GitHub DeepWiki MCP Server

GitHub DeepWiki MCP Server

Q: これは結局何なの？

A: 本家deepwikiがリポジトリオーナー/リポジトリ名入力であるせいでなかなかLLMが使用してくれない問題を解決するためにリポジトリURLで呼び出せるようにしたやつ。ついでにREADMEも取得できる。

A comprehensive GitHub repository documentation server built with FastMCP.

This MCP server provides detailed GitHub repository information by integrating with mcp.deepwiki.com and GitHub API to fetch repository wiki content, README files, and enable intelligent question-answering about repositories.

Features

• Comprehensive Repository Analysis: Fetches wiki information, README files, and architectural details
• Intelligent Q&A: Ask detailed questions about repositories and get structured answers
• Multi-source Integration: Combines data from mcp.deepwiki.com and GitHub API
• Structured Output: Organizes repository information in a readable format
• Fallback Support: Works even when wiki or README information is partially unavailable

Development

To get started, clone the repository and install the dependencies.

Usage

Start the server

If you simply want to start the server, you can use the start script.

However, you can also interact with the server using the dev script.

This will start the server and allow you to interact with it using CLI.

Available Tools

The server provides two main tools:

• get-github-wiki: Retrieves comprehensive GitHub repository wiki information and README files
  • repoUrl (required): GitHub repository URL (e.g., https://github.com/facebook/react)
• ask-github-question: Ask detailed questions about GitHub repositories
  • repoUrl (required): GitHub repository URL
  • question (required): Detailed question about the repository (include context, background, and specific requirements)

Available Resources

• DeepWiki MCP Server Information: General server information and usage guidelines

Available Prompts

• github-analysis: Template for analyzing GitHub repositories
  • repoUrl (required): GitHub repository URL
  • analysisType (optional): Type of analysis (architecture, features, tech stack, etc.)

Testing

The unit tests verify URL validation, success/failure cases, error handling, and formatting. The MCP Inspector CLI tests verify actual server functionality including tool listings, resource reading, and real tool execution.

Linting

Having a good linting setup reduces the friction for other developers to contribute to your project.

This project uses Prettier, ESLint and TypeScript ESLint to lint the code.

Formatting

Use npm run format to format the code.

External Service Integration

This server integrates with the following external services:

mcp.deepwiki.com

Connects via HTTP Streamable transport to access:

• read_wiki_structure: Repository documentation structure
• read_wiki_contents: Detailed wiki content
• ask_question: Repository question-answering

GitHub API

Uses GitHub REST API v3 to fetch:

• README files (README.md, README, readme.md, readme)
• Repository metadata and structure
• Rate limit: 60 requests per hour for public repositories

Configuration

The server is designed to work out of the box with no configuration required. It dynamically fetches content from both mcp.deepwiki.com and GitHub API.

Technical Stack

• FastMCP: MCP server framework
• TypeScript: Type safety and code quality
• Zod: Schema validation
• Vitest: Testing framework
• ESLint + Prettier: Code quality management

How to use with Claude Desktop?

Follow the guide https://modelcontextprotocol.io/quickstart/user and add the following configuration: