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-16T11:39:52Z", "role": "primary", "sha256": "78c19ee712487af60df1913f11572958564e59a1a2cc663b1b30d971855d7a28", "supports": ["details.dataset_profile"], "url": "https://raw.githubusercontent.com/OpenGVLab/GUI-Odyssey/HEAD/introduction.md"}, {"method": "huggingface_card", "retrieved_at": "2026-10-16T11:39:52Z", "role": "primary", "sha256": "fd627ee0a6ee0596e0ee8147996b9a7a3436f754fc95ed63cf5f408d778b2831", "supports": ["licenses.data"], "url": "https://huggingface.co/datasets/hflqf88888/GUIOdyssey/raw/main/README.md"}], "id": "url:511903ebfac45e6751b2", "identifiers": {}, "kind": "dataset", "last_verified_at": "2026-10-16T11:39:52Z", "licenses": {"code": null, "data": {"basis": "publisher model/dataset card metadata; third-party assets retain their own terms", "declared": "cc-by-4.0", "retrieved_at": "2026-10-16T11:39:52Z", "url": "https://huggingface.co/datasets/hflqf88888/GUIOdyssey/raw/main/README.md"}, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/hflqf88888/GUIOdyssey"}], "name": "GUIOdyssey — current data release", "notes": "The project recommends this release over its earlier OpenGVLab/GUI-Odyssey upload. Episode count follows the current data card; conflicting repository prose is documented in the profile.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/hflqf88888/GUIOdyssey/raw/main/README.md", "target_id": "arxiv:2406.08451", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Cross-app Android interaction episodes with fine-grained step instructions, screen descriptions, action context, and target boxes.", "topics": ["mobile-navigation", "cross-app", "data-generation"], "updated_at": null, "url": "https://huggingface.co/datasets/hflqf88888/GUIOdyssey"}
{"aliases": [], "authors": [], "details": {"benchmark_protocol": {"actions": ["mobile interface operations", "clarification requests"], "counts": [{"basis": "reported", "evidence_url": "https://huggingface.co/datasets/vivo-ai/MobileGUI-VBench/raw/main/README.md", "population": "MobileGUI-VBench release", "qualifier": "exact", "unit": "tasks", "value": 150}, {"basis": "reported", "evidence_url": "https://huggingface.co/datasets/vivo-ai/MobileGUI-VBench/raw/main/README.md", "population": "MobileGUI-VBench release", "qualifier": "exact", "unit": "mobile apps", "value": 40}], "evaluation": "Published task definitions include preconditions and clarification requirements; the metadata review does not independently validate the reported agent scores.", "evidence_urls": ["https://huggingface.co/datasets/vivo-ai/MobileGUI-VBench/raw/main/README.md"], "focus": ["mobile task completion", "clarification behavior", "cross-app tasks"], "notes": "The card describes 140 single-app and 10 cross-app tasks. Task and app counts are different units.", "observations": ["mobile app interfaces"], "setting": "interactive"}, "disabled": false, "download_url": "https://huggingface.co/datasets/vivo-ai/MobileGUI-VBench/tree/main", "gated": false, "hub_id": "vivo-ai/MobileGUI-VBench"}, "environments": ["unspecified"], "evidence": [{"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:42:32Z", "role": "primary", "sha256": "422ff6a557a1aabd8f6c6ef26e5fd05b992f4acf0e5128ab8bc437021fcc5ea7", "supports": ["name", "kind", "summary", "environments", "topics", "scope", "relations", "details"], "url": "https://huggingface.co/datasets/vivo-ai/MobileGUI-VBench/raw/main/README.md"}, {"method": "editorial_primary_review", "retrieved_at": "2026-10-17T02:42:32Z", "role": "primary", "sha256": "5850fc58f54c877cffe328dba2a0b5e3e98767637aae6d6efd9a037cf06eabce", "supports": ["identity", "release_status", "details", "licenses.data"], "url": "https://huggingface.co/api/datasets/vivo-ai/MobileGUI-VBench"}], "id": "url:d7a4f34a6a5202af5f64", "identifiers": {}, "kind": "benchmark", "last_verified_at": "2026-10-17T02:42:32Z", "licenses": {"code": null, "data": null, "document": null, "weights": null}, "links": [{"rel": "primary", "url": "https://huggingface.co/datasets/vivo-ai/MobileGUI-VBench"}], "name": "MobileGUI-VBench", "notes": "Public task files are available as JSONL and XLSX with documentation. The card explicitly declares Apache-2.0 for the dataset, despite having no license field in its YAML metadata. It identifies mobile apps without fixing an operating-system taxonomy; model scores are not imported.", "published_at": null, "relations": [{"evidence_url": "https://huggingface.co/datasets/vivo-ai/MobileGUI-VBench/raw/main/README.md", "target_id": "arxiv:2609.12394", "type": "documented_by"}], "review_status": "curated", "scope": "core", "summary": "Mobile-app task definitions covering everyday scenarios, with annotated preconditions and clarification requirements for evaluating agent behavior.", "topics": ["evaluation", "mobile-navigation", "cross-app"], "updated_at": null, "url": "https://huggingface.co/datasets/vivo-ai/MobileGUI-VBench"}