{
  "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"
}