rai-quality-review

To catch semantic bugs missed by automated linters, audit changed story files for logic correctness, type honesty failures, and language-specific idioms. This review evaluates diffs to produce file:line findings with concrete fixes, preventing subtle regressions before code review.

Fragile tests break under harmless refactors. Identify low-quality tests by evaluating test behavior using mutation survival, refactor resilience, and behavior-specification heuristics, which classify tests as Muda, Fragile, or Valuable to isolate maintenance debt.

Audit code for type honesty failures by reviewing changed file diffs after story implementation and automated gates pass. The review detects type honesty issues, evaluates the API surface, and generates a PASS, PASS WITH RECOMMENDATIONS, or FAIL verdict.

This code review approach requires automated gates to pass first. It runs on the set of changed files in a completed story after automated gates pass, requiring access to changed file diffs and project manifest or language detection to evaluate semantics.

The best way to review API surface and security concerns in changed files is to load project patterns and design context alongside the file diffs. This evaluates language-specific semantics and security to produce file:line findings with explanations and a final verdict.