What problem does it solve? Code reviews are inconsistent and often miss critical issues like secrets, silent fallbacks, or architecture violations while wasting time on style nits. This Skill provides a disciplined senior-engineer review procedure with hard token budgets and a prioritized checklist. ## Core Features & Use Cases - Prioritized review axes: Red flags first (secrets, type-safety escapes, silent fallbacks, hardcoding), then architecture conformance, duplication, and complexity. - Mechanical grep-first workflow: Cheap pattern scans run before expensive file reads, with a hard cap on duplication searches to control token usage. - Structured verdict report: Produces APPROVE / FIX FIRST / BLOCK verdicts with path:line findings, plus drift-pattern candidates for the project's standards. - Use Case: After an agent implements a feature per a plan document, run this review to verify the diff matches the design's boundaries, catch a price ?? 0 silent fallback, and confirm no regression-surface consumers broke. ## Quick Start Ask the AI to review the completed change for the current task using the code-review procedure and report a verdict with findings.
