{
  "aliases": [],
  "authors": [],
  "details": {
    "benchmark_protocol": {
      "actions": [
        "pixel (x,y) cursor coordinate"
      ],
      "counts": [],
      "evaluation": "The inspected evaluator scales target boxes from a 1000×1000 reference grid into image pixels and tests pixel predictions, allowing ±10 pixels around the center along any target dimension narrower than 1 pixel after scaling. It reports cumulative hit accuracy by turn (earlier hits remain correct), miss distances, any-turn hit rate, and a correction rate whose denominator is all successfully processed samples. The default budget is five turns; retain prompt/feedback variant, turn budget, and processed/skipped coverage.",
      "evidence_urls": [
        "https://raw.githubusercontent.com/microsoft/precision-cua-bench/HEAD/README.md",
        "https://raw.githubusercontent.com/microsoft/precision-cua-bench/HEAD/eval/test_allturns_api.py",
        "https://api.github.com/repos/microsoft/precision-cua-bench/git/trees/main?recursive=1"
      ],
      "focus": [
        "pixel-precise editor cursor placement",
        "multi-turn visual self-correction"
      ],
      "notes": "The release has four editor/theme JSONL partitions (Cursor/VS Code × dark/light) and screenshot files. Partitions are not train/test splits or independent task totals. README’s initial field table calls bbox pixel coordinates, but the evaluation-format section and actual ratio_h/ratio_w=dimension/1000 code use normalized target boxes; profile follows the inspected evaluator and retains the inconsistency. Source/images were not executed or downloaded as a dataset. MIT is recorded for repository code only.",
      "observations": [
        "code-editor screenshot",
        "cursor-location instruction",
        "red-cross feedback marking previous incorrect prediction"
      ],
      "setting": "offline"
    }
  },
  "environments": [
    "cross-platform"
  ],
  "evidence": [
    {
      "method": "editorial_primary_review",
      "retrieved_at": "2026-10-16T06:40:52Z",
      "role": "primary",
      "sha256": "23d0dc8ea0a10c080cf8ae32121eade2785ae923a6b0d853d166c94c195e789b",
      "supports": [
        "name",
        "kind",
        "summary",
        "environments",
        "topics",
        "scope",
        "relations"
      ],
      "url": "https://github.com/microsoft/precision-cua-bench"
    },
    {
      "method": "editorial_primary_review",
      "retrieved_at": "2026-10-16T06:35:59Z",
      "role": "primary",
      "sha256": "84ad4fb798abb776dce813cf4dab296d6e03248d8ae29388e1f06cccd27d1241",
      "supports": [
        "details.benchmark_protocol"
      ],
      "url": "https://raw.githubusercontent.com/microsoft/precision-cua-bench/HEAD/README.md"
    },
    {
      "method": "editorial_primary_review",
      "retrieved_at": "2026-10-17T14:21:08Z",
      "role": "primary",
      "sha256": "1fda7c3980fe7a7c373478eda6ecfabff227e92c2b1f19c6fe903b408f583797",
      "supports": [
        "details.benchmark_protocol"
      ],
      "url": "https://raw.githubusercontent.com/microsoft/precision-cua-bench/HEAD/eval/test_allturns_api.py"
    },
    {
      "method": "editorial_primary_review",
      "retrieved_at": "2026-10-17T14:14:06Z",
      "role": "primary",
      "sha256": "ede84e3dda47f80fada4234caa2423db08e21ff7667a37599c03aae657b2f84b",
      "supports": [
        "details.benchmark_protocol"
      ],
      "url": "https://api.github.com/repos/microsoft/precision-cua-bench/git/trees/main?recursive=1"
    },
    {
      "method": "editorial_primary_review",
      "retrieved_at": "2026-10-16T06:32:00Z",
      "role": "primary",
      "sha256": "4d09ebfb093e43a4c44adef86c8fa8374f90d907fc33bd8e12a1315e62d5a583",
      "supports": [
        "relations"
      ],
      "url": "https://export.arxiv.org/api/query?id_list=2603.12823%2C2603.14707%2C2603.15039%2C2603.17357%2C2603.18429%2C2603.19191%2C2603.24440%2C2603.25226%2C2603.25864%2C2603.26211%2C2603.26648%2C2603.29318%2C2604.01535%2C2604.02947%2C2604.04399%2C2604.05477%2C2604.06126%2C2604.06182%2C2604.06367%2C2604.07429%2C2604.08455%2C2604.08516%2C2604.08523%2C2604.09155%2C2604.09574%2C2604.09587%2C2604.09815%2C2604.09937%2C2604.11259%2C2604.11784%2C2604.12284%2C2604.13019%2C2604.13822%2C2604.14113%2C2604.14262%2C2604.14956%2C2604.15093%2C2604.15309%2C2604.17817%2C2604.17821%2C2604.18860%2C2604.24964%2C2604.25380%2C2604.25562%2C2604.27151%2C2604.27253%2C2604.27776%2C2605.00642%2C2605.06393%2C2605.06664&max_results=50"
    },
    {
      "method": "license_header",
      "retrieved_at": "2026-10-16T06:41:59Z",
      "role": "primary",
      "sha256": "c2cfccb812fe482101a8f04597dfc5a9991a6b2748266c47ac91b6a5aae15383",
      "supports": [
        "licenses.code"
      ],
      "url": "https://raw.githubusercontent.com/microsoft/precision-cua-bench/HEAD/LICENSE"
    }
  ],
  "id": "github:microsoft/precision-cua-bench",
  "identifiers": {},
  "kind": "benchmark",
  "last_verified_at": "2026-10-16T06:40:52Z",
  "licenses": {
    "code": {
      "basis": "repository LICENSE header; not a weights/data license",
      "declared": "MIT",
      "retrieved_at": "2026-10-16T06:41:59Z",
      "url": "https://raw.githubusercontent.com/microsoft/precision-cua-bench/HEAD/LICENSE"
    },
    "data": null,
    "document": null,
    "weights": null
  },
  "links": [
    {
      "rel": "primary",
      "url": "https://github.com/microsoft/precision-cua-bench"
    }
  ],
  "name": "Precision CUA Bench",
  "notes": "",
  "published_at": null,
  "relations": [
    {
      "evidence_url": "https://export.arxiv.org/api/query?id_list=2603.12823%2C2603.14707%2C2603.15039%2C2603.17357%2C2603.18429%2C2603.19191%2C2603.24440%2C2603.25226%2C2603.25864%2C2603.26211%2C2603.26648%2C2603.29318%2C2604.01535%2C2604.02947%2C2604.04399%2C2604.05477%2C2604.06126%2C2604.06182%2C2604.06367%2C2604.07429%2C2604.08455%2C2604.08516%2C2604.08523%2C2604.09155%2C2604.09574%2C2604.09587%2C2604.09815%2C2604.09937%2C2604.11259%2C2604.11784%2C2604.12284%2C2604.13019%2C2604.13822%2C2604.14113%2C2604.14262%2C2604.14956%2C2604.15093%2C2604.15309%2C2604.17817%2C2604.17821%2C2604.18860%2C2604.24964%2C2604.25380%2C2604.25562%2C2604.27151%2C2604.27253%2C2604.27776%2C2605.00642%2C2605.06393%2C2605.06664&max_results=50",
      "target_id": "arxiv:2604.13019",
      "type": "documented_by"
    }
  ],
  "review_status": "curated",
  "scope": "core",
  "summary": "Evaluation resources for pixel-precise cursor grounding in code editors.",
  "topics": [
    "evaluation",
    "grounding"
  ],
  "updated_at": null,
  "url": "https://github.com/microsoft/precision-cua-bench"
}