DeepWiki, from Cognition (the team behind Devin), generates a wiki for public GitHub repositories. Its MCP server lets an agent read that wiki and ask questions about a codebase it has never cloned — with no API key at all.
https://mcp.deepwiki.com/mcp
Claude Haiku 4.5
Chat with 60+ AI models on the same workflow — switch to a different model mid-conversation and re-run the same prompt, or use Compare mode to put several side-by-side and balance quality vs. cost.
No token required
How models use it and what it is built for.
The server exposes three tools. read_wiki_structure lists the documentation topics DeepWiki has generated for a repository, read_wiki_contents returns the documentation itself, and ask_wiki_question takes a free-form question and answers it from the repository’s wiki. Each tool takes a repository name in owner/repo form. The practical use is the question you would otherwise answer by cloning a dependency and reading source for an hour: how does this library handle retries, where is auth configured, which module owns the cache. Because the answers come from generated documentation rather than the live source, they describe the repository as of DeepWiki’s last index, and on a fast-moving project that can lag the default branch.
Typical tools an AI model can call. Exact names vary by version.
Taken from the official DeepWiki documentation — see DeepWiki MCP documentation for the full reference.
Examples
Streamable HTTP
https://mcp.deepwiki.com/mcpThe current endpoint. Use this unless your client only speaks SSE.
Legacy SSE
https://mcp.deepwiki.com/sseFor older clients that predate Streamable HTTP.
Client configuration
Any MCP client
No credential block needed — just the URL.
{
"mcpServers": {
"deepwiki": {
"type": "http",
"url": "https://mcp.deepwiki.com/mcp"
}
}
}Copy any of these into MCP Agent Studio after connecting.
How does modelcontextprotocol/typescript-sdk handle session ids on the Streamable HTTP transport?
List the documentation topics for vercel/next.js, then explain how the app router caches fetches.
In facebook/react, which package implements the reconciler and how is it split up?
Compare how two of my dependencies handle retries, and tell me which one is configurable per request.
This is not a single-model product: you get the same MCP connection with 60+ models (Claude, GPT, Gemini, DeepSeek, open-weight, and more), you can switch mid-conversation, and you can open Compare mode to run the same prompt against multiple models at once. The card above is a suggested starting point for this server — not the only choice.
Default pick for DeepWiki
Claude Haiku 4.5
Three tools with obvious names, so choosing the right one is easy. The work is phrasing a good question and summarising the answer, which Haiku does well and cheaply.
Listing tools proves the server is reachable, not that a model can work with it. Evals go further: they read every tool on the server, write a test suite from its real schemas, and run it — code decides pass/fail on the responses (schema conformance, error codes, pagination, result caps) while a scoring model grades plain-English tasks driven through the tools.
Get a pass/fail report per tool with the evidence behind each verdict — and replay the same suite after every schema change. Destructive tools are excluded from the run.
Run evalsOpen MCP Agent Studio with the connection pre-filled. Add your token, pick any of 60+ models, and start chatting — no install required.
Open Agent StudioCommon questions about connecting, scoping and using it safely.
Yes. The hosted endpoint at https://mcp.deepwiki.com/mcp needs no account or API key, and it answers an initialize handshake unauthenticated. It covers public GitHub repositories.
Three: read_wiki_structure lists the documentation topics for a repository, read_wiki_contents returns that documentation, and ask_wiki_question answers a free-form question about the repository. Each takes the repository as owner/repo.
It answers from DeepWiki’s generated wiki for the repository, not from a fresh clone. On an actively developed project, check anything version-sensitive against the source or the release notes.
Not through this free endpoint, which is for public repositories. Private repository access is part of Devin, Cognition’s paid product.
Context7 serves version-specific library documentation that maintainers publish. DeepWiki generates documentation from the repository itself and lets you ask questions about the code, so it also covers projects with thin or no docs. They work well side by side.