kiro-review

Code review against task specifications is performed by mapping git diffs to approved boundary constraints and referenced requirement sections, running tests and static checks, and scanning for secrets or placeholders to produce objective pass or fail signals.

Task-local adversarial code review is a process that detects hidden scope changes and spillover by verifying implementations are confined to approved boundaries. Use it when an implementer reports READY_FOR_REVIEW or before accepting a task into feature-level validation.

Detect hidden coupling in git diffs by mapping changed files to task boundary definitions and referenced design sections, running static analysis checks, and validating that modifications align strictly with approved specifications and constraints.

Yes, automated static analysis can run targeted greps for TODO and PLACEHOLDER markers alongside secrets detection to produce objective verification evidence and ensure no incomplete or sensitive data remains in the implementation.

To run a boundary audit on task implementations, you need the task text, git diffs, requirements.md, design.md, the implementer's status report, and any validation commands required to execute tests and static checks.

Code review fails to catch silent design substitutions when it does not map changed files against referenced design sections and boundary constraints, missing hidden coupling or unapproved architectural deviations that fall outside the task scope.