LangGraph WhatsApp Chatbot

A WhatsApp chatbot built with LangGraph, FastAPI, and Twilio that can handle both text and image responses.

Prerequisites

• Python 3.10 or higher
• Docker and Docker Compose (for containerized setup)
• A Twilio account with WhatsApp enabled
• A Tavily API key
• An Ngrok account (for exposing your local server)

Environment Variables

Create a .env file in the root directory with the following variables:

# Tavily API Key
TAVILY_API_KEY=your_tavily_api_key

# Ngrok Configuration
NGROK_AUTHTOKEN=your_ngrok_auth_token
HOST=your-reserved-domain.ngrok.io

Running with Docker Compose

1. Build and start the services:

docker-compose up --build

This will start:

• MCP Server (port 8001)
• WhatsApp Bot (port 8000)
• Ngrok tunnel (port 4040)

Running Manually

1. Install dependencies:

pip install -r requirements.txt

2. Start the MCP server:

python mcp_server.py

3. In a new terminal, start the WhatsApp bot:

python whatsapp_chatbot_client.py

4. In another terminal, start ngrok:

ngrok http 8000

Testing the Bot

1. Configure your Twilio WhatsApp sandbox:

   • Go to your Twilio Console
   • Navigate to Messaging > Try it out > Send a WhatsApp message
   • Set the webhook URL to your ngrok URL + /whatsapp
   • Example: https://your-domain.ngrok.io/whatsapp

2. Send a test message to your WhatsApp bot:

   • Text: "Hello"
   • The bot should respond with a message

Project Structure

.
├── docker-compose.yml    # Docker services configuration
├── Dockerfile           # Container definition
├── requirements.txt     # Python dependencies
├── .env                # Environment variables
├── mcp_server.py       # MCP server implementation
├── whatsapp_chatbot_client.py  # WhatsApp bot implementation
└── graph.py            # LangGraph agent configuration

Features

• WhatsApp integration via Twilio
• LangGraph agent for intelligent responses
• Support for both text and image responses
• Docker containerization for easy deployment
• Ngrok integration for exposing local server

Troubleshooting

1. If you get authentication errors:

   • Verify your Twilio credentials in the .env file
   • Check if your Twilio account is active
   • Ensure your WhatsApp sandbox is properly configured

2. If the bot doesn't respond:

   • Check if all services are running
   • Verify the ngrok tunnel is active
   • Check the logs for any error messages

3. If Docker services fail to start:

   • Ensure Docker and Docker Compose are installed
   • Check if ports 8000, 8001, and 4040 are available
   • Verify your .env file is properly configured

Contributing

Feel free to submit issues and enhancement requests!