Windows MCP Server
Windows automation that actually works.
Let AI Control Windows Apps — By Name, Not Pixels
Windows MCP Server gives your AI assistant direct access to Windows applications through the Windows UI Automation API. The same API screen readers use to read buttons, menus, and text fields.
Your AI says “click Save” and the server finds the Save button by name. No screenshots. No pixel parsing. No coordinate guessing.
What You Can Do
Ask your AI assistant to control any Windows application:
- “Click the Save button in Notepad”
- “Type my email in the login field”
- “Toggle Dark Mode in Settings”
- “Move this window to my second monitor”
- “Read the error message from that dialog”
Works with GitHub Copilot, Claude Desktop, Cursor, and any MCP client.
Why This Approach
Most automation tools take screenshots and ask vision models to find buttons in the pixels. That approach is slow, expensive, and breaks when windows move or themes change.
Windows MCP Server asks Windows directly: “What buttons exist?” Windows knows. It’s deterministic. Same command works every time, regardless of DPI, theme, or resolution.
Tested with Real AI Models
Tool descriptions that seem clear to humans often confuse AI. Parameters get misunderstood. Actions get skipped.
We test every tool with real AI models (GPT-4.1, GPT-5.2) using agent-benchmark. 54 automated tests. 100% pass rate required for release.
If the AI can’t use it correctly, we fix the tool — not the prompt.
View latest LLM test results →
Quick Start
VS Code (Recommended)
Install from VS Code Marketplace →
Other MCP Clients
Download from GitHub Releases. Add to your MCP config:
{ "servers": { "windows": { "command": "path/to/Sbroenne.WindowsMcp.exe" } } }
Tools
| Tool | What It Does |
|---|---|
ui_click |
Click buttons, checkboxes, menu items by name |
ui_type |
Type into text fields |
ui_find |
Discover elements in a window (with timeout/retry) |
ui_read |
Read text (with OCR fallback) |
file_save |
Save files via Save As dialog |
screenshot_control |
Get element metadata (image optional) |
window_management |
Find, activate, move, resize windows |
mouse_control |
Coordinate-based clicks (fallback for games) |
keyboard_control |
Hotkeys and key sequences |
app |
Launch applications |
Key Features
🧠 Semantic UI Access
Find elements by name, type, or ID — not coordinates. Works regardless of DPI, theme, resolution, or window position.
🧪 LLM-Tested
Every tool tested with real AI models before release. 54 automated tests across 7 scenarios. 100% pass rate required.
💻 Broad App Support
Tested against classic Windows apps, modern Windows 11 apps, and Electron apps (VS Code, Teams, Slack). Same commands work across all.
📺 Multi-Monitor
Full support for multiple displays with per-monitor DPI scaling. Move windows between monitors, capture any screen.
🔄 Full Fallback
Screenshot + mouse + keyboard for games and custom controls. Annotated screenshots return element metadata — image omitted by default to save tokens.
⚠️ Caution
This MCP server controls your Windows desktop. Use responsibly.
Related Projects
- agent-benchmark — LLM agent testing framework (powers our integration tests)
- Excel MCP Server — AI-powered Excel automation
- OBS Studio MCP Server — AI-powered streaming control