A WhatsApp customer-support bot that turns a folder of AI agent tools into live HTTP endpoints — drop in a tool, get an API.
Built with Flask + Twilio + Agency Swarm. Designed to deploy in minutes on Railway.
Every Python file you put in ./tools/ is auto-discovered at startup and exposed as its own authenticated POST endpoint — no manual route wiring. Incoming WhatsApp messages (via Twilio) are routed to the right agent tool, which runs and replies.
WhatsApp message ──▶ Twilio ──▶ Flask endpoint ──▶ Agent tool ──▶ reply
git clone https://github.com/dawood-k/Agent-infinity.git
cd Agent-infinity
pip install -r requirements.txt
cp .env.example .env # add your Twilio + DB_TOKEN secrets
python main.pyEndpoints are protected with a bearer token (DB_TOKEN) — every request must send Authorization: Bearer <token>.
- Create a new
.pyfile in./tools/defining an Agency Swarm tool class. - Restart — it's automatically registered as
POST /<ToolName>.
Ships with railway.json and a config for one-click Railway deployment.
Python · Flask · Gunicorn · Twilio · Agency Swarm
MIT