Multi-Runtime Support: Claude Code, Gemini CLI, OpenCode, and Codex

The AI coding assistant landscape is diversifying fast. agents-reverse-engineer (ARE) was designed to work across multiple AI assistants, generating documentation that works seamlessly with all of them.

The AGENTS.md Standard

AGENTS.md is an emerging cross-tool standard for AI-readable project documentation. Like README.md for humans, AGENTS.md provides AI-readable documentation in a consistent, hierarchical structure. ARE generates this through its two-phase pipeline: parallel file analysis followed by hierarchical directory aggregation. Because AGENTS.md is a standard format, any tool that supports it will automatically benefit — even tools that don't exist yet.

Runtime-by-Runtime Breakdown

Claude Code: Full Integration

Claude Code has the most mature integration, with full slash commands (/are-generate, /are-update, etc.) and three automated session hooks. are-check-update.js performs version checks, are-context-loader.js auto-injects parent AGENTS.md files into context, and are-session-end.js triggers auto-updates on git changes. Uses CLAUDE.md pointer file.

Gemini CLI: Full Support

Receives the same level of support with runtime-specific adaptations. Uses GEMINI.md pointer file, identical slash commands, and equivalent session hooks for version checking, context loading, and session-end updates.

OpenCode: AGENTS.md Native

Supports AGENTS.md files directly without additional configuration. Uses OPENCODE.md pointer file, full command suite, and hooks stored in .opencode/hooks/ directory.

Codex: Local Context Rules

Uses AGENTS.override.md pattern instead of runtime-specific pointer files. Works with Codex's local context rule system for integrating generated documentation.

Installation for Multiple Runtimes

Install for all runtimes simultaneously:

npx agents-reverse-engineer@latest --runtime all -g

Or target specific runtimes:

npx agents-reverse-engineer@latest --runtime claude -g

What's the Same Across Runtimes

What's Different

Backend Selection

Runtimes determine where docs are consumed; backends determine which AI generates them. You can mix and match: generate with Claude for quality, consume with Gemini for speed.