Clawdbot: install an open-source AI assistant on your machine or VPS

What is Clawdbot?

Clawdbot is an open-source project that transforms Claude into an autonomous agent capable of interacting with your communication channels: WhatsApp, Telegram, Discord, Slack, Signal, and even iMessage. The idea? An AI assistant always available that can respond to your messages, execute tasks, and act on your behalf.

The project runs via a "Gateway" — a local or remote server that bridges messaging APIs and Claude. You can install it on your Mac, Linux PC, or on a VPS for 24/7 availability.

Prerequisites

Before you start:

• Node.js 22 or higher (mandatory)
• macOS, Linux, or Windows with WSL2 (native Windows not recommended)
• An Anthropic API key, OpenAI Code subscription, or Claude Code
• pnpm (recommended for source builds)

To check your Node version:

node --version
# Should display v22.x.x or higher

Quick Installation (Recommended)

On macOS or Linux

The simplest method uses the official installation script:

curl -fsSL https://clawd.bot/install.sh | bash

This script downloads Clawdbot, installs it globally, and launches the configuration wizard.

Via npm/pnpm

If you prefer more control:

npm install -g clawdbot@latest
# or with pnpm
pnpm add -g clawdbot@latest

Then, run the onboarding:

clawdbot onboard --install-daemon

• Choosing the AI model (Claude Opus, Sonnet, etc.)
• Configuring your API key
• Connecting to channels (WhatsApp, Telegram...)
• Installing the daemon for background operation

On Windows

Native Windows is not officially supported. Use WSL2:

# In PowerShell (administrator)
wsl --install

# Once in WSL2
curl -fsSL https://clawd.bot/install.sh | bash

Installation from Source

For developers who want to contribute or customize:

git clone https://github.com/clawdbot/clawdbot.git
cd clawdbot

pnpm install
pnpm ui:build
pnpm build

pnpm clawdbot onboard --install-daemon

For development mode with hot-reload:

pnpm gateway:watch

VPS Deployment

For 24/7 availability, deploy Clawdbot on a VPS (DigitalOcean, Hetzner, OVH...).

Classic Method

1. Provision a VPS with Ubuntu 22.04+ and Node 22
2. Install Clawdbot:

curl -fsSL https://clawd.bot/install.sh | bash
clawdbot onboard --install-daemon

1. Configure systemd for automatic startup:

# The daemon is automatically installed by onboarding
systemctl status clawdbot

With Docker

The repo includes Docker files:

git clone https://github.com/clawdbot/clawdbot.git
cd clawdbot
docker-compose up -d

Remote Access

To access your instance from outside:

• Tailscale Serve/Funnel: expose your gateway via HTTPS
• SSH tunnels: for secure access without exposing ports

# Start the gateway on a specific port
clawdbot gateway --port 18789 --verbose

Configuration

The config file is located at ~/.clawdbot/clawdbot.json:

{
  "agent": {
    "model": "anthropic/claude-opus-4-5"
  }
}

You can choose from different Claude models depending on your cost and performance needs.

Connecting Your Channels

WhatsApp

clawdbot channels login

Scan the QR code with your phone to link your WhatsApp account.

Telegram / Discord

These platforms require a bot token:

1. Create a bot on @BotFather (Telegram) or Discord Developer Portal
2. Configure the token in Clawdbot
3. Start the pairing

Verify Installation

clawdbot status
clawdbot health

These commands display the gateway status, connected channels, and overall system health.

Important Notes

• Bun has known issues with WhatsApp and Telegram. Use Node for these channels.
• Never store your API key in plain text in versioned files
• Monitor your API consumption — an autonomous agent can generate many requests

Conclusion

Clawdbot opens interesting possibilities for automating your communications. An AI assistant that responds to your WhatsApp messages while you sleep, manages your Telegram DMs, or moderates your Discord — all becomes possible.

The project is actively developed and the community is responsive. If you're looking to experiment with AI agents beyond simple chat, it's a good starting point.

Want help integrating an AI agent into your processes? Let's talk.