Introduction to Grip

Grip is a powerful, standalone CLI that connects your AI assistants to live web browsers through the Model Context Protocol (MCP).

What is Grip?

Grip is an open-source bridge between artificial intelligence and the web. By running as a lightweight background daemon, the Grip CLI exposes your live browser session—including the DOM tree, accessibility data, and interactive capabilities—to AI IDEs and agents.

Instead of AI hallucinating web layouts or being unable to interact with your local development server, Grip gives them precise, structured access to see, click, type, and evaluate JavaScript exactly as a human would.

How it works

Grip acts as a translation layer for the standard Model Context Protocol (MCP) used by AI tools to interact with your web browser.

AI Assistant
MCP
Grip Daemon
CDP
Browser

When you launch Grip, it connects to a debugging port on your browser. Your AI IDE then spawns the Grip CLI as a subprocess to execute tool calls (like "Click the Submit button" or "Read this page"). Grip executes the action via CDP and streams the result back to the AI.

What you can do

  • AI Web Browsing — Let your AI assistant read live documentation, API references, or StackOverflow directly from your browser.
  • Visual Debugging — Ask your AI to inspect the DOM, read console logs, or analyze network requests on your local development server.
  • Automated Testing — Instruct your AI to navigate through a checkout flow or test a form submission just by describing the steps.
  • Universal Support — Works entirely out-of-the-box on Mac, Windows, and Linux with any Chromium-based browser. No Node.js required.

Who it's for

Developers

If you use AI-powered coding tools, Grip supercharges them by granting them vision into your actual browser. You no longer need to copy-paste HTML snippets or console errors; just ask your AI to look at the page.

AI Apps & Agents

Grip provides a clean, standardized MCP interface for programmatic browser automation. It automatically parses complex DOMs into clean, token-efficient interactive accessibility trees (AXTrees) optimized for Large Language Models.

Where to go next

Ready to give your AI access to the web? Start with the quick start guide.