Can I use this for specific architectural patterns?

Yes, the skill references detailed checklists for SOLID principles, API design, and test coverage that can be loaded based on the specific needs of the review.

Does this skill support security audits?

Yes, it includes specialized, OWASP-aligned security checklists to identify vulnerabilities and authentication issues during the review process.

How does it handle false positives?

The skill uses a confidence threshold system where minor or low-impact issues require a very high certainty (up to 95%) before they are reported to the user.

What is the Changed Lines Rule?

It is a core principle that limits the AI's review scope strictly to added or modified lines in a diff, ensuring it does not report pre-existing issues in the codebase.

How are review findings presented?

Findings are reported in a structured table format that includes the file path, line number, issue type, a brief description, and a concrete fix for every item.

Code Review Guidelines

Name: Code Review Guidelines
Author: dnlopes

bydnlopes

0•

Collaboration Tools

Standardizes AI-driven code reviews by enforcing diff-only analysis and pragmatic, actionable feedback principles.

The Code Review Guidelines skill provides Claude with a rigorous framework for evaluating pull requests and code changes with professional precision. It enforces a strict 'Changed Lines Rule' to eliminate noise by preventing the AI from commenting on pre-existing code issues, ensuring focus remains entirely on new or modified lines. By implementing standardized output formats, severity classifications, and confidence thresholds, this skill transforms Claude into a reliable peer reviewer that delivers high-signal, evidence-based feedback across quality, security, and architectural domains.

Key Features

01Modular checklists for security, code quality, and API contract validation

020 GitHub stars

03Strict diff-only focus to eliminate noise from pre-existing code

04Confidence threshold logic to significantly reduce false positive reports

05Severity-based classification system for prioritizing merge-blocking issues

06Standardized reporting format with file paths, line numbers, and fixes

Use Cases

01Providing developers with immediate, actionable feedback before manual peer review

02Automating pull request reviews to catch logic errors and security flaws early

03Ensuring consistent coding standards and maintainability across team repositories

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add dnlopes/claude-code-plugins code-review-guidelines

For use in Claude.ai and ChatGPT

Download Skill