Stats
Actions
Tags
From entropy-reducer
Use this skill when the user mentions "reduce entropy", "simplify code", "reduce code", "less code", "code reduction", "remove dead code", "deduplicate", "consolidate code", "code cleanup without changes", or wants to achieve the same functionality with fewer lines of code. Also applies when user says "entropy reducer" or "reduce-entropy".
How this skill is triggered — by the user, by Claude, or both
Slash command
/entropy-reducer:entropy-reductionThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Orchestrates a multi-agent workflow to analyze a codebase and find ways to achieve the same
Orchestrates a multi-agent workflow to analyze a codebase and find ways to achieve the same functionality with less code — without changing UI/UX, switching languages, or introducing risk.
Always invoke via the /reduce-entropy command, which enters plan mode automatically.
| Phase | Agent | Model | Purpose |
|---|---|---|---|
| 1. Familiarization | codebase-reviewer | haiku | Fast read-through, LOC counting, architecture mapping |
| 2. Analysis | entropy-analyzer | sonnet | Deep analysis of reduction opportunities |
| 3. Safety | safety-auditor | sonnet | Validate each change is safe and performant |
| 4. UI Guard | ui-guardian | haiku | Verify zero UI/UX impact |
Never use opus. Haiku for fast/simple tasks, sonnet for deep analysis.
This skill works with any programming language. If the agents encounter an unfamiliar language, they will use WebSearch to learn its syntax, idioms, and best practices before analyzing.
Never recommend changing the programming language. Reduce entropy within the existing stack.
The final deliverable is a structured plan showing:
The plan is presented to the user for approval before any changes are made.
npx claudepluginhub garretts-apps/entropy-reducer --plugin entropy-reducerRefactors code by removing dead code, extracting duplication, and simplifying deep nesting while preserving behavior. Runs tests before and after to confirm nothing breaks.
Refactors codebases to reduce entropy by diagnosing and incrementally fixing structural, semantic, behavioral, and evolutionary disorder without breaking behavior.
Simplifies complex/AI-generated code by removing duplication, dead code, over-engineering, and bloat patterns like verbose error handling. Use after features, on long functions (>40 lines), deep nesting (>3), or repeated patterns.