rc:diff-review

To automate code review and fix loops, use a tool that compares your feature branch against main and executes a five-round cycle of validation and auto-fix. This orchestrates review, classifies findings, and applies verified fixes until the code is clean.

Yes, you can use sub-agents to classify code review findings automatically. Intelligent validation categorizes findings as to_fix, dismissed, or deferred based on project-specific rules and confidence gating, ensuring only verified issues are remediated.

Automated code review works with git and bash for quality assurance. It requires these dependencies to compare branches, execute shell scripts, and perform structural analysis and verification during the multi-round remediation cycle.

The best way to handle P1/P2 security checks during refactoring is an automated multi-round review loop. It identifies logic or security flaws, validates them against project rules, and applies verified fixes until the code meets quality standards.

Yes, you need jq installed to run automated branch comparisons. Along with git and bash, jq is a required dependency for parsing and processing the JSON data generated during the validation and auto-fix cycle.

You should not use an automated code review loop when your project lacks clear quality assurance rules or when changes require subjective human judgment. The system relies on project-specific rules and confidence gating to classify and remediate findings.