repo-structurer

Cleans up and structures a repository for production. Adds README files to every directory, modularizes code, removes dead files, and makes the repo LLM-smart. Use when the user asks to "clean up", "s

SKILL.md

GitHub

Repo Structurer

Makes any repository clean, documented, and agent-friendly by adding READMEs, improving organization, and removing clutter.

How It Works

  1. Analyze the current directory structure
  2. Identify directories missing READMEs
  3. Generate context-aware READMEs for each directory
  4. Identify dead code and unused files
  5. Suggest modularization improvements
  6. Create agent configuration files (AGENTS.md, .cursorrules)

Step-by-Step Procedure

Step 1: Map the Directory Structure

find . -type d -not -path './.git/*' -not -path './node_modules/*' -not -path './.next/*' -not -path './dist/*' -not -path './__pycache__/*' -not -path './venv/*'

Step 2: Check README Coverage

# Find directories without READMEs
for dir in $(find . -type d -not -path './.git/*' -not -path './node_modules/*' -not -path './.next/*' -maxdepth 3); do
  if [ ! -f "$dir/README.md" ]; then
    echo "MISSING: $dir"
  fi
done

Step 3: Generate READMEs

For each directory missing a README, create one following this template:

# {Directory Name}

{One sentence describing what this directory contains and why it exists.}

## Contents

{List the key files/subdirectories with one-line descriptions.}

## How It Works

{2-3 sentences explaining the role of this directory in the overall project.}

## Key Files

| File | Purpose |
|------|---------|
| `{file}` | {description} |

Rules for README generation:

  • Read 2-3 files in the directory to understand purpose
  • Keep it under 30 lines
  • Focus on WHAT and WHY, not HOW (the code shows how)
  • Make it useful for AI agents navigating the codebase
  • Include relationships to other directories if relevant

Step 4: Identify Dead Code

  • Files not imported anywhere
  • Commented-out code blocks > 10 lines
  • Unused dependencies in package.json / requirements.txt
  • Empty files or placeholder files
  • Duplicate files with slightly different names

Step 5: Suggest Modularization

Look for:

  • Files > 300 lines that could be split
  • Functions > 50 lines that could be extracted
  • Mixed concerns (API routes + business logic + database queries in one file)
  • Utility functions scattered across multiple files

Step 6: Create Agent Config

AGENTS.md (if not exists):

  • Describe the project purpose
  • List the tech stack
  • Document key commands (build, test, deploy)
  • Set boundaries (what agents should/shouldn't modify)

.cursorrules (if not exists):

  • Define code style preferences
  • Set framework-specific patterns
  • Include file organization rules

Output Format

## 🧹 Structure Report

### README Coverage
- Before: X/Y directories documented
- After: Y/Y directories documented
- Files created: {list}

### Dead Code Found
- {file}: {reason}

### Modularization Suggestions
1. {suggestion with rationale}

### Agent Config
- AGENTS.md: {created/updated/already exists}
- .cursorrules: {created/updated/already exists}

Important

  • ALWAYS ask before deleting files
  • ALWAYS explain WHY a change is recommended
  • Keep READMEs concise — agents have limited context windows
  • Prioritize directories that other developers/agents interact with most