What problem does it solve? Code reviews done by the same agent that wrote the code tend to rubber-stamp it, and single-pass reviews miss cross-cutting issues. This Skill runs independent, dimension-focused reviewers with adversarial verification so findings are confirmed before they reach the report. ## Core Features & Use Cases - Two review modes: Light mode dispatches one independent reviewer against quick checklists for ordinary PR reviews; Deep mode runs a full multi-agent pipeline (dimension reviewers, verification, duplicate consolidation, structured report) only on explicit request. - Fourteen review dimensions: logic, security, performance, release risk, compatibility, UX, observability, reuse-architecture, and more, each with executable rules and pruning criteria. - Anti-hallucination verification: candidate findings are independently falsified and labeled confirmed, false positive, or need-more-context, with severity, likelihood, and release-blocking status. - Use Case: After finishing a feature branch, ask for a deep review to get a structured report with a merge verdict, P0/P1/P2 findings, pre-deploy checklist, and legacy issues handed off to their owners. ## Quick Start Ask the agent to run a deep review on the current branch or a specific PR, for example: run deep review on PR #123.
