Connect Claude to WhatsApp in 30 minutes, no code
A WhatsApp agent that answers your customers 24/7. The repo does the heavy lifting — you just follow the 7 steps from your terminal.
May 24, 2026 · 4 min read
A custom WhatsApp agent — no dev, no paid no-code
whatsapp-agentkit is an open source repo (by Hainrixz) that hands you all the plumbing for a turnkey WhatsApp AI agent: a FastAPI server, SQLite memory per phone number, Twilio or Meta Cloud API integration, and Claude (sonnet-4-6) as the brain. You drop in your API keys, describe your business, and the agent answers your customers for you.
Two paths depending on your setup: Claude Code, which scaffolds everything on your machine autonomously, or Cowork (the Claude app), which acts as a copilot to copy-paste the commands into your terminal. Same result, same repo — just two levels of autonomy.
The Claude AI Lab is my Skool community where I share my Claude systems and the more advanced modules. Entry is free.
Join the Lab →The agent handles conversational memory per customer (each number has its own history), loads your business files (PDF menu, FAQ, pricing) into its context, and you can give it a custom personality via a YAML. Final deployment on Railway to go to production, with a free Twilio sandbox to test without a credit card.
The repo is built to be driven by Claude. Whether you use Claude Code or Cowork, Claude reads the project, asks you 10 questions about your business, and configures everything based on your answers. All you do is approve and copy-paste.
Path A · Claude Code — the autonomous setup
python3 --version and node --version. On Mac, brew install python node does the trick.npm install -g @anthropic-ai/claude-codegit clone https://github.com/Hainrixz/whatsapp-agentkit.git then cd whatsapp-agentkit && bash start.shclaude in the terminal. Claude reads the repo and asks you 10 questions about your business (name, industry, tone, hours, the agent's goal, files to ingest).sk-ant-...) from platform.anthropic.com. Twilio (Account SID + Auth Token + Phone, free sandbox) or Meta Cloud API (Access Token + Phone Number ID + Verify Token).python tests/test_local.pyrailway.app, and dictates the webhook URL to paste into Twilio or Meta. You get your first real message.Path B · Cowork — the copilot setup
No Claude Code on your machine? The Cowork app (Claude.ai web or desktop) can dictate the exact same steps. You paste into your terminal, send it back the output, and it unblocks you. It's more manual, but it works.
claude.ai or the desktop app. Switch to the Sonnet 4.6 or Opus 4.7 model for execution quality.github.com/Hainrixz/whatsapp-agentkit and write: "Walk me through cloning this repo step by step, installing the dependencies, configuring my business, and deploying the agent. I don't have Claude Code, so I'll copy-paste each command into my terminal."knowledge/. It prepares the content of config/business.yaml and config/prompts.yaml for you to paste..env template with the variables (ANTHROPIC_API_KEY, TWILIO_ACCOUNT_SID, TWILIO_AUTH_TOKEN, TWILIO_PHONE, or META_ACCESS_TOKEN, META_PHONE_NUMBER_ID, META_VERIFY_TOKEN). You fill in your real keys and save.python tests/test_local.py and paste the output into Cowork. If it crashes, it reads the error and gives you the fix. If the agent answers oddly, it tweaks the prompts.yaml YAML with you.Cowork adds 10 to 15 minutes vs Claude Code (copy-paste back and forth), but stays well under an hour. If you're doing this only once, Cowork is enough. If you're building several, install Claude Code.
Want to go further?
In the Lab, I share my Claude agent and sub-agent setups, and how to run them in parallel without breaking things.
A dedicated session or program, tailored to your tools and use cases.
And day-to-day, I post one reel a day on Instagram: @quentin_iamarketing