Provides full GUI control over macOS, including screenshots, mouse and keyboard input, application management, clipboard access, and multi-display support.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@computer-use open Safari and navigate to example.com"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Feature | Description | |
Vision | Screenshot & Display | Capture any display, enumerate monitors, zoom into regions |
Input | Mouse & Keyboard | Click, drag, scroll, type, key combos, hold keys — with IME-safe clipboard routing |
Apps | Application Control | Launch apps, detect frontmost app, list installed/running apps, tiered permission model |
Clipboard | Read & Write | Full clipboard access for paste-based workflows |
Batch | Action Batching | Chain multiple actions in a single MCP call for speed |
Runtime | Zero-Config Bootstrap | Auto-creates Python virtualenv and installs dependencies on first run |
Portable | Skill Packaging | Ships as a standalone skill — install once, works without the source repo |
Public | No Private Dependencies | Built entirely on public packages: Node.js, Python, pyautogui, mss, Pillow, pyobjc |
Related MCP server: MacWright
1. Clone & build
git clone https://github.com/wimi321/macos-computer-use-skill.git
cd macos-computer-use-skill
npm install && npm run build2. Run the MCP server
node dist/cli.jsOn first launch the server automatically creates a Python virtualenv in .runtime/venv and installs all runtime dependencies. No Claude desktop app, no private native modules.
3. Or install from ClawHub
clawhub install computer-use-macosmacOS requiresAccessibility and Screen Recording permissions for the host process. The server checks both on startup and reports status through MCP.
flowchart LR
A[AI Agent / MCP Client] --> B[MCP Server<br/>TypeScript + stdio]
B --> C[Tool Layer<br/>28 MCP tools]
B --> D[Python Bridge<br/>auto-bootstrapped venv]
D --> E[pyautogui]
D --> F[mss + Pillow]
D --> G[pyobjc<br/>Cocoa + Quartz]
E --> H[Mouse / Keyboard]
F --> I[Screenshots]
G --> J[Apps / Displays<br/>Clipboard / Windows]Tool | Description |
| Capture the current display as a JPEG image |
| Crop and zoom into a region of the last screenshot |
| Switch the active capture target to a different monitor |
Tool | Description |
| Left-click at a coordinate |
| Double-click at a coordinate |
| Triple-click (select paragraph/line) |
| Right-click (context menu) |
| Middle-click |
| Click-and-drag between two points |
| Press and hold the left mouse button |
| Release the left mouse button |
| Move the cursor without clicking |
| Scroll in any direction at a coordinate |
| Type text (clipboard-routed on macOS to avoid IME corruption) |
| Press a key combo (e.g. |
| Hold a key for a duration |
| Get the current cursor coordinates |
Tool | Description |
| Launch a macOS application by name |
| Request access to interact with an application |
| List apps the current session has permission to control |
| Read the system clipboard |
| Write to the system clipboard |
| Pause for a specified duration |
Tool | Description |
| Execute multiple actions in a single call |
| Request elevated access for teaching workflows |
| Single-step action in teach mode |
| Batch actions in teach mode |
Add to your MCP client config:
{
"mcpServers": {
"computer-use": {
"command": "node",
"args": ["/absolute/path/to/macos-computer-use-skill/dist/cli.js"],
"env": {
"CLAUDE_COMPUTER_USE_DEBUG": "0",
"CLAUDE_COMPUTER_USE_COORDINATE_MODE": "pixels"
}
}
}
}See examples/mcp-config.json for a ready-to-use template.
This project ships as a self-contained skill at skill/computer-use-macos.
From ClawHub:
clawhub install computer-use-macosFrom the repo:
bash skill/computer-use-macos/scripts/install.shThe installer copies the full project to ~/.codex/skills/computer-use-macos/project — the skill keeps working even if the original clone is removed.
Variable | Default | Description |
|
| Enable verbose debug logging |
|
| Coordinate mode: |
|
| Prefer clipboard-based typing (IME-safe) |
|
| Animate mouse movement |
|
| Hide overlay windows before actions |
Requirement | Version |
macOS | 12+ (Monterey or later) |
Node.js | 20+ |
Python | 3.10+ (ships with macOS or via Homebrew) |
Permissions | Accessibility + Screen Recording |
Python dependencies (pyautogui, mss, Pillow, pyobjc) are installed automatically into an isolated virtualenv on first run.
macos-computer-use-skill/
├── src/
│ ├── cli.ts # Entry point
│ ├── server.ts # MCP server setup
│ ├── session.ts # Session context factory
│ ├── computer-use/
│ │ ├── executor.ts # macOS executor (bridges to Python)
│ │ ├── pythonBridge.ts # Venv bootstrap + Python IPC
│ │ ├── hostAdapter.ts # Host adapter factory
│ │ └── ...
│ └── vendor/computer-use-mcp/
│ ├── mcpServer.ts # MCP server factory
│ ├── toolCalls.ts # Tool dispatch logic
│ ├── tools.ts # MCP tool schemas
│ └── ...
├── runtime/
│ ├── mac_helper.py # Python runtime (pyautogui + pyobjc)
│ └── requirements.txt
├── skill/
│ └── computer-use-macos/ # Portable skill package
├── examples/
│ ├── mcp-config.json
│ └── env.sh.example
├── assets/
│ └── hero.svg
├── package.json
└── tsconfig.jsonApp icon extraction without private APIs
Stronger nested helper-app filtering
Automated MCP integration test suite
Pre-built release artifacts for easier distribution
Contributions are welcome. See CONTRIBUTING.md for guidelines.
This project extracts and adapts reusable TypeScript computer-use logic from the Claude Code workflow, replacing the private native runtime with a fully standalone, publicly installable macOS implementation. Built on top of the Model Context Protocol.
This server cannot be deployed
A comprehensive Model Context Protocol (MCP) server that enables AI assistants to interact with yo…
Use your own Mac from ChatGPT, Claude or Codex: files, commands, documents, and a browser.
Nifty's MCP server — exposes tasks, projects, messages, and files as tools for AI agents.
MCP server connecting AI agents to 100+ apps (Gmail, Slack, Notion, GitHub) via one-click OAuth.
