MCP Windows, developed by Sbroenne, is a free server tool built with C# and requiring the .NET 10 runtime. It connects to Windows applications and the operating system, utilizing the Windows UI Automation API. This platform-specific MCP server extends automation capabilities to Windows desktop environments, including content within Edge and Chrome browsers. It is distributed as a VS Code Extension, a shared plugin bundle for GitHub Copilot CLI and Claude Code, or a standalone executable from GitHub Releases, catering to various deployment needs for AI agents and developers.

What does MCP Windows connect to?

MCP Windows connects directly to Windows applications and the Windows operating system itself. It achieves this by using the Windows UI Automation API. This server tool specializes in Windows desktop automation, extending its reach to include content within popular browsers such as Edge and Chrome. Its core function is to enable AI agents and other clients to interact with and control the Windows environment programmatically, making it a central component for desktop-focused automation tasks.

Core features and tools

The tool exposes several callable functions through its "wincli" command-line entry point. These include window_management, which helps in finding and organizing windows. ui_click and ui_type enable interaction with user interface elements. For locating specific elements, ui_find is available. Additionally, screenshot_control provides visual analysis capabilities, and mouse_control offers precise cursor manipulation. These tools collectively allow for detailed control over the Windows desktop experience.

Who is it for?

MCP Windows targets AI agents that need to interact with Windows, developers creating such agents, and users who wish to automate Windows applications using natural language commands. It provides the underlying mechanisms for AI to perform tasks on a Windows desktop. This includes automating routine operations or complex workflows across different applications. Skip this if you are not working with Windows desktop automation or AI agent integration.

How to get started

Authentication for LLM integration tests is granted through GitHub, requiring either a GITHUB_TOKEN environment variable or an existing "gh" login. Local execution implies support for the "stdio" transport protocol, while "streamable HTTP" is a current preferred remote transport for MCP servers.

Use cases

• Controlling Windows applications by name, such as "Click the Save button in Notepad."
• Automating browser content, like "Open my banking site in Edge and click Transfer, then fill out the form on this page in Chrome."
• Performing multi-monitor window management, for example, "Move this window to my second monitor."
• Providing a fallback mechanism for games and custom controls using screenshot, mouse, and keyboard actions.

MCP Windows provides a specialized server for automating Windows desktop and browser tasks using AI agents. Its C# foundation and .NET 10 requirement enable direct interaction with the Windows UI Automation API. The tool supports various AI clients and offers essential controls for managing applications, UI elements, and even fallback actions. Try automating a simple task like "Click the Save button in Notepad" to begin using its capabilities.