Clawd Cursor
clawdcursor.com
What it is
Clawd Cursor gives AI agents local control of desktop applications for GUI tasks where a native API or command-line interface is unavailable. It builds a screen map from accessibility data and OCR, assigning stable element IDs and confidence scores; screenshots are used only when needed. Agents can click and type, and actions can be checked against an expected outcome. The maker says every tool call goes through one safety layer and describes the software as local, localhost-only by default, and without telemetry. It supports Windows, macOS 12 or later, and Linux X11 or Wayland. Connections use stdio MCP or an HTTP MCP daemon at 127.0.0.1:3847/mcp. Listed hosts include Claude Code, Cursor, Windsurf, OpenClaw, and Zed. The tool offers seven recommended compound tools and 98 granular tools for compatibility and debugging; the maker says it works with any model. It is free and MIT licensed. Installation requires Node.js 20 or later. Desktop control needs one-time consent; macOS also requires Accessibility permission, while Screen Recording is optional for vision fallback.
Spec sheet
- What it does
- Clawd Cursor gives AI agents local control of desktop apps through a screen map, actions such as clicks and typing, and action verification.Read 29 Sep 2026 · clawdcursor.com
- Integrations
- The site lists Claude Code, Cursor, Windsurf, OpenClaw, and Zed as MCP hosts.Read 29 Sep 2026 · clawdcursor.com
- License
- The site identifies Clawd Cursor as MIT licensed.Read 29 Sep 2026 · clawdcursor.com
- Connection modes
- Agents can connect through stdio MCP or an HTTP MCP daemon on 127.0.0.1:3847/mcp.Read 29 Sep 2026 · clawdcursor.com
- Installation requirement
- The GitHub README says Node.js 20 or later is required.Read 29 Sep 2026 · github.com
- Intended use
- The README describes it as a last-mile option for GUI tasks where a native API or CLI is unavailable.Read 29 Sep 2026 · github.com
- Local operation
- The maker describes it as local, localhost-only by default, and without telemetry.Read 29 Sep 2026 · clawdcursor.com
- Model support
- The maker says it works with any model; the daemon can also use a configured built-in LLM.Read 29 Sep 2026 · clawdcursor.com
- Operating systems
- It supports Windows x64/ARM64, macOS 12 or later, and Linux X11/Wayland.Read 29 Sep 2026 · clawdcursor.com
- Permissions
- Desktop control requires one-time consent; macOS also requires Accessibility permission, while Screen Recording is optional for vision fallback.Read 29 Sep 2026 · github.com
- Safety
- Every tool call passes through one safety layer, and actions can be checked against an expected outcome.Read 29 Sep 2026 · clawdcursor.com
- Screen reading
- It combines accessibility data and OCR into a confidence-scored map with stable element IDs, using screenshots only when needed.Read 29 Sep 2026 · clawdcursor.com
- Tools
- It offers 7 recommended compound tools and 98 granular tools for compatibility and debugging.Read 29 Sep 2026 · clawdcursor.com
Plans and prices
Strong points and limits
Strong points
- Maps screens using accessibility data and OCR
- Actions can be checked against expected outcomes
- Localhost-only by default, without telemetry
- Supports Windows, macOS, and Linux
Limits
- Requires Node.js 20 or later
- Desktop control requires one-time consent
- macOS requires Accessibility permission
Who it suits
Clawd Cursor suits users building AI-agent workflows for desktop tasks that lack a native API or CLI. It may fit teams using a listed MCP host and able to meet its installation and permission requirements.
Verdict
Clawd Cursor offers local desktop control with screen mapping and action checks. Check its Node.js requirement and platform permissions before adding it to an agent workflow.
Where Specifiction places it
- AI Computer Use Agents#5 of 248.0