bmad-code-review

To automate an adversarial code review, run the Python scripts included in the Skill to execute parallel review layers. This process thoroughly analyzes code changes and categorizes findings into actionable triage categories.

Parallel review layers are specialized analysis agents—Blind Hunter, Edge Case Hunter, and Acceptance Auditor—that simultaneously evaluate code changes. This adversarial approach ensures thorough code analysis by targeting different aspects of code quality.

The code review process categorizes findings into structured triage buckets: decision_needed, patch, defer, and dismiss. This classification helps teams prioritize code changes and address quality issues efficiently.

Yes, you can configure pre-flight loads and compliance checks by editing the customize.toml file. This allows you to set personalized settings and tailor the code review workflow to your specific project requirements.

Yes, Python scripts are required to execute and customize the code review process. The scripts drive the parallel analysis layers and handle the categorization of code changes.

The best way to triage code quality issues is using a structured categorization system that separates findings into patch, defer, dismiss, or decision_needed actions, ensuring efficient and accurate review processing.