review-evidence-skill

You collect code review evidence without final gate judgments by constructing lane-specific evidence packages across static-semantic, test, security, and complexity/performance dimensions. This approach outputs a standardized review-evidence report for downstream aggregation rather than blocking deployment.

Synthesizing scoped security and test review evidence requires applying fallback-aware SubAgent dispatch and evidence triage rules. This process promotes only risks threatening acceptance or validation credibility while absorbing low-severity items into residual risk with explicit rules.

Yes, you can configure review breadth by selecting a review_profile ranging from light to deep. This selects specific review lanes and ensures each lane records coverage status and tracks missing evidence instead of collapsing everything into an untraceable summary.

Handling missing evidence during a code security review involves applying lane coverage, fallback, and completeness tracking. This mechanism ensures each selected lane records its coverage status and explicitly documents missing evidence for the downstream gate-skill evaluation.

Use an evidence collection workflow during the Verify implementation phase when you need to gather and synthesize scoped review data for downstream gate-skill aggregation. It is specifically designed for situations requiring evidence collection rather than a final pass or fail gate decision.