What is progressive disclosure in this context?

It is a design principle where you keep only the most essential, universal information at the root level and move specific, detailed guidelines to secondary files, which are linked and accessed only when relevant.

Does it work with files other than CLAUDE.md?

Yes, it is designed to refactor any agent-facing instruction file, including AGENTS.md, COPILOT.md, or custom project documentation used by AI agents.

How does this help with AI agent performance?

Large, monolithic instruction files waste context tokens and can confuse agents with conflicting rules. Streamlining the root file provides the agent with a clearer, faster, and more accurate project context.

Will this skill delete my existing instructions?

The skill identifies redundant, vague, or outdated instructions for potential deletion in Phase 5, but it flags them for your review so you maintain full control over your documentation.

Agent MD Refactor

Name: Agent MD Refactor
Author: softaworks

bysoftaworks

•

347

Productividad y Flujo de Trabajo

Refactors bloated agent instruction files into an organized, multi-file structure using progressive disclosure principles.

Acerca de

The Agent MD Refactor skill helps developers manage massive, monolithic instruction files like CLAUDE.md or AGENTS.md. By applying progressive disclosure, it extracts core project essentials into a minimal root file while delegating detailed conventions—such as TypeScript patterns, testing guidelines, and architecture rules—into specialized, linked sub-documents. This process reduces token waste, resolves instructional contradictions, and ensures AI agents receive only the most relevant context for any given task.

Características Principales

Prunes redundant or vague instructions to optimize AI context usage
Automatically categorizes detailed rules into specialized Markdown sub-files
Identifies and resolves conflicting instructions within monolithic files
347 GitHub stars
Extracts critical project metadata for a minimal, high-impact root file
Creates a clean, linked documentation hierarchy for easier maintenance

Casos de Uso

Standardizing instruction files across multiple microservices or repositories
Organizing chaotic developer notes into actionable, categorized agent guidelines
Slimming down a 500-line CLAUDE.md file that is causing token bloat

Acerca de

Características Principales

Prunes redundant or vague instructions to optimize AI context usage
Automatically categorizes detailed rules into specialized Markdown sub-files
Identifies and resolves conflicting instructions within monolithic files
347 GitHub stars
Extracts critical project metadata for a minimal, high-impact root file
Creates a clean, linked documentation hierarchy for easier maintenance

Casos de Uso

Standardizing instruction files across multiple microservices or repositories
Organizing chaotic developer notes into actionable, categorized agent guidelines
Slimming down a 500-line CLAUDE.md file that is causing token bloat