What problem does it solve? Choosing the right depth and type of code review for each change is error-prone: teams either under-review risky diffs (auth, payments, AI-authored code) or over-review trivial ones, and AI-generated code hides defect classes that pass the happy path. This Skill routes every change to the correct review agent(s) so reviews are mandatory, proportionate, and consistent before merge. ## Core Features & Use Cases - Review agent routing: Maps your change to one of 25+ specialized reviewers (pr, spec, tests, security, language, CI, IaC, API contract, E2E, accessibility, and more), with /review all running the four PR-scoped agents in parallel. - AI-authored code posture: Treats every diff as AI-generated by default, checking for hallucinated APIs, hardcoded secrets, missing authorization, and happy-path-only logic, with mandatory security review for auth, payments, PII, and security boundaries. - Dispatch discipline: Enforces file-path-based artifact passing, single fix-agent per findings list, re-review until PASS, and severity-based action (Critical / Important / Minor). - Use Case: After implementing a feature with subagents, run /review pr after each task to catch issues early, then /review all before merge — no merge proceeds with any Critical finding. ## Quick Start Ask the AI to review your current changes by saying "review my PR" and it will route to the appropriate review agents automatically.
