Context7 vs DeepWiki MCP

When to pick Context7

• You are implementing a library API and need documentation for the version your project uses. Include that version in your prompt so Context7 can match the documentation before your agent writes code.
• Your agent suggests an API that may belong to an older release. Use query-docs to retrieve documentation and code examples for the library, then check the proposed implementation against those results.
• You know the package name but need to find its documentation entry. resolve-library-id takes the library name and your task, using the task to rank matching results. If you already know the Context7 ID, you can skip this step.
• You need examples in a particular programming language. The optional language parameter on query-docs prioritizes examples in that language, alongside the library ID and the question you supply.

When to pick DeepWiki MCP

• You are getting familiar with a public GitHub repository. Start with read_wiki_structure to see which topics its documentation covers before deciding what to investigate.
• You want repository documentation inside your coding agent. read_wiki_contents returns wiki text that the agent can use while discussing the project.
• You have a specific question about how a public repository works. ask_question returns an AI-generated answer grounded in that repository's context, giving you a question-and-answer route alongside wiki reading.
• You want to connect a repository documentation service without creating credentials or running a local server package. DeepWiki MCP offers a free remote endpoint without authentication. Its scope is public repositories; private repository access belongs to the separate Devin MCP service.

Main differences

• Context7 organizes retrieval around a library and can match a version named in your prompt. DeepWiki organizes documentation around a public GitHub repository. Choose according to whether you need API usage guidance or an explanation of a particular project.
• Context7 follows a library-resolution and documentation-query workflow. DeepWiki separates topic discovery, wiki reading and questions into distinct tools. Reading the structure first lets you narrow your investigation before requesting prose.
• Context7 can run as a local Node process over stdio or connect through its hosted HTTP endpoint. DeepWiki runs remotely over Streamable HTTP. Its older SSE endpoint is being deprecated, so use /mcp for a new connection.
• Context7 recommends a free API key for higher rate limits and documents a bearer authorization header for manual hosted connections. DeepWiki requires no authentication and describes the service as free. The supplied Context7 documentation does not specify a full pricing schedule.
• Context7 returns documentation passages and code examples for your agent to use. DeepWiki can return wiki content or generate an answer through ask_question. This changes whether you request reference material or a repository-grounded response.

Setup

For Context7, use npx -y @upstash/context7-mcp as the local server command in your client's stdio configuration. For guided setup, run npx ctx7 setup and select MCP mode; it authenticates through OAuth and generates an API key. A manual hosted connection uses https://mcp.context7.com/mcp with Authorization: Bearer YOUR_API_KEY.

For DeepWiki MCP in Claude Code, run claude mcp add -s user -t http deepwiki https://mcp.deepwiki.com/mcp. In Cursor and most other clients, configure https://mcp.deepwiki.com/mcp in the url field. Devin Desktop also accepts serverUrl; using an unsupported field can cause the client to ignore the server. No API key is needed.