What types of issues can this skill detect?

It is designed to categorize issues into specific domains including Security (SEC), Architecture (ARC), Testing (TST), UX (UXD), Readability (RDY), Documentation (DOC), and Performance (PRF).

Does this skill provide actual code fixes?

Yes, a core requirement of the methodology is that every finding must include a concrete 'Suggestion' field providing a specific fix rather than vague advice.

How does the confidence scoring work?

Findings are scored on a 0-100 scale based on evidence strength and reproducibility. To ensure high quality, only findings with a confidence score of 80 or higher are reported.

What are the severity levels used in the review?

The skill uses four levels: P1 for critical issues (security/crashes), P2 for important feature breaks, P3 for minor bugs, and P4 for trivial or cosmetic changes.

Structured Code Reviewer

Name: Structured Code Reviewer
Author: tenzir

bytenzir

•

Seguridad y Pruebas

Performs code reviews using a rigorous methodology based on confidence scoring and severity ratings.

The Reviewing Changes skill equips Claude with a professional-grade methodology for evaluating code changes. It utilizes a dual-dimension rating system that separates the severity of an impact from the confidence in the finding, ensuring that only high-certainty, high-impact issues are reported. By following a standardized output format that includes specific evidence, reproducibility criteria, and actionable suggestions, this skill transforms code reviews into objective, data-driven assessments that help teams prioritize security, architecture, and performance fixes.

Características Principales

012 GitHub stars

02Evidence-based scoring criteria that factors in reproducibility, impact, and rule violations

03Mandatory actionability requirements ensuring every finding includes a concrete fix

04Dual-dimension rating system combining severity (P1-P4) with confidence scoring (0-100%)

05Standardized issue identification using specific prefixes for security, architecture, performance, and more

06Automated noise reduction by filtering out findings with less than 80% confidence

Casos de Uso

01Conducting deep security audits and architectural reviews on new pull requests

02Automating the identification of performance bottlenecks and readability issues with concrete evidence

03Standardizing team review outputs to ensure consistent quality and actionability

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add tenzir/claude-plugins reviewing-changes

For use in Claude.ai and ChatGPT

Download Skill