reducing-entropy

Detects and deletes dead code, unused abstractions, and over-engineered patterns across codebases.

Updated Aug 2, 2026
One-click install
npx skills add https://github.com/leonardoacosta/skills --skill reducing-entropy-leonardoacosta
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: reducing-entropy
Source: https://github.com/leonardoacosta/skills/tree/main/leo-core/skills/reducing-entropy
Command: npx skills add https://github.com/leonardoacosta/skills --skill reducing-entropy-leonardoacosta

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Codebases accumulate dead code, unused abstractions, speculative features, and copy-pasted components that increase maintenance cost. This Skill provides a systematic contract for finding and safely deleting that entropy, with a bias toward deletion over abstraction. ## Core Features & Use Cases - Dead Code Detection: Identifies unused tRPC procedures, dead monorepo packages, over-parameterized Drizzle schemas, redundant types, and valueless barrel exports, with optional fallow static-analysis integration. - Safe Deletion Rules: Enforces verification of dynamic callers (createCaller, raw SQL, string interpolation) before deletion, plus PAGNI exceptions for infrastructure that is expensive to retrofit. - Ranked Cut Report: Outputs findings in a cut-format ranked biggest-cut-first with delete/stdlib/native/yagni/shrink tags and a net line-count summary. - Use Case: During a code review of a T3 Turbo monorepo that has grown messy, run a repo-wide sweep to find unused tRPC procedures and single-consumer packages, then delete them after verifying no dynamic callers exist. ## Quick Start Ask the agent to audit this codebase for dead code and over-engineering, then report the biggest cuts first.

Frequently Asked Questions about reducing-entropy

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I find and remove dead code in a TypeScript monorepo?

Run a static-analysis sweep with fallow dead-code detection or grep-based heuristics to find unused exports, procedures, and packages. Verify each finding has no dynamic callers before deleting, then report cuts ranked by size.

How to detect unused tRPC procedures in a codebase?

List routers with grep for createTRPCRouter, then search for each procedure name across apps and packages. Also check createCaller call sites and string-interpolated names before deleting, since static grep alone can miss dynamic usage.

When should I not apply YAGNI to code deletion?

YAGNI does not apply to PAGNIs: patterns like created_at timestamps, audit logs, API versioning, and pagination that are dramatically cheaper to add at creation than to retrofit. Check the expensive-to-add-later reference before deleting infrastructure-like patterns.

What is the difference between reducing-entropy and simplify skills?

Reducing-entropy runs a repo-wide dead-code and over-engineering sweep ranked biggest-cut-first. Simplify is scoped to recently changed code in a diff. Use simplify for reviewing a changeset and reducing-entropy for whole-codebase cleanup.

Can dead code deletion break dynamic tRPC or SQL callers?

Yes, if deletion relies on static grep alone. The skill requires checking createCaller call sites, string-interpolated procedure names, and raw SQL template literals before removing procedures or Drizzle columns.