Windows-MCP

Connects AI agents to the Windows operating system for file navigation, application control, and UI interaction.

Windows-MCP screenshot 1

Windows-MCP is an open-source Model Context Protocol (MCP) server that connects AI agents to the Windows operating system. It allows Large Language Models to interact with the desktop environment to perform tasks such as file navigation, application control, and UI interaction. The software is designed to work with various AI clients, including Claude Desktop, Perplexity Desktop, Gemini CLI, and Qwen Code, acting as a bridge between the LLM and the local machine.

The application runs as a server on Windows 7, 8, 10, and 11. It can be installed via PyPI using the uv package manager or run directly from source. Users can configure it to run as a background task that starts automatically at login via a per-user Scheduled Task or execute it manually via the command line using different transport methods like SSE or streamable-HTTP. This flexibility allows it to be integrated into diverse AI workflows depending on the preferred connection method.

Key features

  • Native interaction with Windows UI elements and application control
  • Support for any LLM without requiring computer vision or fine-tuned models
  • Toolset for simulating mouse and keyboard operations
  • DOM mode for browser automation in Chrome, Edge, and Firefox
  • Ability to capture current window and UI states
  • Low latency action execution ranging from 0.2 to 0.5 seconds
  • Integration with the MCP Registry for easier discovery

Built with Python 3.13, the server acts as a bridge that translates LLM requests into system-level actions. It avoids reliance on traditional vision techniques, making it compatible with text-based models and reducing the complexity of the initial setup. The inclusion of a specific DOM mode allows agents to filter out browser UI elements and focus exclusively on web page content for cleaner automation, utilizing IAccessible2 fallbacks for Firefox. It is intended for developers, automation engineers, and QA testers who need to automate Windows workflows using AI agents.

Windows-MCP serves as a lightweight infrastructure layer for enabling computer-use capabilities on the Windows platform.

Last Modified
Software TypeWeb App / Server
Platform
Last Activity16 days ago
Repository Age1 year
LicenseMIT
Open Source Software.io

Join our newsletter to get shiny new open source software delivered to your inbox. Unsubscribe anytime.