awesome-skills/code-review-skill

What it solves

Code Review Skill transforms AI-assisted code reviews from vague, inconsistent feedback into a structured, expert-level process. It bridges the gap between automated linting and human oversight by providing a comprehensive, modular framework that guides Claude Code through a consistent review workflow.

How it works

The tool uses a "progressive disclosure" approach to keep the AI's context window efficient. A small core skill (approx. 220 lines) is loaded initially, and specific, detailed language or framework guides (ranging from 200 to 1,100 lines) are only pulled in when needed. The review follows a four-phase process: gathering context, performing a high-level architectural review, conducting a line-by-line analysis, and delivering a structured summary with actionable feedback.

Who it’s for

This tool is designed for developers and teams using Claude Code who want to standardize their code review process. It is ideal for those working across diverse technology stacks who need consistent, security-conscious, and professional feedback on their pull requests.

Highlights

• Broad Compatibility: Supports over 20 programming languages and frameworks, including React, Java, Go, Rust, and C++.
• Severity Labeling: Categorizes feedback into six levels, ranging from blocking issues to praise for excellent work.
• Security-First: Includes dedicated security checklists for every supported language ecosystem.
• Collaborative Tone: Prioritizes helpful questions and suggestions over rigid commands.
• Automation Awareness: Clearly distinguishes between issues that should be caught by automated linters and those requiring human-level judgment.