Microsoft Docs

The Microsoft Docs MCP Server connects AI assistants directly to Microsoft’s official documentation through the Model Context Protocol.

When you ask a question, the MCP can ground its answer in the latest, most accurate information straight from the source.

Features

• 🔍 Semantic Search: Advanced vector search finds contextually relevant documentation across all Microsoft sources
• 📚 Comprehensive Coverage: Access Microsoft Learn, Azure documentation, Microsoft 365 docs, and other official sources
• ⚡ Real-time Updates: Always current with the latest published Microsoft documentation
• 🎯 Optimized Content Delivery: Returns up to 10 high-quality content chunks (max 500 tokens each) with article titles and URLs
• 🔌 Universal Compatibility: Works with any IDE or tool supporting the Model Context Protocol
• ☁️ Cloud-hosted: No local installation required – connects via remote HTTP endpoint
• 🤖 AI-optimized: Content chunking designed specifically for AI assistant consumption

Use Cases

• Azure Development Workflows: Quickly reference Azure CLI commands, ARM template syntax, and service configurations while writing infrastructure code without opening multiple browser tabs
• Microsoft 365 Integration Projects: Access Graph API documentation, SharePoint development guides, and Teams app development resources directly within your coding environment
• Troubleshooting Complex Microsoft Technologies: Get instant access to official troubleshooting guides and best practices when debugging issues with Microsoft services or tools

How to Use It

The Microsoft Docs MCP Server operates as a remote endpoint, so no local installation is required. You simply configure your MCP-compatible client to connect to the service.

Server Endpoint:

https://learn.microsoft.com/api/mcp

Basic Configuration (JSON):

{
  "microsoft.docs.mcp": {
    "type": "http",
    "url": "https://learn.microsoft.com/api/mcp"
  }
}

IDE-Specific Installation

VS Code & VS Code Insiders:

1. Click the one-click install button in the official repository
2. The extension automatically configures the MCP connection
3. Restart VS Code to activate the server

Cursor:

1. Use the provided Cursor install button for automatic setup
2. The server appears in your available MCP tools list
3. No additional configuration needed

Visual Studio:

1. Manually add the configuration to your MCP settings
2. Use "type": "http" in your configuration
3. Restart Visual Studio to recognize the new server

Roo Code:

1. Add the server configuration manually
2. Use "type": "streamable-http" for Roo Code compatibility
3. Save and restart the application

Using the MCP Server

1. Activate Agent Mode: In VS Code, open GitHub Copilot and switch to Agent mode
2. Verify Connection: Check that “Docs MCP Server” appears in your tools list
3. Query Documentation: Use prompts like “What are the az cli commands to create an Azure container app according to official Microsoft Learn documentation?”
4. Review Results: The assistant retrieves and references official Microsoft documentation in its responses

FAQs

Q: Why do I get a “405 Method Not Allowed” error when I paste the URL into my browser?
A: The URL is an API endpoint designed for programs like your IDE’s AI assistant to communicate with, not for direct browser access. This error is expected and means the endpoint is working correctly, just not for web browsers.

Q: The tool isn’t appearing in GitHub Copilot. What should I do?
A: First, restart VS Code. That often resolves discovery issues. Also, confirm you are in “Agent” mode in the Copilot Chat window, as tools are only available in this mode. Finally, double-check your MCP JSON configuration for any typos.

Q: What specific tool does this server provide?
A: It provides one main tool: microsoft_docs_search. This tool takes a single string parameter called query to perform a semantic search against the documentation.