review-code

To generate a multi-dimension code review report, instruct Claude to review a specific directory, such as "review code in src/". The Skill applies a rule-based engine to analyze correctness, readability, performance, security, testing, and architecture, outputting a structured Markdown report.

A structured code quality review analyzes six dimensions: correctness, readability, performance, security, testing, and architecture. Each dimension is evaluated using JSON rule files to surface actionable findings within a final Markdown summary.

Yes, this code review tool applies to codebases across languages and architectures. It uses a generic rule-based engine with JSON rule files under specs/rules to evaluate code quality consistently regardless of the specific programming language.

Static analysis handles architecture and security findings by applying rule-based checks from JSON rule files under specs/rules. It surfaces prioritized, actionable issues within these specific dimensions and compiles them into a structured Markdown report.

The automated code review is driven by JSON rule files located under the specs/rules directory. You can modify or add JSON rule files within this directory structure to customize the analysis criteria across the six supported review dimensions.

The best way to trigger the code review process is by using natural language commands like "review code" or "code review" followed by the target directory. This seeds the automated governance workflow and produces a repeatable quality signal.