Deepwiki MCP Server

What is Deepwiki MCP Server?

Deepwiki MCP Server is an unofficial MCP server that accepts a Deepwiki URL via MCP, crawls all relevant pages, converts them to Markdown, and returns either a single aggregated document or a structured list of pages. It is intended for developers using MCP-compatible clients such as Cursor to fetch documentation from Deepwiki.

How to use Deepwiki MCP Server?

Configure the server in your MCP client (e.g., add to .cursor/mcp.json) with npx -y mcp-deepwiki@latest as the command. Then invoke the tool deepwiki_fetch with parameters url (required), mode (optional: "aggregate" or "pages"), and maxDepth (optional). You can also use natural language prompts like deepwiki fetch how can i use gpt-image-1 with "vercel ai" sdk.

Key features of Deepwiki MCP Server

• Domain safety: only processes URLs from deepwiki.com
• HTML sanitization strips headers, footers, navigation, scripts, and ads
• Link rewriting adjusts links to work in Markdown
• Supports multiple output formats: one document or structured pages
• Fast crawling with adjustable concurrency and depth
• Provides progress events during crawling

Use cases of Deepwiki MCP Server

• Fetch complete documentation for a Deepwiki repository as a single Markdown document
• Retrieve structured, page‑wise Markdown for a Deepwiki repository
• Quickly answer specific questions about a library using its Deepwiki documentation
• Integrate Deepwiki documentation into AI‑assisted coding workflows (e.g., Cursor)

FAQ from Deepwiki MCP Server

What does Deepwiki MCP Server do that the official DeepWiki MCP server does not?

This server is an unofficial alternative. As of the notice in the README, it is currently not working because DeepWiki has cut off the ability to scrape it. The official DeepWiki MCP server is recommended at https://docs.devin.ai/work-with-devin/deepwiki-mcp.

What are the runtime requirements?

The server requires Node.js and is run via npx (npx -y mcp-deepwiki@latest). It can also be run from source after cloning the repository and running npm install and npm run build.

Where does the data come from and how is it processed?

The server only accepts URLs from the deepwiki.com domain. It crawls the given repository, sanitizes the HTML (stripping headers, footers, navigation, scripts, and ads), and converts the content to Markdown. It does not store data externally.

What are the known limitations?

The server is currently broken because DeepWiki has blocked scraping. By design, it only works with deepwiki.com domains. For large repositories, timeout errors may occur; these can be mitigated by adjusting environment variables such as DEEPWIKI_REQUEST_TIMEOUT and DEEPWIKI_MAX_CONCURRENCY.

What transports and authentication are supported?

The server supports stdio transport (default for MCP) and HTTP transport (via --http flag). No authentication is required or mentioned. It can be configured with environment variables for concurrency, timeout, and retry settings.