{
  "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"
}