Microsoft recently released a MCP (Model Context Protocol) server that gives AI assistants direct access to their official documentation. This means Claude can search through Microsoft Learn, Azure docs, and other official sources in real-time. Here’s how to set it up in Claude Code.

What You’re Getting

The Microsoft Docs MCP server provides a single tool called microsoft_docs_search that:

  • Searches across all official Microsoft/Azure documentation
  • Returns up to 10 relevant content chunks (max 500 tokens each)
  • Includes article titles, URLs, and self-contained content excerpts
  • Uses semantic search to find contextually relevant docs

The Setup

First, check what MCP commands Claude Code supports:

claude mcp

The command we need to enter is as follows:

claude mcp add --transport http MicrosoftDocs https://learn.microsoft.com/api/mcp

Verifying the Connection

After adding it, check that it’s connected:

claude mcp list

You should see something like:

  • Status: ✔ connected
  • URL: https://learn.microsoft.com/api/mcp
  • Capabilities: tools
  • Tools: 1 tools

You can also inspect what tools are available:

claude mcp get MicrosoftDocs

Using It in Practice

Once connected, you can:

  • Just ask Claude Code questions about Microsoft tech - it should automatically use the tool when relevant. You can always nudge it to use the Microsoft Docs MCP server.
  • Explicitly request searches: “Search Microsoft docs for Azure Container Apps”

The search is semantic, so natural language queries work perfectly. Results are optimised for AI consumption - you get clean, relevant chunks rather than entire documents.

Connecting Claude Desktop to the Microsoft Docs MCP

You can also add the Microsoft Docs MCP server in Claude Desktop by clicking on the Connect Apps button on that chat screen, and then choosing Add integration. Fill out a name of your choice, and add the URL https://learn.microsoft.com/api/mcp, it’s that simple.