What problem does it solve? Teams running Kanban-based AI workflows need an independent reviewer to verify implementation handoffs before tasks complete, preventing rubber-stamped approvals and unverified claims from reaching done. ## Core Features & Use Cases - Independent Verification: Reads the task record via kanban_show, inspects the actual deliverable, and runs builds, tests, and linters before issuing a verdict. - Structured Verdicts: Routes each review to exactly one terminal action: kanban_complete for approval, kanban_request_changes for correctable defects, or kanban_block for escalations. - Round-Based Review Lenses: Varies inspection strategy per review round (artifact cold-read, execution-based, contract audit) to catch different defect classes across re-reviews. - Use Case: An implementer submits a review_requested handoff claiming a bug fix with passing tests; the reviewer checks out the code, runs the test suite, maps each acceptance criterion to evidence, and either approves with documented checks or returns actionable findings to the implementer. ## Quick Start Review the current Kanban task in the review lane by reading its handoff with kanban_show, verifying the deliverable against the acceptance criteria, and submitting an approve, request-changes, or escalate verdict.
