概要
The Reviewing Code skill provides a rigorous framework for auditing code changes through multiple dimensions including security, logic correctness, and performance. It utilizes a 100-point scoring system and specialized review strategies—ranging from single-agent analysis for small PRs to parallel multi-agent reviews for large-scale changes. By cross-referencing implementation against local documentation such as API contracts and feature specs, it ensures that every code submission remains architecturally sound and functionally accurate while prioritizing critical vulnerabilities and broken logic.
