lh-deepwiki_skill

Overview

This skill queries the DeepWiki MCP service over Streamable HTTP to retrieve GitHub repository documentation, wiki structure, and AI-driven Q&A. It fixes older protocol issues by using the /mcp endpoint, updates parameter names, and increases default timeouts for complex queries. The skill supports public GitHub repos without authentication and is maintained as an open-source integration.

How this skill works

The skill sends Streamable HTTP requests to https://mcp.deepwiki.com/mcp, using the repoName parameter to identify the target public GitHub repository. It can fetch the wiki tree, read specific documentation files, and invoke an AI Q&A endpoint that streams answers for natural-language queries. Timeouts and parameter handling are optimized to handle larger docs and longer answer generation.

When to use it

• You need AI-powered answers grounded in a public GitHub repository's documentation or wiki.
• You want to explore or export a repository's documentation structure and file contents.
• You want to integrate repository Q&A into a CLI or automation workflow without API keys.
• You need streamed responses for long or interactive question sessions.
• You are working with archived or versioned documentation and need a read-only view.

Best practices

• Provide owner/repo in the form owner/repo when calling the skill to avoid lookup errors.
• Use the structure command first to find relevant paths before requesting file contents.
• Allow up to 120 seconds for complex questions; increase client-side timeout if needed.
• Only target public GitHub repositories; private repos are not supported.
• Treat responses as read-only archival data and validate critical facts against source files.

Example use cases

• Ask natural-language questions about a project's purpose and design by querying its docs.
• Export a repository's wiki tree to catalog archived documentation across versions.
• Read a specific documentation page content for automated summarization or migration.
• Integrate repository Q&A into developer tooling or chatbots for on-demand docs lookup.
• Troubleshoot documentation gaps by querying topics the wiki does not clearly cover.