What is the 'zero issues' policy in this skill?

The skill is designed to treat all flagged items—including minor ones—as mandatory fixes. It will continue the review-fix loop until the code reviewer agent confirms that no issues remain.

How does it handle large files or context limit timeouts?

If a review fails due to size, the skill automatically retries with a focused scope, looking only at modified files or splitting the review into smaller logical chunks.

What happens if the AI code reviewer makes a mistake?

The workflow allows for human intervention. You can push back with technical reasoning, provide evidence through tests, or adjust your workflow policy to accept specific findings.

Can I use this skill while working on multiple features simultaneously?

Yes. It uses a session-isolated SCRATCHPAD_DIR to ensure that scratch files and temporary data from one review session do not interfere with others.

Automated Code Review & Fix Loop

Name: Automated Code Review & Fix Loop
Author: ed3dai

byed3dai

•

协作工具

Automates the code review lifecycle by dispatching specialized subagents to identify issues, apply systematic fixes, and verify resolutions.

The requesting-code-review skill implements a robust, autonomous feedback loop designed to catch and resolve software defects before they reach production. By orchestrating a 'review → fix → re-review' workflow, it ensures that every implementation meets specific requirements and quality standards. The skill manages session isolation via namespaced directories, handles complex git diffs, and integrates directly with bug-fixing agents to achieve a 'zero issues' state. It is particularly effective for maintaining high code quality during long-running feature development or as a final gatekeeper before merging critical pull requests.

主要功能

015 GitHub stars

02Autonomous review-fix-verify loop that persists until all issues are resolved

03Multi-level issue categorization covering critical bugs, important logic, and minor style

04Persistent issue tracking to ensure previous bugs don't regress during fix cycles

05Intelligent retry logic that handles context timeouts by narrowing review scope

06Session isolation using SCRATCHPAD_DIR to prevent collisions in parallel workflows

使用场景

01Automating the remediation of linter, test, and architectural violations discovered during development

02Conducting a comprehensive final audit of a feature branch before merging to main

03Validating individual task completion during complex implementation plans

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add ed3dai/ed3d-plugins-testing requesting-code-review

For use in Claude.ai and ChatGPT

Download Skill