What Is REA (Reverse Engineer Anything)? A Plain-English Guide

REA lets your AI agent open an app, read how it really works and report back with evidence, instead of guessing.

You see an app with a feature you love, and one question keeps coming back: how does it actually work? You cannot see the code, the maker is not going to explain it, and guessing from screenshots only gets you so far. That is the exact gap REA fills, and it is a tool most people building with AI have never heard of.

I found REA while looking for a way to make Claude Code understand an app instead of guessing about it. In this guide I explain what REA is, what it can look at, what a first session feels like, and where the limits are, in words that do not need a technical background.

What is REA?

REA stands for Reverse Engineer Anything. It is a free, open source tool (MIT license) that connects to your AI coding agent, such as Claude Code, and gives it a set of inspection tools. Instead of answering from general knowledge, your agent can open an app, read what is really inside it, and report back with evidence.

Reverse engineering simply means working out how something was built by studying the finished thing. A mechanic who takes an engine apart to see how it runs is reverse engineering. REA lets your AI do the same with software.

The official sources are the REA page on GitHub, the REA website and the rea-agents package on npm. The author warns that a similar repository name or a copied README does not make a download official, so only install from those three places.

How does REA work with Claude Code?

REA uses something called MCP (Model Context Protocol), which is a standard way to plug extra tools into an AI agent. Once REA is set up, this is the loop:

1. You give your agent a target and a question, for example "how does the search feature in this app work?"
2. The agent calls REA's tools to inspect the target.
3. REA returns what it found, along with the evidence and its own limits, so you can see what it knows and what it is unsure about.
4. The agent explains it to you, or writes and tests code based on it.

REA analyzes targets locally on your computer. Your agent receives the results, and the company behind your AI model has its own data policy, so check yours if the target is sensitive.

What I like most is the evidence part. A normal chat answer sounds confident even when it is guessing. REA hands the agent real findings to point at, which makes the answers easier to trust and easier to check.

What can REA look at?

A lot. The official README lists many target types. These are the ones that matter most for beginners:

• Websites. Page structure, scripts, network activity and screenshots, using a Chrome-family browser.
• JavaScript and Electron apps. Many desktop apps (the kind built with web technology) are really packaged websites. REA can map their modules, routes and how their parts talk to each other.
• Android apps (APK files). The manifest, classes and decompiled methods.
• Native programs on your computer. Pseudocode, assembly, strings and symbols. This deeper work uses a separate analysis program such as Hopper, Ghidra or IDA.
• .NET programs, saved network captures, SQLite databases and more.

You do not need to understand most of that list. For a first session, a website or a simple app is plenty. If you are new to Claude Code itself, start with what Claude Code is and Claude Code for beginners first.

What does a first session look like?

The setup is one command. REA needs Node.js (a free program that runs JavaScript tools) in a recent version, then you run this in your terminal:

npx rea-agents setup

It asks which agents you use, shows the changes it wants to make, and backs up your existing configuration. You approve it, restart your agent, and REA appears in the tool list. The full walkthrough with screenshots is in my tutorial, Reverse Engineer Any App with REA and Claude Code. If you have not installed Claude Code yet, do that first with How to Install Claude Code.

Then you ask in plain words. The official README gives this example:

Understand how search works in the Notes app, show me the evidence, and build a similar feature for my project.

Swap in your own app and feature. When I ran my own case study, I pointed REA at a city life game and Claude Code reported that it was a Next.js app with a 3D map and sign-up flow before it built a working version. It was close on the first go, not perfect. One prompt rarely builds a whole app, so expect to follow up with several small edits. That is normal, and it is where most of the learning happens.

What can you use REA for?

Here are the uses I think are worth your time as a beginner or a business owner:

• Learning. Understand how a feature you admire is built, then practise building your own version in your own words.
• Rescue work. Recover how an old app of your own worked after the source code was lost.
• Interoperability. Work out how two programs talk to each other so you can connect them.
• Auditing. Check what software you own or are responsible for really does.
• Checking an idea. Study how a public site behaves before you build something of your own.

If you want a done-for-you method with ready prompts, the playbook Clone/Rebuild Any Website or Mobile App Using REA AI goes deeper. The free tutorial above is enough to get your first result.

What are the limits, and what is off limits?

REA is a powerful tool, so be honest with yourself about both what it cannot do and what you should not do.

Technical limits. Deep analysis of native programs needs a separate provider like Hopper, Ghidra or IDA, and some of those are commercial products you must license legally. Features differ by operating system. REA changes quickly, so update it often and read its own guides at rea.tools/guides when something behaves differently from this article.

Responsible use. REA's own disclaimer says it is for lawful research, education, interoperability, software preservation and analysis of software you own or are authorized to examine. It does not support bypassing DRM or license checks, tampering with online services, getting around logins, or copying proprietary software and its protected code and assets. Terms of service and local laws differ, and you are responsible for your own use.

My simple rule: learn from how apps work, then build something original of your own. Do not copy someone else's product and pass it off as yours. This guide is for educational purposes only.

Where to go next

Pick one small target you are allowed to study, such as a public website or an app of your own, install REA and ask one question about it. The first answer with real evidence behind it is the moment this tool clicks.

I put together a free cheat sheet with the targets REA can handle, ten starter prompts and a short responsible-use checklist, so you do not start from a blank box. Then try the hands-on REA tutorial and tell me what you pointed it at.