A Retrieval-Augmented Generation (RAG) chatbot integrated with Mattermost, using LlamaIndex, Qdrant, OpenAI embeddings, and Anthropic's Claude for response generation.
- Mattermost Integration: Seamlessly integrates with Mattermost via slash commands and bot interactions
- RAG Architecture: Uses state-of-the-art RAG techniques to provide context-aware responses
- Incremental Updates: Supports both bulk ingestion and incremental updates of channel messages
- Vector Search: Utilizes Qdrant for efficient similarity search of message embeddings
- Modern Tech Stack:
- OpenAI's text-embedding-3-large for embeddings
- Anthropic's Claude for response generation
- LlamaIndex for RAG orchestration
- FastAPI for the web server
- Docker for containerization
- Python 3.10+
- Docker and Docker Compose
- Mattermost server with bot account and access token
- OpenAI API key
- Anthropic API key
- Clone the repository:
git clone <repository-url>
cd mmrag2- Create and configure the
.envfile:
cp .env.example .env
# Edit .env with your configurationRequired environment variables:
MATTERMOST_URL: Your Mattermost server URLMATTERMOST_BOT_TOKEN: Bot account access tokenOPENAI_API_KEY: OpenAI API key for embeddingsANTHROPIC_API_KEY: Anthropic API key for Claude- See
.envfile for additional configuration options
- Build and start the containers:
docker-compose up -d --build- Check the logs:
docker-compose logs -f- Create a virtual environment:
python -m venv venv
source venv/bin/activate # On Windows: venv\Scripts\activate- Install dependencies:
pip install -r requirements.txt- Run the development server:
python -m uvicorn app.main:app --reload --host 0.0.0.0 --port 5000The bot responds to the following slash commands in Mattermost:
/rag <query>: Ask a question to the RAG system- Example:
/rag What was discussed about the project timeline?
POST /api/slash_command: Handles Mattermost slash commandsPOST /api/ingest: Triggers message ingestion from a channelGET /health: Health check endpoint
- Initial bulk ingestion:
curl -X POST "http://localhost:5000/api/ingest" \
-H "Content-Type: application/x-www-form-urlencoded" \
-d "channel_id=your_channel_id"- Incremental update:
curl -X POST "http://localhost:5000/api/ingest" \
-H "Content-Type: application/x-www-form-urlencoded" \
-d "channel_id=your_channel_id&since_time=1234567890000"mmrag2/
├── app/
│ ├── __init__.py
│ ├── main.py # FastAPI application
│ ├── rag_engine.py # RAG implementation
│ ├── processing.py # Text processing and chunking
│ ├── mattermost_client.py # Mattermost API client
│ └── utils.py # Utility functions
├── tests/
│ ├── __init__.py
│ ├── test_api.py
│ ├── test_processing.py
│ ├── test_rag_engine.py
│ └── test_mattermost_client.py
├── scripts/ # Utility scripts
├── docker-compose.yml # Docker Compose configuration
├── Dockerfile # Docker build configuration
├── requirements.txt # Python dependencies
└── README.md # This file
pytestThe application is configured through environment variables. See the .env file for available options.
- Fork the repository
- Create a feature branch
- Commit your changes
- Push to the branch
- Create a Pull Request
[Your License Here]
- LlamaIndex for the RAG framework
- Qdrant for the vector database
- OpenAI for embeddings
- Anthropic for Claude
- Mattermost for the chat platform