OpenClaw Browser Automation

OpenClaw includes a built-in Chrome browser that can be controlled via CLI commands. This is useful for testing web apps, taking screenshots, and automating browser interactions.

Setup

The browser profile is configured in openclaw.json:

{
  "browser": {
    "enabled": true,
    "executablePath": "/usr/bin/google-chrome-stable",
    "headless": true,
    "noSandbox": true,
    "defaultProfile": "openclaw"
  }
}

Key Commands

# Start the browser
openclaw browser --browser-profile openclaw start

# Check status
openclaw browser --browser-profile openclaw status

# Navigate to a URL
openclaw browser --browser-profile openclaw navigate http://localhost:3000

# Take a snapshot (accessibility tree with refs)
openclaw browser --browser-profile openclaw snapshot

# Click an element by ref
openclaw browser --browser-profile openclaw click e14

# Take a screenshot
openclaw browser --browser-profile openclaw screenshot

Workflow for Testing Web Apps

1. Start the dev server (npm run dev)
2. Start the browser (openclaw browser start)
3. Navigate to the page
4. Take a snapshot to identify element refs
5. Interact with elements via click, type, etc.
6. Take screenshots to verify results

Lessons Learned

• The browser runs on CDP port 18800 by default
• Screenshots are saved to ~/.openclaw/media/browser/
• The gateway must be running for browser commands to work
• If the gateway closes unexpectedly, restart it before retrying browser commands

Referenced from: Mar 20, 2026