bmad-code-review

Adversarial code review uses parallel layers like Blind Hunter and Edge Case Hunter to proactively find hidden issues and compliance violations in code changes. It actively hunts for flaws rather than passively reading, generating structured actionable suggestions for software development workflows.

Run the review skill to start a session that automatically executes multiple adversarial review steps across parallel layers on your code changes. It then organizes all findings into structured, actionable categories for efficient issue triage and resolution.

Yes, this approach is designed for software development workflows requiring code quality and compliance checks. It tracks code quality issues by organizing adversarial findings into structured triage categories for efficient resolution.

Adversarial review layers actively probe code changes from different perspectives like edge cases and acceptance criteria, while standard analysis applies fixed rule sets. This multi-layer approach provides comprehensive analysis and structured triage of findings.

A limitation is that handling findings from different adversarial layers requires executing multiple review steps and manual oversight to resolve triaged issues. It focuses on identifying issues and suggesting improvements rather than automatically fixing the code.