{
  "aliases": [],
  "authors": [],
  "details": {},
  "environments": [
    "desktop"
  ],
  "evidence": [
    {
      "method": "editorial_primary_review",
      "retrieved_at": "2026-10-16T06:36:02Z",
      "role": "primary",
      "sha256": "2248b2048b9166ce0348b24e81b2587fdf373a958fd86ab870ba85825425b02f",
      "supports": [
        "name",
        "kind",
        "summary",
        "environments",
        "topics",
        "scope",
        "relations"
      ],
      "url": "https://raw.githubusercontent.com/niuzaisheng/ScreenAgent/HEAD/README.md"
    }
  ],
  "id": "github:niuzaisheng/screenagent",
  "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/niuzaisheng/ScreenAgent"
    }
  ],
  "name": "ScreenAgent",
  "notes": "",
  "published_at": null,
  "relations": [
    {
      "evidence_url": "https://raw.githubusercontent.com/niuzaisheng/ScreenAgent/HEAD/README.md",
      "target_id": "arxiv:2402.07945",
      "type": "documented_by"
    }
  ],
  "review_status": "curated",
  "scope": "core",
  "summary": "Environment and agent in which a vision-language model controls real desktops over VNC with mouse and keyboard in plan, act and reflect stages, plus a hand-annotated dataset of daily computer tasks.",
  "topics": [
    "planning"
  ],
  "updated_at": null,
  "url": "https://github.com/niuzaisheng/ScreenAgent"
}