What problem does it solve? It enforces consistent, evidence-grounded code review inside a multi-agent software engineering workflow, covering implementation source review before API/E2E, proportional test-code review after successful runs, and failure-origin classification after failed runs. ## Core Features & Use Cases - Three review entry points: full implementation-source and structural review, lightweight durable test-code review, and focused API/E2E failure-origin analysis, each with distinct standards. - Canonical report templates: produces code-review-report.md, api-e2e-test-review-report.md, and a chronological code-review-revision-record.md with CRR-* entries tracking every review round. - Classification and routing: classifies outcomes as Local Fix, Design Impact, Requirement Gap, or Unclear and routes handoffs to implementation, API/E2E, or solution design owners. - Use Case: After an implementation engineer hands off a feature change, the reviewer traces the production behavior path, applies structural and design checks, scores ten quality categories, and either passes the package to API/E2E testing or routes classified findings back to the owning specialist. ## Quick Start Review the implementation handoff in the task workspace against the design spec and produce the code review report with findings, scorecard, and routing decision.