What is REA?
REA is an open-source toolkit (npm package rea-agents, GitHub morluto/rea) that exposes one CLI and MCP server with 100+ reverse-engineering commands: static bundle analysis, CDP browser capture, process capture, Electron scenarios, APK resource decoding, decompilation via Ghidra/IDA/Hopper, and more. The pitch is simple: instead of wiring up Ghidra scripts yourself, your AI coding agent (Claude Code, Codex, Cursor, Gemini CLI) calls REA commands directly.
Why this site exists
When we evaluated REA, every search result was either the official README or thin AI-generated content with no evidence anyone had ever run it. So we ran it — on our own projects — and published the inputs, the outputs, the errors, and the workarounds. If a page here says "9 scripts analyzed, 40,388 AST nodes visited", those numbers came from an actual terminal session you can reproduce.
The compliance line we don't cross
Every workflow on this site targets software we own or operate (including this site's sibling projects). Reverse engineering is a legitimate skill for security research, interoperability and debugging — but only on software you own or are explicitly authorized to analyze. Don't be the reason tools like this get locked down.
FAQ
Do I need Ghidra or IDA to use REA?
No. A large subset of REA — all the web, JavaScript and Electron analysis commands — works with zero external engines. Native binary decompilation providers (Ghidra, IDA, Hopper, jadx) are optional add-ons.
Does REA work with Claude Code?
Yes — that's its primary use case. Run npx rea-agents@latest setup and it registers an MCP server + skill with detected agents. It also works standalone from any terminal.
Is it free?
REA itself is open source and free. Optional analysis engines have their own licenses (Ghidra is free; IDA and Hopper are commercial).