Docs › Skills › code-review-and-quality

code-review-and-quality

Conducts multi-axis code review. Use before merging any change. Use when reviewing

138 lines guide: CHANGE-SIZING guide: FIVE-AXIS guide: REVIEW-PROCESS

Overview

Code review and quality conducts multi-dimensional review across five axes: correctness, readability, architecture, security, and performance. Every finding is labeled with a severity level — blocking, important, nit, suggestion, learning, or praise — and the reviewer must state when verification tools cannot reach the real world (TOOL_GAP).

Use this skill before merging any PR or change, after completing a feature, when evaluating code from another agent, or after any bug fix. It includes an AI-generated code checklist targeting hallucinated dependencies, subtle correctness gaps, superficial error handling, and over-engineered abstraction.

The output is a structured review report with severity-labeled findings, covering all five axes, with blocking items clearly identified. The approval standard is that the change must definitely improve overall code health — not that it must be perfect.