
OpenClaw Chrome MCP
OpenClaw's powerful Chrome DevTools MCP integration. Attach your local AI agent to your real signed-in Chrome browser (with cookies, tabs, and logins intact) for full web control — click, type, read pages, take snapshots, and automate any task directly in your live browser.
Overview
OpenClaw Chrome MCP is the native browser integration in OpenClaw — one of the most popular open-source personal AI assistants (300k+ GitHub stars). It uses Google's official Chrome DevTools MCP server to let your local agent attach directly to your running Chrome (or Brave/Edge) session.
This gives the agent real-time access to your actual browser environment: open tabs, logged-in accounts, cookies, extensions, and live DOM — eliminating the limitations of isolated/headless browsers.
Key Features
- Existing-Session Attachment: Connect to your real Chrome via the built-in
userprofile (or custom profiles) - Full Browser Control: Open/close tabs, navigate, click, type, drag, scroll, upload files
- Rich Observation: AI-optimized snapshots (with element refs), screenshots, console logs, network inspection
- State Management: Cookies, localStorage, geolocation, device emulation, headers
- Multiple Profiles: Isolated
openclawmode +existing-session(Chrome MCP) + remote options - CLI-First:
openclaw browser --browser-profile user ...commands for scripting and interactive use
Quick Setup
# 1. Enable remote debugging in Chrome (chrome://inspect)
# 2. Configure ~/.openclaw/openclaw.json
{
"browser": {
"profiles": {
"user": {
"driver": "existing-session",
"attachOnly": true
}
}
}
}
# 3. Start and attach
openclaw browser --browser-profile user start
openclaw browser --browser-profile user snapshot --format ai
OpenClaw automatically starts the Chrome DevTools MCP server with --autoConnect behind the scenes.
Use Cases
- Personal Automation: Check Gmail, manage calendars, shop online, or handle any logged-in service
- Research & Scraping: Browse multiple authenticated sites with full context
- Complex Workflows: Multi-step tasks across tabs and services (e.g., "transfer money from my bank tab to the invoice site")
- Development Assistance: Debug web apps, test UIs, or interact with internal tools in real sessions
- Agent Orchestration: Combine with NotebookLM MCP, Stitch MCP, or GitHub tools for end-to-end automation
Why OpenClaw Chrome MCP?
Most browser agents use isolated or headless instances that lose your real login state. OpenClaw Chrome MCP bridges that gap — giving your local AI superpowers inside your actual daily browser. Fully local, private, and incredibly practical.
Works with Ollama or any local model. Requires Chrome 144+. Actively maintained with a huge community.