Access LLM Documentation Locally or Remotely

DocsMCP

A Model Context Protocol (MCP) server that provides documentation access to LLMs.

Overview

DocsMCP enables Large Language Models (LLMs) to access and query documentation from specified sources, whether from local files or remote URLs. It uses the Model Context Protocol (MCP) to facilitate communication between the LLM and documentation sources.

Usage

Cursor MCP Configuration

You can also configure DocsMCP in your Cursor project by creating a .cursor/mcp.json file:

{
  "mcpServers": {
    "docs-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "docsmcp",
        "'--source=Model Context Protocol (MCP)|https://modelcontextprotocol.io/llms-full.txt'"
      ]
    }
  }
}

This configuration allows Cursor AI to use the documentation MCP server automatically when you open your project.

Note:

When specifying a source that contains spaces, ensure to wrap the entire string in quotes. For example:
'--source=Model Context Protocol (MCP)|https://modelcontextprotocol.io/llms-full.txt'

VS Code MCP Configuration

You can configure DocsMCP in VS Code by adding a configuration to your .vscode/mcp.json file:

{
  "servers": {
    "documentation-mcp-server": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "docsmcp",
        "--source=Model Context Protocol (MCP)|https://modelcontextprotocol.io/llms-full.txt"
      ]
    }
  }
}

This configuration allows VS Code extensions that support MCP to use the documentation server automatically.

Available Tools

The MCP server provides two main tools:

getDocumentationSources

Lists all available documentation sources that have been configured.

getDocumentation

Fetches and parses documentation from a given URL or local file path.

Parameters:

• url: The URL or file path to fetch the documentation from