Self-hosted personal AI agent framework in Python with a browser WebUI, terminal client, chat app channels, tools, memory and MCP support.
csr1010/whatsapp-ai-agent-sdk
Open-source TypeScript SDK for building a self-hosted AI agent that runs on WhatsApp, with your own model key and data.
- Stars
- 252
- Forks
- 53
- Open issues
- 1
- Last commit
- 5 days ago
What is it?
Wappy is a TypeScript SDK and project generator for building an AI agent that talks to users over WhatsApp through the official WhatsApp Cloud API. Running `npm create @wappy_ai/agent` produces a project that wires up the agent loop, memory, tool calling and reply formatting, and you supply your own model provider key (Anthropic, OpenAI, Gemini, or a local Ollama model).
Conversations and memory are stored locally by default in SQLite/LibSQL, with BM25 retrieval for ingested documents. Cognee can be chosen instead as a self-hosted or cloud knowledge graph backend. An optional Gmail and Google Calendar assistant ships as a read-only tool example, using your own Google OAuth credentials.
The project is pre-1.0 and split into packages: core contracts, the harness that runs the agent, a WhatsApp package handling formatting, retries and webhooks, and the installer. Your business logic, such as ticketing or booking integrations, is added on top of the generated project.
Key features
- 01Generates a WhatsApp agent project with one `npm create` command.
- 02Supports Anthropic, OpenAI, Gemini, and local Ollama models with your own keys.
- 03Stores conversations and ingested knowledge locally in SQLite/LibSQL by default.
- 04Offers Cognee as an optional self-hosted or cloud knowledge graph backend.
- 05Includes a read-only Gmail and Google Calendar tool example.
Use cases
- Building a private personal assistant that remembers context across WhatsApp conversations.
- Answering questions over your own notes or documents without sending them to a third-party cloud.
- Prototyping a customer support or booking bot reachable on WhatsApp.
- Running a fully offline agent against a local Ollama model.
Quick start
- 1
Generate the project with the installer
npm create @wappy_ai/agent - 2
Copy the env sample and fill in your model key and WhatsApp credentials
cp .env.sample .env - 3
Install dependencies
npm install - 4
Start the dev server and open the tunnel, then set the printed URL as Meta's webhook
npm run dev
Things to know
- The project is pre-1.0 and still in early days, so expect changes.
- It requires Node.js 22 or newer and a WhatsApp Cloud API app from Meta.
- You need your own model API key, or a local Ollama setup, to run the agent.
In collections
Topics
FAQ
- What is whatsapp-ai-agent-sdk?
- Open-source TypeScript SDK for building a self-hosted AI agent that runs on WhatsApp, with your own model key and data. Wappy is a TypeScript SDK and project generator for building an AI agent that talks to users over WhatsApp through the official WhatsApp Cloud API. Running `npm create @wappy_ai/agent` produces a project that wires up the agent loop, memory, tool calling and reply formatting, and you supply your own model provider key (Anthropic, OpenAI, Gemini, or a local Ollama model).
- Is whatsapp-ai-agent-sdk free to use? What license does it have?
- whatsapp-ai-agent-sdk is open source under the MIT license. Check the license terms before using it in commercial products.
- Is whatsapp-ai-agent-sdk still maintained?
- Actively maintained. The last commit to whatsapp-ai-agent-sdk was on Nov 2, 2026.
- What can I use whatsapp-ai-agent-sdk for?
- Building a private personal assistant that remembers context across WhatsApp conversations. Answering questions over your own notes or documents without sending them to a third-party cloud. Prototyping a customer support or booking bot reachable on WhatsApp. Running a fully offline agent against a local Ollama model.
- How do I install whatsapp-ai-agent-sdk?
- Generate the project with the installer: npm create @wappy_ai/agent Copy the env sample and fill in your model key and WhatsApp credentials: cp .env.sample .env Install dependencies: npm install Start the dev server and open the tunnel, then set the printed URL as Meta's webhook: npm run dev
- Where can I get whatsapp-ai-agent-sdk?
- The source code and installation instructions are on GitHub: https://github.com/csr1010/whatsapp-ai-agent-sdk
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
A local-first desktop and Docker AI agent app where you run personal bots that use any model via one Modelflare key, reachable from WeChat, Feishu, Telegram, Slack and Discord.
A local-first personal AI assistant in readable Python, with its own agent loop, SQLite-based memory, and built-in evals.
Open-source desktop AI office assistant that plans and carries out file, document and command tasks on your machine from a one-sentence request.
A local-first personal assistant with a Rust core for the agent loop, memory and tools, plus TypeScript chat gateways and a dashboard.
Open-source personal AI assistant and Agent Harness that plans tasks, runs tools and skills, and keeps memory across web and IM channels.