WhatsApp Context Extractor

Extract WhatsApp chat context for AI agents with RAG. Parse exports, extract decisions, tasks, and key information, then convert to embeddings for semantic search.

Features

• Parse WhatsApp exports (.txt format)
• Classify messages: decision, task, info, question, other
• Extract key information: decisions, tasks, important links
• Create embeddings with Ollama or MiniMax
• Store in ChromaDB for semantic search
• CLI interface for easy usage

What Makes It Unique

Unlike generic WhatsApp RAG solutions, this tool is designed for AI agent context:

• Selective extraction: Only important content (decisions, context, agreements)
• Rich metadata: Author, date, thread/chat ID
• Organized by project/topic: Index multiple chats by theme
• Incremental updates: Add new messages without reprocessing everything
• Direct context injection: Export to markdown for immediate AI agent use

Installation

git clone https://github.com/pepeneif/whatsapp-context-extractor.git
cd whatsapp-context-extractor
pip install -r requirements.txt

Usage

# Parse a WhatsApp export
wce parse chat.txt --output chat.json

# Create embeddings and store in vector DB
wce embed chat.json --collection my_chat

# Search for context
wce search "what decisions were made" --collection my_chat

# Export to markdown for direct context injection
wce export --collection my_chat --format markdown

Configuration

Copy .env.example to .env and configure:

# Ollama (local, free)
EMBEDDER=ollama
OLLAMA_MODEL=nomic-embed-text

# Or MiniMax (API)
# EMBEDDER=minimax
# MINIMAX_API_KEY=your_key

Project Structure

whatsapp-context-extractor/
├── src/wce/
│   ├── __init__.py
│   ├── parser.py       # Parse WhatsApp export
│   ├── extractor.py    # Extract decisions, tasks, links
│   ├── chunker.py      # Chunk with metadata
│   ├── embedder.py     # Create embeddings
│   ├── vector_store.py  # ChromaDB integration
│   └── cli.py          # CLI interface
├── examples/
├── tests/
├── requirements.txt
└── README.md

Use Cases

• AI Agent Memory: Give your AI agents context from past conversations
• Project Documentation: Extract decisions and tasks from chat history
• Knowledge Base: Build searchable archive of important discussions
• Team Onboarding: Quick context for new team members

Contributing

Contributions welcome! Please open an issue or PR on GitHub.

License

MIT