code-council

Parallel AI code review runs 2-3 AI reviewers simultaneously on the same code to identify security vulnerabilities and logic errors. The host agent verifies each finding to ensure accuracy before merging into the main branch.

Multi-backend code review uses the code-dispatcher to run parallel execution across multiple AI agents. It verifies findings across different backends and synthesizes a final actionable report highlighting performance issues and code quality concerns.

AI code review verifies findings by having the host agent validate each issue reported by the AI reviewers. Instead of just listing potential code quality concerns, it consolidates and presents only verified findings in a synthesized report.

To run a parallel code review on a specific file like src/main.py, execute the council skill targeting that file path. The skill dispatches the file to multiple AI backends for simultaneous analysis and synthesizes the verified results.

Yes, you can use parallel code review for a new feature branch. It is designed to provide a comprehensive code review before merging, automatically catching security vulnerabilities, logic errors, and performance issues early in development.

A limitation of using AI agents for code review is the potential for unverified false positives, which is why the host agent verifies each finding. Users must still review the synthesized report to ensure the identified issues match their specific application context.