CodeRabbit brings automated code review to Google Antigravity

A review and fix loop for AI-generated code

CodeRabbit said the integration is designed to help teams validate agent-generated changes as quickly as coding agents produce them. Its review acts as a governance and control layer, helping developers assess whether changes deliver the intended behavior, fit the architecture and introduce risks.

In the workflow described by CodeRabbit, Antigravity generates code, CodeRabbit reviews the changes, and Antigravity uses the review findings to make fixes. The coding agent can then request another review and report any unresolved findings, keeping the process within the same agentic loop.

CodeRabbit said this approach is intended to reduce interruptions and context switching between writing code and understanding the changes that require attention.

What the CodeRabbit plugin includes

Developers can install the CodeRabbit plugin from the Antigravity Plugin Marketplace Customizations tab or through the Google Antigravity CLI. The plugin provides skills and agents for CodeRabbit code reviews, while the separately installed CodeRabbit CLI runs local reviews.

• Code review: Reviews local code changes with CodeRabbit, guides Antigravity through fixing findings and reviews the fixes.
• Autofix: Applies existing CodeRabbit review feedback from a GitHub pull request, with approval required for each change.

The autofix workflow requires the GitHub CLI to be installed and signed in. To install the plugin through the Antigravity CLI, developers can run agy plugin install coderabbit. CodeRabbit also instructs first-time users to install its CLI and sign in to a CodeRabbit account before triggering a review.

Reviews focus on code changes in context

According to CodeRabbit, its independent code review is designed to identify functional and logical errors, race conditions, security vulnerabilities and gaps in test coverage. It also provides architectural feedback.

The company said reviews use codebase context, functional dependencies, project structure, team coding guidelines and learned preferences to assess changes in relation to the software being built and its intended outcome. Inside Antigravity, those findings become actionable guidance for the coding agent, including what needs attention, why it matters and the context required to address it.

For developers using the integration, the source describes a workflow in which they ask Antigravity in natural language to review changes, fix selected findings and verify the fixes with another review. CodeRabbit directs users to its setup guide to install the plugin, connect an account and run a first review.