Windows-MCP

Why we included this project

Windows-MCP plugs an AI assistant directly into the Windows desktop, so it can click buttons, type text, move the mouse, switch between windows, and work with files, all over the Model Context Protocol. That is useful when the goal is an agent doing real work inside existing Windows programs rather than just answering questions. Teams running automated QA passes, checking for UI regressions, or building a desktop helper that walks through multi-step tasks in everyday apps are the audience here. It reads the native UI state instead of guessing from a vision model, which means it works with whatever LLM you already have, and since it ships on PyPI, most MCP clients can start it with a single command.