A powerful tool to index Markdown documentation from GitHub repositories and ask questions with AI-powered answers and citations.

• 📚 Index .md and .mdx files from any GitHub repository
• 🔍 Semantic search with vector embeddings
• 💬 AI-powered question answering with source citations
• 🎯 Multiple LLM providers (OpenAI, Anthropic)
• 🚀 Fast retrieval with ChromaDB
• 🌐 Web UI and CLI interface

1. Clone this repository
2. Install dependencies:

pip install -r requirements.txt

3. Copy .env.example to .env and add your API keys:

cp .env.example .env

4. Edit .env with your API keys

python -m deepwiki index <github_repo_url>

Example:

python -m deepwiki index https://github.com/anthropics/anthropic-sdk-python

python -m deepwiki ask "How do I use streaming with the SDK?"

python -m deepwiki serve

Then open http://localhost:8000 in your browser.

python -m deepwiki list

python -m deepwiki clear

1. Crawling: Fetches all .md and .mdx files from the specified GitHub repository
2. Chunking: Splits documents into manageable chunks with overlap
3. Embedding: Generates vector embeddings using OpenAI or similar models
4. Indexing: Stores embeddings in ChromaDB for fast retrieval
5. Question Answering: Retrieves relevant chunks and uses LLM to generate answers with citations

deepwiki/ ├── __init__.py ├── __main__.py # CLI entry point ├── config.py # Configuration management ├── crawler.py # GitHub repository crawler ├── indexer.py # Document chunking and indexing ├── retriever.py # Search and retrieval ├── qa.py # Question answering with citations └── api.py # FastAPI web server 

Edit .env to customize:

• LLM_PROVIDER: Choose openai or anthropic
• EMBEDDING_MODEL: Embedding model to use
• CHUNK_SIZE: Size of text chunks for indexing
• CHUNK_OVERLAP: Overlap between chunks

• Python 3.8+
• OpenAI API key or Anthropic API key
• GitHub token (optional, for higher rate limits)

MIT