Deepwiki MCP Server

Welcome to the documentation for the Deepwiki MCP Server, an unofficial tool designed to fetch content from deepwiki.com and convert it into clean, LLM-readable Markdown. This server acts as a bridge, allowing AI agents and other developer tools to easily consume and understand documentation hosted on Deepwiki.

⚠️ IMPORTANT NOTICE: As noted in the project's README, this server is currently non-functional because DeepWiki has restricted web scraping. We recommend using the official DeepWiki MCP server provided by Devin AI until this unofficial version is updated. This documentation is provided for archival and educational purposes.

Overview

The primary goal of this project is to provide a seamless interface for programmatic access to Deepwiki content. It exposes a set of tools through the Model Context Protocol (MCP) that can be triggered with simple commands. The server handles the complexity of web crawling, HTML parsing, content sanitization, and Markdown conversion, returning a clean and structured result.

Key Features

  • MCP Tool Integration: Exposes a deepwiki_fetch tool that can be called by any MCP-compatible client (like Cursor).
  • Flexible Input: Accepts full Deepwiki URLs, shorthand owner/repo formats, and even single library keywords, which are automatically resolved to the correct repository using NLP and the GitHub API.
  • Content Sanitization: Intelligently strips away unnecessary HTML elements like headers, footers, navigation, scripts, and ads, focusing purely on the core documentation content.
  • Markdown Conversion: Transforms the sanitized HTML into high-quality, GitHub-Flavored Markdown, making it ideal for language models.
  • Multiple Output Formats: Delivers content either as a single, aggregated document or as a collection of individual pages, depending on the user's need.
  • Configurable Crawling: Offers control over crawling depth, request concurrency, and timeouts through environment variables.
  • Multiple Transport Layers: Can be run over standard I/O (stdio), HTTP, or Server-Sent Events (SSE) for flexible integration.