deepwiki

What This Skill Does

The DeepWiki MCP skill provides OpenClaw with the ability to query public GitHub repositories using AI-grounded insights. Instead of manual code exploration, this skill interfaces with the DeepWiki documentation service to retrieve explanations, architectural overviews, and source code analysis. It allows you to treat any public open-source project as a searchable, conversational knowledge base. Whether you are debugging an unfamiliar framework or trying to understand how a core module interacts with its dependencies, DeepWiki provides structured, reference-backed answers.

Installation

To install the DeepWiki MCP skill, use the ClawHub command-line interface provided in the OpenClaw environment:

clawhub install openclaw/skills/skills/chunhualiao/deepwiki-mcp

Ensure that your OpenClaw agent has permission to execute shell scripts located within the skill directory, as the agent relies on the deepwiki.sh wrapper to communicate with the MCP endpoint at https://mcp.deepwiki.com/mcp.

Use Cases

• Architecture Analysis: Understand the design patterns or module boundaries of a repository you have never worked on before.
• Quick Debugging: Get pointers to specific code paths or files that implement a feature when you encounter errors.
• Learning New Libraries: Use DeepWiki as an interactive tutorial to explore how a library handles specific tasks, such as authentication or state management.
• Documentation Discovery: Use the topics action to map out the available documentation within a complex repository before diving into the code.

Example Prompts

1. "How does the request routing mechanism work in the expressjs/express repository?"
2. "Look up the authentication flow in the supabase/auth-js codebase."
3. "Can you list the available documentation topics for the reactjs/react repo?"

Tips & Limitations

• Scope: This skill is strictly for public GitHub repositories. Private repositories are not supported.
• Latency: DeepWiki's indexing is not real-time. If you are inquiring about a commit made within the last 24-48 hours, the information may not yet be reflected in the service.
• Efficiency: While docs provides full content, it is often better to use the ask action for targeted, high-precision answers. docs can produce large outputs that may clutter your chat history.
• Fallback: If the provided helper script fails, you can perform direct queries against the MCP endpoint via standard curl commands, provided the agent environment has network access.