bmad-code-review

Automate adversarial code review by prompting the tool to run parallel review layers on pull requests, diffs, or branch comparisons, which surfaces edge cases and generates actionable findings. It orchestrates multiple perspectives to evaluate code changes before merging.

Adversarial code review applies multiple parallel perspectives, such as Blind Hunter and Edge Case Hunter, to scrutinize code changes. This mechanism surfaces hidden gaps and acceptance criteria violations by intentionally probing for weaknesses across diverse viewpoints.

Yes, you can review git diffs and branch comparisons across any project. The tool applies deterministic triage to code changes universally, requiring no specific project configurations to execute its multi-layered review process.

The best way to triage code review findings automatically is using a structured tool that orchestrates parallel review layers and applies deterministic triage. This categorizes edge cases and acceptance violations into actionable findings for faster decision-making.

Start a code review on a diff by prompting with 'run code review' or 'review this code'. This initiates the automated adversarial process, applying memory-based state tracking to ensure repeatable reviews across the provided changes.