Skip to main content

​
Quickstart

Get your personal AI assistant running in under 5 minutes. This guide will walk you through installation, onboarding, and sending your first message.

​
Prerequisites

Before you begin, ensure you have:
  • Node.js ≥ 22: OpenClaw requires Node.js version 22 or higher
  • Package Manager: npm, pnpm, or bun
  • Operating System: macOS, Linux, or Windows (via WSL2)
  • AI Provider Account: Anthropic (Claude) or OpenAI (ChatGPT) subscription
For the best experience, we recommend Anthropic Pro/Max + Opus 4.6 for long-context strength and better prompt-injection resistance.

​
Installation

1

Install OpenClaw globally

Choose your preferred package manager:
npm install -g openclaw@latest
Using pnpm? Make sure your shell’s PATH includes ~/Library/pnpm (macOS) or ~/.local/share/pnpm (Linux).
2

Run the onboarding wizard

The wizard will guide you through setting up the Gateway, configuring your AI model, and connecting your first channel:
openclaw onboard --install-daemon
The --install-daemon flag automatically installs the Gateway as a background service (launchd on macOS, systemd on Linux) so it stays running.
The onboarding wizard:
  1. Creates the configuration file at ~/.openclaw/openclaw.json
  2. Sets up your AI model credentials (Anthropic or OpenAI)
  3. Configures the Gateway on port 18789
  4. Optionally sets up your first messaging channel
  5. Installs the Gateway daemon for automatic startup
3

Start the Gateway

If you didn’t install the daemon, start the Gateway manually:
openclaw gateway --port 18789 --verbose
You should see output indicating the Gateway is running:
Gateway listening on ws://127.0.0.1:18789
Control UI available at http://127.0.0.1:18789
The Gateway binds to localhost (127.0.0.1) by default for security. See the deployment guide for remote access options.

​
Your First Message

Once the Gateway is running, you can interact with your assistant in several ways:

​
Via CLI (Direct)

Send a message directly to the agent: 
oclaw agent --message "Hello, what can you help me with?"

​
Via Messaging Channel

If you configured a messaging channel during onboarding, just send a message to your bot on that platform:
  • WhatsApp: Send a message to your linked device
  • Telegram: Message your bot at @yourbotname
  • Discord: Mention your bot in a channel or DM
  • Slack: DM your bot or mention it in a channel

​
Via Web UI

Open your browser to http://127.0.0.1:18789 to access the Control UI and WebChat interface.

​
Verify Installation

Check that everything is working correctly: 
# Check Gateway status
oclaw gateway status

# Check connected channels
oclaw channels status

# Run diagnostics
oclaw doctor

​
Configuration File

Your configuration is stored at ~/.openclaw/openclaw.json. Here’s a minimal example: 
{
  "agent": {
    "model": "anthropic/claude-opus-4-6"
  },
  "gateway": {
    "port": 18789,
    "bind": "loopback"
  }
}
See the configuration guide for all available options.

​
Common Issues

If port 18789 is already in use, specify a different port:
openclaw gateway --port 18790
Update your config file to use the new port:
{
  "gateway": {
    "port": 18790
  }
}
OpenClaw requires Node.js ≥ 22. Check your version:
node --version
If it’s older, install Node 22+ using nvm or your system’s package manager.
Check the logs for errors:
openclaw logs --follow
Common issues:
  • Config file syntax errors (run openclaw config validate)
  • Missing AI provider credentials
  • Port conflicts
  • File permission issues

​
Next Steps

Add More Channels

Connect WhatsApp, Telegram, Discord, and more

Configure Your Gateway

Customize models, routing, and security settings

Explore Platform Apps

Install the macOS app or iOS/Android nodes

Deploy to Production

Run OpenClaw on a VPS or cloud instance

​
Learn More