How does the skill identify which code to review?

It uses git SHAs to define a range (Base SHA to Head SHA), allowing the AI to focus specifically on the changes made since the last milestone or task.

When is it mandatory to request a code review?

Review is mandatory after completing each task in subagent-driven development, after finishing a major feature, and immediately before merging changes into the main branch.

What is a code-reviewer subagent?

A code-reviewer subagent is a specialized AI persona configured to review your work objectively, comparing the actual code changes against the project's requirements or design documents.

How should I handle feedback from the AI reviewer?

Critical issues should be fixed immediately, and Important issues should be addressed before proceeding. If you believe the reviewer is incorrect, you can provide technical reasoning and evidence to push back.

Can this skill be used for ad-hoc development?

Yes, it is highly effective for ad-hoc development to check for logic errors before a merge or to get a baseline check before starting a major refactor.

Automated Code Review

Name: Automated Code Review
Author: marcos-abreu

bymarcos-abreu

•

コラボレーションツール

Dispatches a specialized code-reviewer subagent to analyze implementations against requirements and catch issues before they cascade.

The Requesting Code Review skill enhances the development workflow within Claude Code by automating the feedback loop. By invoking a dedicated subagent, it performs thorough inspections of code changes against original plans or technical requirements using git commit history. This skill is designed to be integrated into subagent-driven development, ensuring that every task is validated for quality, logic, and architectural consistency. It categorizes feedback into critical, important, and minor issues, allowing developers to address potential bugs or technical debt early in the lifecycle before merging into the main branch.

主な機能

01Workflow integration for task-based development

02Subagent-driven implementation analysis against original plans

03Categorized issue reporting (Critical, Important, Minor)

04Git-integrated diff comparison using HEAD and Base SHAs

05Structured feedback loop for resolving technical issues

061 GitHub stars

ユースケース

01Gaining a fresh technical perspective when stuck on complex bugs or refactoring

02Performing a final quality audit before merging a major feature into main

03Verifying individual task completion in subagent-driven development

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add marcos-abreu/connect_kit requesting-code-review

For use in Claude.ai and ChatGPT

Download Skill