What problem does it solve? Code changes are often declared complete immediately after implementation without verifying correctness, integration, or edge cases, leading to regressions, broken pipelines, and fabricated completion claims. This Skill enforces a mandatory self-review and verification cycle before any task is marked done. ## Core Features & Use Cases - Mandatory Verification Workflow: Enforces a staged pipeline from requirement understanding through diff inspection, requirement-by-requirement verification, integration review, edge-case analysis, testing, and regression checks. - Domain-Specific Reviews: Includes dedicated checklists for RAG pipelines, trace analysis, multi-worker concurrency, security, and documentation consistency. - Structured Final Report: Produces a standardized completion report with requirement status (PASS/PARTIAL/FAIL/NOT VERIFIED), test evidence, remaining risks (P0-P3), and an explicit final verdict. - Use Case: After fixing a retrieval bug in a RAG application, use this Skill to inspect the actual git diff, verify each requirement against runtime evidence, run regression tests, and produce an evidence-backed completion report instead of assuming the fix works. ## Quick Start Ask the AI to apply the self-review skill to verify the code change just implemented before declaring the task complete.
