reducing-entropy

Overview

Reducing Entropy is a manual-only skill from the softaworks/agent-toolkit designed to minimize total codebase size by biasing every decision toward deletion and simplification. It challenges common development instincts by measuring success not by effort or organization, but by the final amount of code in the codebase after changes are applied.

Key Features

• End-State Measurement - Evaluates changes based on the final codebase size rather than effort, churn, or perceived organization improvements
• Three Core Questions - Systematic framework asking "What's the smallest codebase?", "Does this result in less total code?", and "What can we delete?"
• Red Flag Detection - Identifies rationalizations like "keep what exists" (status quo bias), "adds flexibility" (YAGNI), "better separation" (more code), and "type safety" (worth how many lines?)
• Reference Mindset System - Requires loading at least one philosophical reference mindset from the references directory before proceeding with any refactoring
• Clear Exceptions - Acknowledges when this approach does not apply: already minimal codebases, frameworks with strong conventions, and regulatory requirements

Who is this for?

This is for developers conducting code reduction exercises who want a systematic approach to minimizing codebase size without losing functionality. It is ideal for teams fighting codebase bloat who need an opinionated framework that challenges assumptions about code organization, flexibility, and separation of concerns.