wiki_question

A powerful tool for asking questions about GitHub repositories using DeepWiki's AI analysis. This tool provides intelligent answers based on deep codebase analysis, with the ability to retrieve specific code snippets and file contents. DeepWiki's AI examines the entire repository structure, code patterns, and relationships to provide accurate, context-aware answers.

When to use this tool:

• Understanding how a specific feature or system works in a repository

• Getting explanations of complex code implementations

• Finding where specific functionality is implemented

• Understanding architectural decisions and patterns

• Retrieving exact code snippets that answer your questions

• Getting full file contents for detailed analysis

• Following up on previous questions with cached results

• Going deeper on existing queries for more detailed analysis

• Sending follow-up questions to continue conversations

⚠️ Important: Do NOT ask for specific files in new questions (e.g., "show me auth.js"). Instead ask conceptual questions and use queryId to retrieve referenced files.

Key features:

• Ask natural language questions about any GitHub repository

• Retrieve exact code snippets that DeepWiki referenced in its answer

• Get complete file contents or specific line ranges

• Use cached results for efficient follow-up queries

• Optional deep research mode for comprehensive analysis (3-15 minutes) - consider running in tmux for background processing

• Go deeper functionality for existing queries to get more detailed analysis

• Follow-up questions to continue conversations with existing queries (supports deep research)

• Save responses to files for documentation or sharing

• Flexible output control to show only what you need

CRITICAL: Understanding when to use queryId vs new question:

• USE queryId when: Getting references/files from previous response, asking follow-ups about the same topic, going deeper, sending follow-up questions

• USE new question when: Asking something completely different, no previous query exists

Usage examples:

1. Ask a new question: question: "How does the authentication system work?", includeReferencesList: true

2. Get specific references from previous query: queryId: "query-12345", referencesNumbers: [1, 2], includeAnswer: false

3. Get full file content from previous query: queryId: "query-12345", contextFiles: ["src/auth/login.ts"], includeAnswer: false

4. Get specific line ranges: queryId: "query-12345", contextFiles: ["src/index.ts"], contextRanges: {"src/index.ts": {"start": 100, "end": 150}}

5. Deep research with file saving: question: "Explain the entire data flow architecture", useDeepResearch: true, saveToFile: "save-and-show"

6. Go deeper on existing query: queryId: "query-12345", goDeeper: true

7. Send follow-up question (shows only new response): queryId: "query-12345", followUpQuestion: "Can you explain how error handling works?", includeFullConversation: false

8. Send follow-up with deep research (shows full conversation): queryId: "query-12345", followUpQuestion: "What are the security implications?", useDeepResearch: true, includeFullConversation: true

9. Get all code snippets without the explanation: queryId: "query-12345", referencesAll: true, includeAnswer: false, includeReferencesList: false

Important notes:

• referencesNumbers gets the EXACT snippets DeepWiki used (not full files)

• contextFiles gets COMPLETE file contents (not just snippets)

• Always check the queryId in responses for follow-up queries

• Deep research mode provides more comprehensive analysis but takes significantly longer - consider using tmux for background execution

• goDeeper creates a new query with deeper analysis on existing queries - returns a new queryId

• followUpQuestion continues existing conversation with same queryId, supports deep research

• includeFullConversation=false (default) shows only new response; true shows complete conversation

• Use queryId instead of question when following up on previous responses