adversarial-review

Adversarial code review challenges work from multiple critical lenses by spawning opposite-model reviewers to critique changes, helping catch edge cases and quality gaps without altering the original code.

To run an adversarial code review, trigger the process after large diffs or planning sessions to spawn 1–3 opposite-model reviewers via CLI, apply distinct lenses, and synthesize a verdict guiding decisions.

Yes, you can spawn opposite-model reviewers during or after planning sessions to assess resilience and ensure multiple critical lenses are applied to your proposed work before implementation.

No, adversarial review does not modify your original code; it spawns distinct reviewers to challenge the work and delivers a synthesized verdict to guide your decisions.

You can spawn 1–3 reviewers on the opposite model to critique your changes, each applying a distinct lens to assess resilience and identify quality gaps.

Avoid using adversarial review when you lack an opposite-model CLI environment, as it requires spawning external reviewers and loading brain principles to synthesize a verdict without internal delegation.