Windows-Use: LLM-Powered Windows GUI Automation via System APIs

Windows-Use enables large language models to directly control the Windows graphical user interface by leveraging system-level UI automation interfaces instead of screenshot-based visual guessing. It retrieves control names and positions from the accessibility API, making interactions more reliable than purely visual approaches. Tasks are described in natural language, and the model autonomously plans and executes window opening, clicking, typing, and dragging. It also supports running system commands, reading and writing files, switching virtual desktops, and voice input/output. Compatible with Claude, GPT, Gemini, and local models via Ollama, it runs on Windows 7 through 11. The project is available on GitHub.