judgment-day

A parallel adversarial code review uses two independent sub-agents to review the same codebase simultaneously. This dual-reviewer mechanism identifies issues missed by a single reviewer to ensure higher code quality assurance.

To run an adversarial code review, trigger the process using commands like 'judgment day' or 'review adversarial'. The system then launches parallel reviews to evaluate your code against project-specific quality standards.

If your code fails an adversarial review, the system supports iterative fixes and re-reviews. It will continue evaluating your updated code until it passes both independent reviews or escalates the persistent issues.

Yes, parallel adversarial code review requires a project-specific skill registry and standards. These dependencies establish the baseline criteria that the two independent sub-agents use to evaluate your codebase.

Parallel adversarial review is better than a single code review because it deploys two independent sub-agents simultaneously. This approach catches issues that a single reviewer might miss, ensuring more robust quality assurance.

Issues found during iterative code reviews should be escalated when they persist after multiple fix and re-review cycles. The system escalates if the code continuously fails to pass both independent sub-agent reviews.