iterative-review

Automated code review before a merge is handled by spawning a review subagent and a fix subagent. This cycle identifies and addresses code issues, repeating up to three iterations until Critical or High findings reach zero.

An iterative code review cycle repeatedly scans target files for quality issues and automatically fixes them. It operates by spawning subagents that review code, apply fixes, and re-review until quality gates are met or the iteration limit is hit.

Yes, automated subagents can perform repeated quality checks across repositories. The skill applies to feature development, pull requests, and automated workflows, using YAML frontmatter to define discovery and operation constraints for each repository.

The automated review-fix cycle runs a maximum of three iterations. It stops early if the review subagent confirms that all Critical and High severity findings have been resolved to zero by the fix subagent.

The main limitation is the strict maximum of three iterations for the review-fix cycle. If complex code issues require more passes to resolve, the automated process will stop at the limit regardless of whether all Critical or High findings are fixed.

You configure discovery and operation constraints for automated code review using YAML frontmatter. This defines the target files and operational boundaries for the review and fix subagents during the iterative cycle.