Home / MCP / Windows MCP Server
Offers Windows automation and UI control via MCP server interfaces for AI agents.
Configuration
View docs{
"mcpServers": {
"windows_mcp": {
"command": "uvx",
"args": [
"windows-mcp"
],
"env": {
"ANONYMIZED_TELEMETRY": "false"
}
}
}
}Windows-MCP is a lightweight open-source MCP server that lets AI agents control and automate tasks on Windows. It provides native Windows UI integration, app control, UI interaction, and testing capabilities, enabling you to drive the OS from an LLM or automation agent with minimal setup.
You connect an MCP client to the Windows-MCP server using a local stdio configuration. This lets your agent send commands to click, type, move the mouse, launch applications, capture window state, and interact with browser content when using DOM mode. The server exposes a set of tools for UI automation and state capture, and it works with any LLM without requiring specialized computer vision models.
Prerequisites: install Python 3.13+ and the UV package manager from Astra. You also need an English environment as the default language for Windows, or disable the App-Tool in the MCP Server if you use other languages.
# Prerequisites
# Install Python 3.13+
# Install UV package manager (choose one)
pip install uvx
# or
curl -LsSf https://astral.sh/uv/install.sh | sh
```} ,{You can install and run Windows-MCP via multiple client environments. The following examples show how to configure MCP connections in several desktop environments.
# Option A: Run from PyPI via UVX (recommended for Claude Desktop)
# This config runs the latest Windows-MCP binary from PyPI
{
"mcpServers": {
"windows-mcp": {
"command": "uvx",
"args": [
"windows-mcp"
]
}
}
}To run from source directly, use UV with a directory path to the Windows-MCP source and start the server.
# Option B: Run from source
{
"mcpServers": {
"windows-mcp": {
"command": "uv",
"args": [
"--directory",
"<path to the windows-mcp directory>",
"run",
"windows-mcp"
]
}
}
}
```You can connect Windows-MCP to various desktops and CLIs by supplying the same stdio configuration in the client’s settings. The key is to run the server using the appropriate command and path for your environment.
Clicks on screen coordinates to simulate mouse interactions.
Types text into an element, with optional clearing of existing text.
Scrolls vertically or horizontally within windows or regions.
Drags the cursor from one point to another.
Moves the mouse pointer to a specified location.
Sends keyboard shortcuts like Ctrl+C or Alt+Tab.
Pauses execution for a defined duration.
Captures a snapshot of language, active apps, and UI state, with optional DOM/vision modes.
Launches, resizes, moves, and switches between applications.
Executes PowerShell commands.
Scrapes page content for information.