What is the 'Iron Law' in this skill?

The Iron Law dictates that tests must never stay red; if a change breaks a test, the skill immediately reverts to the last green state to ensure codebase stability.

Does it use static analysis to find references?

It integrates directly with LSP (Language Server Protocol) for precise symbol searching, renaming, and dependency mapping, which is faster and more accurate than grep.

Can this skill handle large-scale refactors across many files?

Yes, for changes involving 10 or more files, it automatically suggests creating a PLAN.md and can delegate tasks to subagents for managed execution.

How does it identify what needs to be refactored?

It includes an Analysis Mode that scans for code smells like God components, kitchen-sink utilities, duplicate patterns, and redundant abstractions before presenting findings to the user.

Atomic Code Refactor

Name: Atomic Code Refactor
Author: cinjoff

bycinjoff

•

開発者ツール

Restructures and optimizes existing codebases while ensuring zero regressions through test-driven atomic changes and precise LSP analysis.

This skill enables developers to perform safe, structural improvements to their code without altering external behavior. It follows a strict 'Iron Law' where tests must remain green at every step, utilizing Language Server Protocol (LSP) for precision and memory-aware agents to analyze complex dependencies. From identifying 'God components' to executing multi-file renames and extracting patterns, it provides a disciplined framework for technical debt reduction and architectural evolution, ensuring every change is verified and committable.

主な機能

01Atomic Step Execution: Breaks large refactors into small, testable, and independently committable chunks.

021 GitHub stars

03LSP-Powered Analysis: Uses precise symbol lookups and reference tracking for accurate scope identification.

04Iron Law Enforcement: Automatically reverts changes if tests fail to guarantee zero regressions.

05Automated Anti-Pattern Detection: Identifies God components, duplicate patterns, and redundant abstractions.

06Context-Aware Memory: Leverages project history to recall past architectural decisions and avoids repeating mistakes.

ユースケース

01System-wide renaming or restructuring of APIs and utility functions with high confidence.

02Breaking down monolithic 'God classes' into smaller, single-responsibility modules.

03Eliminating code duplication and extracting shared logic into clean, reusable abstractions.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add cinjoff/fhhs-skills refactor

For use in Claude.ai and ChatGPT