code-reviewer_skill

Overview

This skill provides a technology-agnostic, systematic code review methodology for assessing correctness, security, maintainability, and performance. It defines multi-pass review steps, severity classifications, checklists, feedback patterns, and a quality scoring model to produce actionable, prioritized review results.

How this skill works

Reviews follow a multi-pass process: correctness, security, maintainability, and performance. Each pass applies focused checks and checklists, classifies findings by severity, and produces constructive feedback using a what/why/how/severity template. A quality score is computed by weighting categories and deducting points for issues to summarize overall health.

When to use it

• During pull request reviews to standardize assessment across teams
• As a pre-merge gate to block critical or major issues
• For self-review by developers before submitting changes
• When auditing code for security and data-protection risks
• To generate objective quality reports for deliverables

Best practices

• Define project-specific natural-language constraints up front and load them into review configuration
• Run the multi-pass review in sequence and record findings per pass to avoid missed coverage
• Classify each finding with the provided severity levels and take the prescribed action
• Use the constructive feedback template: describe what, explain why, propose how, and set severity
• Acknowledge good patterns and keep tone focused on the code, not the author

Example use cases

• Block a merge when a security vulnerability or data-loss risk is detected (critical)
• Request changes for missing error handling or a major logic bug (major)
• Suggest stylistic or documentation improvements that can be merged (minor)
• Evaluate a release candidate by computing an overall quality score to decide readiness
• Automate checklist checks for input validation, auth, and test coverage during CI