Shayan Hashemi/MCP_MAC_USE
Built by Metorial, the integration platform for agentic AI.
Shayan Hashemi/MCP_MAC_USE
Server Summary
Execute AppleScript commands
Control mouse movement and clicks
Simulate keyboard input
Open applications
Retrieve system information
Natural language control via LLM
This is a Model Context Protocol (MCP) server that allows controlling macOS through an AI interface. It provides tools for executing AppleScript, controlling the mouse and keyboard, and retrieving system information.
Install dependencies:
npm install
Run the MCP server:
npm start
For LLM integration: Set your LLM API key and run the LLM server:
export LLM_API_KEY=your-api-key-here
npm run serve
We've added a web interface that allows you to control your Mac using natural language commands through an LLM. To use it:
Run the LLM server:
npm run serve
Open your browser and navigate to http://localhost:3000
Enter natural language commands like:
For more detailed instructions, see the LLM Integration Guide.
The server can be configured through environment variables or through the Smithery configuration panel:
ENABLE_MOUSE_CONTROL
: Enable or disable mouse control functions (default: true)ENABLE_KEYBOARD_CONTROL
: Enable or disable keyboard input functions (default: true)ENABLE_APPLESCRIPT
: Enable or disable AppleScript execution (default: true)This MCP server follows the standard MCP protocol:
{
"type": "tool_call",
"data": {
"id": "unique-id",
"name": "tool_name",
"params": {
"param1": "value1"
}
}
}
{
"type": "tool_result",
"data": {
"id": "unique-id",
"result": {
"key": "value"
}
}
}
This repository includes the necessary configuration files for deployment on Smithery:
Dockerfile
: Defines the container buildsmithery.yaml
: Configuration for the Smithery platformSince this MCP server can control your Mac, it should be used with caution:
MIT