Code Review for Requirements and Design

Problem

After implementation, code review often relies on judgment: a requirement “looks done,” and a design change “should be fine,” without item-by-item evidence. This skill is aimed at engineering review, aligning the implementation with feature docs, requirement / clarification docs, and high-level design docs to surface missing requirements, unhandled edge cases, and changes that deviate from the design.

How It Works

• Material gating: it first lists the recognized inputs; if key documents are missing, it stops the workflow, explains the impact, and asks whether to continue.
• Review baseline: extract feature items, business rules, and edge cases from requirements; extract interfaces, data flow, logic changes, and change scope from the design.
• Located code review: use code locations from the docs to inspect the implementation, checking requirement coverage and design consistency, while looking for common risks such as null checks, edge values, exception handling, concurrency safety, hard-coded values, and resource release.
• Conclusion output: return review passed or review not passed, categorize issues as incomplete, bug, or design deviation, and tie each issue to a concrete code location.

Fit and Limits

Best suited when requirements, design, and code locations are available. If inputs are only verbal descriptions, requirements are unclear, or design docs are missing, the reliability of the conclusion drops.