MousePilot
Enables AI agents to control a macOS 14+ desktop via MCP or CLI, including taking screenshots, clicking, typing, scrolling, dragging, and using keyboard shortcuts.
This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
MousePilot
Contents · Demos · Install · Use
Demos
Windows · Edge InPrivate

macOS · Safari Private Browsing

Recorded on QA Practice Hub: scrolling, typing, keyboard shortcuts, clicks, and dragging. Silent, with pauses trimmed.
Related MCP server: mcp-macos-cua
Install
Install uv and download or clone this repository. Run the backend on the computer you want to control, while logged in and unlocked.
Windows
From the project folder in PowerShell:
uv sync --locked --no-dev
.\.venv\Scripts\mousepilot-server.exe --trayTo build a standalone executable instead:
.\scripts\build-windows.ps1
.\dist\mousepilot-server.exe --traymacOS
From the project folder on the Mac:
./scripts/build-macos.sh
mkdir -p ~/Applications
ditto dist/MousePilot.app ~/Applications/MousePilot.app
open ~/Applications/MousePilot.appIn the MousePilot menu, choose Grant permissions…. Enable Accessibility, Input Monitoring, and Screen Recording in Privacy & Security.
Use
On the computer running your agent (Linux, macOS, or Windows):
uv sync --lockedFollow configuration to connect your device over SSH and add MousePilot to your MCP client.
Ask your agent to take a screenshot, then click, type, scroll, or drag.
Mouse coordinates use the returned image's pixels and its screenshot_id.
type_text does not press Enter.
For manual control after configuring a device:
uv run mousepilot --device windows screen
uv run mousepilot --device windows click 640 400 --frame SCREENSHOT_ID
uv run mousepilot --device windows type 'hello'Use --device mac for the Mac profile. More commands are in usage.
Emergency stop: Ctrl+Alt+Shift+F11 on Windows; Ctrl+Option+Shift+F11 on macOS. Resume explicitly through MCP or the tray/menu.
AI-assisted development
AI tools have been used during development. Contributors remain responsible for reviewing changes and identifying third-party sources. See Contributing.
Updates and contributions
MousePilot is an independent personal project. Updates follow my needs as I use it. There is no fixed release schedule. Use the repository's Issues tab to report a bug or request a feature. Describe the problem and the result you need. You can also fork the project and submit a pull request. Feature requests and pull requests do not guarantee an update or acceptance.
See Contributing for setup and checks.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
This server cannot be deployed
Maintenance
Related MCP Connectors
Connect AI assistants to your approved computer for files, commands, browser and desktop tasks.
Connect ChatGPT, Claude, Cursor or VS Code to your Windows PC: files, terminal, browser, apps.
Web scraping, browser automation, screenshots and AI data extraction for AI agents via MCP.
Use your Mac, Windows or Linux computer from ChatGPT, Claude or Codex: files, commands, documents.
Related MCP Servers
- AlicenseAqualityDmaintenanceEnables LLM agents to capture screenshots, control mouse/keyboard, and manage windows on desktop platforms, primarily Windows, via an MCP server.1616 PyPI1MIT
- AlicenseAqualityDmaintenanceEnables AI agents to control macOS desktop apps via screenshots, mouse clicks, keyboard input, accessibility queries, and AppleScript.1112 npmMIT
- AlicenseNot gradedqualityBmaintenanceEnables AI agents to control remote desktops through screen capture, mouse movement, and keyboard input.MIT
- AlicenseNot gradedqualityCmaintenanceEnables AI agents to capture multi-monitor screenshots with coordinate grids and perform mouse, keyboard, scrolling, dragging, and window-management actions on Windows via MCP tools or CLI. It provides pixel-accurate desktop automation for computer-use agents.MIT