Quick Start
1
Open DeepWiki Interface
Navigate to http://localhost:3000 in your browser.You should see the main DeepWiki interface with:
- Repository URL input field
- Model provider selection
- Generate Wiki button
The interface loads without errors and shows all input fields.
2
Choose a Test Repository
For your first wiki, start with a small, well-structured repository:
Start with repositories under 100 files for faster initial generation and testing.
3
Configure AI Model
Select your AI model provider and specific model:
Google Gemini (Recommended for Beginners)
Google Gemini (Recommended for Beginners)
- Provider: Google
- Model:
gemini-2.0-flash
(default) - Why: Fast, reliable, generous free tier
- Best for: General documentation, good balance of speed and quality
OpenAI GPT Models
OpenAI GPT Models
- Provider: OpenAI
- Model:
gpt-4o
(default) - Why: High-quality, detailed documentation
- Best for: Complex codebases requiring detailed analysis
OpenRouter (Multiple Models)
OpenRouter (Multiple Models)
- Provider: OpenRouter
- Model:
openai/gpt-4o
oranthropic/claude-3.5-sonnet
- Why: Access to multiple providers through one API
- Best for: Experimenting with different models
Ensure you have valid API keys configured for your chosen provider.
4
Generate the Wiki
Click the “Generate Wiki” button and watch the progress:
- Repository Analysis: DeepWiki clones and analyzes the repository structure
- Code Processing: Creates embeddings for intelligent code understanding
- AI Generation: Generates documentation using your selected AI model
- Diagram Creation: Creates Mermaid diagrams for visualization
- Wiki Assembly: Organizes everything into a structured wiki
Generation typically takes 30 seconds to 2 minutes for small repos, 5-10 minutes for larger ones.
Understanding the Generated Wiki
Once generation completes, you’ll see your interactive wiki with several key sections:Wiki Structure
What you’ll see:
- Repository summary and purpose
- Key features and functionality
- Architecture overview
- Main technologies used
- Project description
- Installation instructions
- Usage examples
- Contributing guidelines
Navigation Features
Wiki Tree View
Hierarchical navigation showing all wiki pages and sections
Search Functionality
Full-text search across all generated documentation
Cross-References
Links between related pages and code sections
File References
Direct links to source files and line numbers
Using the Ask Feature
The Ask feature lets you chat with your repository using RAG (Retrieval Augmented Generation):1
Access Ask Interface
Look for the “Ask” or chat interface in your generated wiki.
2
Ask Questions
Try these example questions:
3
Review Responses
The AI will provide answers based on your actual repository code, including:
- Direct code references
- File locations
- Relevant context
- Example usage
Deep Research Mode
For complex analysis, try the Deep Research feature:1
Enable Deep Research
Toggle the “Deep Research” switch in the Ask interface.
2
Ask Complex Questions
Deep Research is perfect for questions like:
- “Analyze the security implementation across the entire codebase”
- “How is performance optimized in this application?”
- “Explain the complete data flow from user input to database”
3
Review Multi-Turn Analysis
Deep Research conducts multiple investigation rounds:
- Initial Analysis: Basic research and findings
- Deep Dive: Detailed investigation of specific aspects
- Cross-References: Analysis of relationships and dependencies
- Final Synthesis: Comprehensive conclusion with recommendations
Customizing Your Wiki
Model Selection
Switching Models Mid-Generation
Switching Models Mid-Generation
You can change models and regenerate sections:
- Go to model configuration
- Select a different provider/model
- Regenerate specific sections
- Compare results from different models
Custom Model Parameters
Custom Model Parameters
For advanced users, you can adjust:
- Temperature (creativity vs. consistency)
- Top-p (response diversity)
- Max tokens (response length)
- Custom model names (for compatible providers)
Private Repository Access
1
Generate Personal Access Token
- Go to GitHub Settings → Developer settings → Personal access tokens
- Generate new token (classic)
- Select scopes:
repo
(for private repos) - Copy the generated token
2
Add Token to DeepWiki
- Click ”+ Add access tokens” in the DeepWiki interface
- Paste your personal access token
- The token is stored locally and used for repository access
3
Generate Private Repo Wiki
Now you can generate wikis for your private repositories using the same process.
Tokens are stored locally in your browser. Never share your access tokens.
Troubleshooting Your First Wiki
Generation Failed
Generation Failed
Common causes:
- Invalid repository URL
- Private repository without access token
- API key issues
- Repository too large
- Verify the repository URL is accessible
- Check API key configuration
- Try a smaller repository first
- Check the browser developer console for errors
Poor Quality Results
Poor Quality Results
Possible issues:
- Wrong model selected
- Repository has unusual structure
- Limited code documentation
- Try a different AI model
- Use OpenAI GPT-4 for complex repositories
- Enable Deep Research mode
- Ask specific questions to get better answers
Slow Generation
Slow Generation
Normal for:
- Large repositories (1000+ files)
- Complex codebases
- Detailed analysis requests
- Start with smaller repositories
- Use faster models like
gemini-2.0-flash
- Avoid Deep Research for initial testing
- Check your internet connection
Example Walkthrough
Let’s generate a wiki for the Express.js repository:1
Setup
- Repository:
https://github.com/expressjs/express
- Model: Google Gemini 2.0 Flash
- No authentication needed (public repo)
2
Expected Results
The generated wiki should include:
- Express.js framework overview
- Middleware architecture explanation
- Router and route handling
- Request/response cycle diagrams
- API reference for core methods
3
Testing Ask Feature
Try asking:
- “How does Express.js middleware work?”
- “Show me how to create a simple server”
- “Explain the routing system”
Next Steps
Explore Advanced Features
Learn about advanced wiki generation options and customization
Set Up Multiple Models
Configure multiple AI providers to compare results
Production Deployment
Deploy DeepWiki for team or production use
API Integration
Integrate DeepWiki into your development workflow