structured-code-review

The Structured Code Review skill transforms how AI agents evaluate pull requests by applying a rigorous, multi-stage methodology. Instead of performing a single, unstructured scan, this skill enforces a systematic inspection process that ensures no aspect of code health is overlooked. It is designed for developers, software engineers, and automated agents who need to maintain high standards in collaborative environments, ensuring that every submission is reliable, maintainable, and secure.

• Performs a distinct five-stage inspection process: Requirements Compliance, Correctness, Code Quality, Testing, and Security/Performance.

• Uses targeted checklists for each phase to catch logic bugs, scope creep, naming inconsistencies, edge case handling, and vulnerabilities like SQL injection or exposed secrets.

• Categorizes feedback into clear levels: Blocker (must fix), Major (should fix), Minor (suggestion), and Nit (style/trivial).

• Provides a standard, actionable feedback template that identifies the issue, explains why it matters, and suggests a specific improvement.

• Includes a comprehensive Review Checklist Summary for consistent reporting at the end of each session.

• Integrates with other professional engineering skills like verification gates, test patterns, and anti-pattern analysis to create a cohesive development workflow.

• Trigger this skill by asking the agent to 'review my code', 'check this PR', 'review my changes', or 'provide code feedback'.

• Use it specifically when you need to ensure code follows project conventions, meets functional requirements, and passes security audits.

• The output is structured to be developer-friendly, encouraging direct actionable changes rather than general observations.

• Ideal for use during the pre-merge phase to prevent regression and technical debt.

• Works best on clean, modularized codebases; highly effective at spotting missing edge case handling and suboptimal architectural abstractions.