Reverse Engineer Anything (REA) is the viral GitHub repo that lets AI agents like Claude Code and Codex look inside an app and rebuild it. I show what REA is, how its skill and MCP tools work, how to install it with Hopper for binary analysis, and how I used it to clone Wispr Flow, the voice-to-text app I use daily, with a hotkey, note taker and local transcription. My take: code is no longer the moat. Your brand and users are.

👋 𝗛𝗲𝘆, 𝗜’𝗺 𝗘𝗿𝗶𝗰.
I spent seven years as a senior software engineer at Amazon and Microsoft. Now I build AI apps like bookzero.ai with Claude Code.

🔗 𝗟𝗶𝗻𝗸𝘀

• Reverse Engineer Anything on GitHub: https://github.com/morluto/rea
• Join our Skool community: https://www.skool.com/erictech/about
• Check out bookzero.ai: https://bookzero.ai

⌚️ 𝗦𝘁𝗮𝗺𝗽𝘀
00:00 Intro
00:44 What is REA?
01:30 How REA works
02:06 REA Setup & Installation
05:01 Clone demo
07:30 Outro

#reverseengineering #claudecode #aiagents

---

Transcript

0:00 If you want to reverse-engineer any

0:01 applications with AI that has the

0:03 highest accuracy, well, you got to check

0:05 out this repository called

0:06 reverse-engineer-anything.

0:08 This repository has got over 60,000

0:10 stars on GitHub just a couple days after

0:12 this repository has released. And people

0:14 were actually able to cloning any

0:15 application using this like Adobe

0:17 Premiere's and Photoshop and so many

0:19 more. And the most interesting part here

0:21 is that it's able to reverse-engineer

0:23 any applications like desktop

0:24 applications, games, web applications,

0:27 even mobile applications, and so much

0:29 more. So, that's why in this video I'm

0:31 going to show exactly what is REA, how

0:33 does it work behind the scene, and also

0:34 walk you through how I was able to clone

0:36 a Whisper flow, which is something that

0:38 I'm currently using every single day,

0:39 and try to clone it exactly the same.

0:41 So, with that being said, if you're

0:42 interested, let's get into the video.

0:44 Now, really quickly, what is REA? Well,

0:46 REA right here you can see it stands for

0:48 reverse-engineer-anything.

0:50 And just like what it says in the

0:51 description, let's say you see a feature

0:53 that you like. It could be a desktop

0:54 applications, web applications, or a

0:56 mobile applications. It will get down to

0:58 the binary level, understand exactly how

1:01 this application works, and giving AI

1:03 much more context to build a much actual

1:05 results. Now, in the past we already

1:07 have something like spec-driven

1:08 development before, right? Which we know

1:10 that if we were to have AI here to

1:12 building applications with the highest

1:13 accuracy, we need to have more context.

1:16 And that context is really the key. And

1:18 that's exactly what REA is trying to do.

1:20 If you want to have a clone anything,

1:22 well, you got to give more context. You

1:23 got to give more evidence on exactly

1:25 what this application works behind scene

1:27 to getting more accuracy in terms of the

1:29 actual building process. Now,

1:31 essentially the way how it works for REA

1:32 is you can treat it as like a skill. And

1:34 this skill here has many MCPs. Like for

1:37 example, Hopper for desktop, JavaScript

1:39 or browser here for web applications,

1:41 maybe there's mobile applications,

1:43 hardwares, and so many MCPs. And REA

1:46 here is a skill that teach AI agent on

1:48 how to using these tools and try to

1:50 collecting informations from those

1:51 applications itself. All right. So, now

1:53 you know exactly what is REA and how

1:55 does it work behind scene, let's take a

1:57 look at how we can use it to set it up

1:58 and actually use it to see what is the

2:00 accuracy result look like when it comes

2:02 to cloning an existing applications that

2:04 I'm currently using every single day.

2:06 All right, so the first step we're going

2:07 to do here is to run this command inside

2:08 of our repo. Now, before you do so

2:10 though, I highly recommend to check your

2:12 current node version. Simply just type

2:14 in node -v and it's going to tell you

2:16 exactly the current version you're using

2:18 inside of your node.js. Now, the

2:20 requirement for this is that you have to

2:22 be above 24, so version 24 above. And if

2:25 you want to upgrade it, so simply just

2:27 type in brew upgrade node and it's going

2:29 to upgrade that to the latest version.

2:31 So, once you have the latest version for

2:33 node, simply just copy this command

2:35 right here and what you're going to do

2:36 here is going to come back to your

2:37 terminal, paste that command and it's

2:39 going to run it.

2:40 So, what's going to happen here is it's

2:42 going to install this onto your local

2:44 machine. First of all, we're just going

2:45 to pick the ideal AI agent you're using,

2:48 so I'm choosing claw code and codex. So,

2:50 feel free to choosing open code or any

2:52 other AI agent harness you want. So,

2:54 once you are done that and then that's

2:56 what we have here is the install hopper

2:58 here for deep binary analysis. Now, you

3:00 can think of hopper here is kind of like

3:02 a third-party tool, right? That this REV

3:05 here uses or REA, sorry, uses that

3:07 basically just try to analyze exactly

3:09 what this application do, right? So,

3:11 usually the way how it works is

3:12 developer writes a source code and then

3:14 this source code here makes the

3:15 application work. Well, because we're

3:17 trying to reverse engineer this, so it's

3:19 going to using a tool called hopper here

3:21 to actually analyze the applications and

3:23 turning this application requirements

3:25 into proper so that we can be able to

3:27 using our AI agent here to kind of

3:29 reverse engineer it and try to build

3:30 that out, right? So, in this case we're

3:32 going to say yes and we're going to say

3:33 install this and then you can see here

3:35 that right now it's asking for applying

3:37 for five different changes. So,

3:39 installing the hopper deep analysis

3:40 providers as well as adding claw code,

3:43 codex, whatever coding agents you have

3:45 and also reverse engineering skills. So,

3:47 in this case I'm going to say yes, apply

3:49 for to both and it's going to go ahead

3:50 and do that. Now, right away you can see

3:53 that it's going to pop out this Hopper

3:54 Disassembler, and you can see that we

3:56 simply just click on try the demo, and

3:59 it's going to have everything good to

4:00 go. I'm going to say check automatically

4:02 here, and then what's going to happen

4:04 here, you can see we have everything's

4:05 are fully installed, right? So, you can

4:08 see the deeper analysis here, granted,

4:10 reverse engineer, pretty much

4:11 everything's are good to go.

4:14 Now, before we continue, I recently

4:15 launched our school community where I

4:17 help you to master AI agents,

4:18 automations, and building real AI

4:20 products. This is coming from someone

4:22 who used to work as a senior AI software

4:24 engineer at companies like Amazon and

4:26 Microsoft. And in this community, you're

4:28 going to get full AI builder roadmap

4:30 where you're going to start with AI

4:31 automations, then later on moving into

4:33 AI agents, research systems, SaaS

4:36 buildings, and eventually how to

4:38 productize and market your AI skills.

4:40 You're going to get the templates,

4:41 workflows, weekly live calls, and direct

4:44 community supports. So, when you get

4:46 stuck, you can ask questions and get

4:47 help right away. And the goal here is

4:49 not just watch more videos. It's

4:51 actually build, post your progress, give

4:53 feedback, and start to level up. So, if

4:56 you're interested, make sure you jump

4:57 right in, start the first course, and

4:58 I'll see you in a community.

5:01 So, now what we can do here is we can

5:02 open our claw code and maybe we just

5:04 explain any applications, and it's going

5:06 to basically try to show the evidence on

5:08 how to reverse engineer and build that

5:10 application out. So, in this case, I'm

5:11 going to clear the terminal and type in

5:13 claw here, and now you can see we have

5:15 our claw code. So, what we're going to

5:16 do here is I'm going to just try to give

5:18 it a prompt, and let's try to see how we

5:20 can be able to clone it. Now, the

5:22 application we're trying to clone here

5:23 is called Whisper Flow. Now, it's a

5:25 voice-to-text AI applications, and it

5:27 lives on my desktop app. So,

5:29 essentially, you can see that this icon

5:30 right here, if I click on it or if I do

5:32 a hotkey, it's just going to

5:34 automatically try to translate, right?

5:36 Try to translate any speech that I say

5:38 into proper text. And it's also able to

5:40 translate meeting notes, right? So,

5:42 those are really helpful. So, what I

5:44 want to do here is I want to kind of

5:45 like using that skill, and see if we can

5:48 be able to actually reverse engineer

5:49 applications on my local machine. And

5:51 right here you can see I have prepared a

5:53 prompt. It's called find Whisper flow on

5:55 my entire desktop app, right desktop

5:57 folder. Use RE here to basically inspect

5:59 the app and use the official website as

6:01 the background. And also explain what

6:03 you can be approved from the application

6:04 files, right things like that. And

6:06 here's to get a small working version

6:08 here working. So, I'm going to basically

6:10 open my Claude here and paste the

6:12 prompts, and just send it over, and let

6:15 Claude here do do the rest. So, in this

6:17 case, I'm going to let it go, and let's

6:18 see what it does. All right, so now you

6:20 can see that once we have our

6:21 application is fully complete here, if I

6:23 were to open the app, here is what it

6:25 looks like. So, we have the exact clone

6:27 of Whisper flow right here in my desktop

6:29 app. So, you can see that currently we

6:31 have all the conversation, everything

6:33 that I said is inside of this. I can

6:35 also click on note taker here to see the

6:37 past notes that I have took, right? So,

6:39 I can view the entire transcript or

6:40 resume if I want to.

6:42 And I can also be able to start to using

6:45 a shortcut, right a short key. So, I'm

6:47 going to type in the short key right

6:48 here, which is option plus one.

6:50 So, I want to start a new application.

6:55 And you can see that we have the text

6:56 here showing, okay? And I can say

6:59 different languages, like uh "Jintian

7:01 shi Zhouliu."

7:03 So, you can see that we have English and

7:05 Chinese here also in the same uh input,

7:08 right? So, we can be able to also have

7:10 the ability here do that. And also it's

7:11 really, really fast. So, it's not like,

7:13 you know, it takes a long time here to

7:14 load. It's transcribed locally on my

7:16 machine, so the conversation or the

7:19 voice here does not go on the internet,

7:20 which you can see here, okay? So, as you

7:22 can see that it's actually able to using

7:24 reverse engineering tool here to analyze

7:26 exactly how it works. And you can see

7:28 that it's able to creating a similar

7:29 versions of this using our AI agents.

7:31 All right, so pretty much that's it for

7:32 this video on the preview for REA, what

7:35 it is, how it works, and how you can be

7:37 able to set it up. Now, the last part

7:38 what I want to talk about here is just

7:39 my take, my my two cents on this. I feel

7:41 like this is going to be really

7:42 revolutionary because now the

7:44 application here, the code, is no longer

7:46 the moat. The moat is actually the brand

7:49 and the actual users, right? The actual

7:51 data is the moat. And you can see that

7:53 this is a clear example. We have

7:55 something called Photo Graph, which is

7:57 going viral recently. And essentially

7:59 what this application do is cloning the

8:01 entire suite, right, of Adobe Premiere,

8:04 Adobe Photoshop, everything from Adobe

8:06 suites, and basically try to clone that

8:08 using entirely in Rust. And you know

8:10 Rust is actually a lot faster than any

8:12 other languages, and it's able to do

8:13 this, right, one to one with a lot of

8:16 tools that we're currently using every

8:17 single day. And I think that's something

8:18 that you also have to consider. And one

8:20 more thing that you have to consider

8:21 here is building better marketing,

8:23 building better brands. Because if you

8:24 have a stronger brand, you that's

8:26 something that they cannot be able to

8:27 take away, right? So, building better

8:30 brands, marketing, building

8:31 applications, and how you can really

8:32 scale your business, these are things

8:34 that we talk about inside of our school

8:36 community for every tech head builders.

8:38 And of course, if you want to network

8:40 with other business owners and

8:41 professionals, this community here can

8:43 also help you do so. Now, if you're

8:45 interested for this, I'll make sure to

8:46 put the school community link in the

8:47 description below. And of course, if

8:49 you're looking to scale your marketing

8:51 for your business, you can check out

8:52 this video right here. I have reviewed

8:54 40 different skills for marketing. And

8:56 of course, if you're interested for more

8:58 content like this, make sure to

8:59 subscribe to this channel, like this

9:00 video, and I'll see you in the next

9:01 video.