Skip to content
Google Antigravity Docs

Explore Antigravity

Getting StartedInstall & launch your first agentRemote ControlDrive desktop & CLI sessions from any browserAgent Skills & RulesReusable workflows & workspace instructionsSubagents & TeamworkParallel multi-agent orchestrationAntigravity SDKBuild custom Python agents & hooksPermissions & SandboxFile access controls & safe execution
Try searching
    Remote Control
    NewGemini 4 Hydrogen and Orchestrator now in Google AntigravityRead post
    Browser recordings - Antigravity IDEantigravity-ideAntigravity IDEFeaturesBrowser recordings - Antigravity IDE Antigravity IDE FeaturesAntigravity IDE
    Available on:
    Antigravity IDE

    Browser recordings

    Section titled “Browser recordings”

    In Antigravity IDE (v2.6.0+), the browser subagent runs when you type /browser in the Antigravity panel. It drives your own Google Chrome through the Chrome DevTools MCP server (npx), so Node.js must be on your PATH and Chrome must already be running locally with remote debugging turned on. For setup steps, refer to Set up Chrome.

    When the subagent interacts with a page, such as clicking, typing, or filling in a form, Antigravity IDE can record those actions as a WebM video. Recording isn’t guaranteed: sessions that only open pages or take screenshots don’t produce a recording, and a session that interacts with a page might not produce one either.

    When a recording is produced, it’s saved as recording.webm in the browser subagent’s artifacts after the subagent finishes, and the agent lets you know it’s available.