Getting Started with agents-reverse-engineer in 5 Minutes
agents-reverse-engineer (ARE) is a CLI tool that generates AI-friendly documentation for your codebase automatically. Instead of manually explaining your project to AI coding assistants, ARE analyzes your code and creates comprehensive documentation that AI models can read instantly.
Prerequisites
You'll need Node.js 18 or higher and an AI coding assistant like Claude Code, Gemini CLI, OpenCode, or Codex. ARE works with existing codebases in any state.
Step 1: Install agents-reverse-engineer
ARE integrates directly with your AI assistant through slash commands. Run:
npx agents-reverse-engineer@latest
The installer prompts two questions:
- Select your runtime (Claude Code, Gemini CLI, OpenCode, Codex, or all)
- Installation location: Global (~/.claude/) for all projects, or Local (./.claude/) for this project only
For non-interactive installation, use flags:
# Global installation for Claude Code
npx agents-reverse-engineer@latest --runtime claude -g
Step 2: Initialize Configuration
In your AI assistant, navigate to your project and run:
/are-init
This creates .agents-reverse-engineer/config.yaml with defaults that work for most projects.
Step 3: Discover Your Codebase
Run the discovery command:
/are-discover
ARE scans your project, respects .gitignore patterns, filters binaries, and creates a generation plan showing checkboxes for each file and directory.
Step 4: Generate Documentation
Run the generation command:
/are-generate
ARE executes a two-phase pipeline:
Phase 1: File Analysis - ARE analyzes each source file in parallel, extracting imports, exports, purpose, and dependencies. It writes a .sum file next to each source file.
Phase 2: Directory Aggregation - ARE builds directory summaries from deepest to shallowest, ensuring child documentation is available when building parent summaries.
What You Get
After generation, you have three documentation types:
1. File Summaries (.sum files) - Each source file gets a .sum file with YAML frontmatter and structured sections covering purpose, public interface, dependencies, and implementation notes.
2. Directory Summaries (AGENTS.md) - Each directory gets an aggregated summary of its contents and architecture.
3. Root Pointer (CLAUDE.md) - A simple @AGENTS.md pointer at your project root that tells your AI assistant to automatically load project documentation.
Keeping Documentation Updated
As you modify code, run /are-update. This compares file hashes, regenerates only changed files, updates parent directories, and removes documentation for deleted files.
Next Steps
You now have a fully documented codebase that your AI assistant understands. The documentation lives alongside your code and updates as your project evolves.