Overview

Model Context Protocol (MCP) is a standardization of how AI assistants interact with APIs, and ReadMe is bringing this capability to your developer hub. With MCP servers, you can convert your API documentation into a structured, resource that AI assistants can understand and interact with programmatically.

Key Features

• OpenAPI Integration: Generate an MCP server from your existing OpenAPI specification.
• AI assistants can connect to your MCP server to:
  • Read and understand your OpenAPI spec.
  • Execute API calls.
  • Search docs with Owlbot.
• AI Tools Included:
  • OpenAPI Tools:
    • Execute_request_body - Make API calls directly from your specification
    • Get_endpoint - Pull in detailed endpoint information on demand
    • Get_request_body - Access structured request parameters
    • Get_response_schema - Understand what your API returns
    • List_endpoints - Browse all available API endpoints
    • List_security_schemes - Access authentication requirements
    • Search_schema - Find exactly what you need in your API spec
  • Documentation Search:
    • search_documentation - Leverage Owlbot to search your entire knowledge base

How It Works

We create a dedicated MCP server that connects to your OpenAPI specification and Owlbot search functionality. This creates a bridge between your API documentation and AI assistants, making your API instantly more accessible and understandable to AI tools.

Getting Started

You can share your MCP URL with your developers, and they can connect their AI assistants and tools directly to your API.

ReadMe hosts a remote MCP server at https://docs.readme.com/mcp. Configure your AI development tools to connect to this server using your ReadMe API key.

{
  "mcpServers": {
    "ReadMeAPI": {
      "url": "https://docs.readme.com/mcp"
    }
  }
}

💁‍♂️

If you want to execute API requests you will need to pass in an api key. You can use mcp-remote to do this. Update your mcp config to look like this

mcp_config.json

{
  "mcpServers": {
    "readmeAPI": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://docs.readme.com/mcp",
        "--header",
        "Authorization: ${AUTH_HEADER}"
      ],
      "env": {
        "AUTH_HEADER": "Bearer rdme_xn..."
      }
    }
  }
}

Note: Cursor and Claude Desktop (Windows) have a bug where spaces inside args aren't escaped when it invokes npx, simply remove the whitespace after 'Authorization:' to work around this bug.

Testing Your MCP Setup

Once configured, you can test your MCP server connection:

1. Open your AI editor (Cursor, VS Code, etc.)
2. Start a new chat with the AI assistant
3. Ask about ReadMe - try questions like:
   • "How do I [common use case]?"
   • "Show me an example of [API functionality]"
   • "Create a [integration type] using ReadMe"

The AI should now have access to your ReadMe account data and documentation through the MCP server.

Auto-generating MCP documentation for end-users

Once you have activated your MCP server, can automatically generate access instructions for your end-users by clicking the "Generate MCP Template" button. This creates a new unpublished doc called "MCP" in your Guides showing how to connect to your MCP server in tools such as Cursor and Claude Desktop. You can find the doc at the bottom of your Guides or API Reference in a new category called "MCP SERVER."