Browser overview

Google Antigravity can open, read, and actuate a local Chrome browser, enabling you to test development websites, read documentation sources, and automate a variety of browser tasks.

---

Core mechanisms

Type /browser in the Agent panel to hand a task to the browser subagent. The subagent drives your own Google Chrome through the Chrome DevTools MCP server, which Antigravity IDE starts with npx. Before you use it, install Node.js so that npx is on your PATH, and start Chrome with remote debugging turned on, as described in Set up Chrome.

The subagent can take screenshots, and when it interacts with a page, it can record its actions as a WebM video. For details, refer to Browser recordings.

To control what the subagent can do, click Antigravity - Settings in the status bar and select Browser. Browser Javascript Execution Policy controls whether the agent can run custom JavaScript in the page, and Browser Actuation Rules sets the URLs the subagent is allowed or denied.