Name: ChatDev
Uploaded: 2024-12-20T05:59:00.617Z
Duration: 2 min 33 s
Description: An open-source framework utilizing large language models to automate the entire software development process through multi-agent collaboration.

chatdev

An open-source framework utilizing large language models to automate the entire software development process through multi-agent collaboration.

ChatDev

This might be the most quirky AI coding assistant that I've come across. It's called ChatDev. It's by Open BMB and its purpose is to simulate an entire software company. Everyone from the CEO to designers to product managers and all of these AI agents work together to create a product. 

So I'm going to show you how to install it. Then I'm going to show you how to use it. Let's go. So this is their GitHub page. I just want to show this to you really quickly. Communicative agents for software development and really what sets this apart are two things. One, it's multiple AI agents working together that take on different roles within a software company. 

And second, it comes with this really cool pixelated user interface that actually illustrates the different members of the team working together. So it's really cool. It has a ton of functionality. It has 9,3000 stars already and it's very.

Popular it's on GitHub trending. So let me show you how to install it. So the first thing you're going to do is open up a new terminal. I'm using a Mac but you can use Windows. It should work just the same and I'll highlight when the commands are different for Windows. 

Now you should already have Anaconda installed. I use cond in every one of these projects. So if you're watching my video you probably have it installed. But if you don't go ahead and install it. Just Google how to install Anaconda. 

Now the last thing I'm going to mention is I'll drop a gist in the description below with all the commands I'm going to be using. So you can just copy paste from there. The first thing we need to do is clone the repository. So get clone and then the URL. 

And to grab the URL back to the GitHub page you click this green code button and then right here you click copy. Switching back to terminal.

We're going to hit enter and there it's downloading all the files and folders necessary. Done. Next we're going to create a new cond environment for this project and that's going to allow us to have all of our dependencies isolated in a really neat way. All right and it's done. 

The next thing we're going to do is grab this command right here: cond activate chatdev condore env and then just hit enter and we know it's active cuz it says so right there. Next we're going to CD change directory into chatDev, so CD chatDev, enter and now we're in that folder. 

Next we're going to install all of the requirements, so pip 3 install d r requirements.txt and then just hit enter. Now I've already installed it so that's why it says requirement already satisfied, but if this is your first time installing it, it'll actually show the installation in progress. The next thing we need is a chat.

GPT API key so switch over to open Ai. If you don't already have an account, go ahead and sign up. Go to the API key section and we're going to create a new secret key. I'm going to call this chat Dev 2 because I already have a chat dev. Then we're going to copy this and don't worry, I'm going to revoke this key before publishing the video. 

Since I'm on a Mac, I'm going to type export and then open AI API key with underscores equals and then as a string the API key. Now if you're on Windows, it's slightly different and I'll put that on the screen right now. But you're going to type EnV colon open AI API key allcor equals and then in a string again put the API key. Then we hit enter and that's it. 

So the next thing we're going to do is actually create our application. So on a Mac, we're going to type Python 3 run.py --task and then this is where you can write out a description.

Of the app that you want to create and you can write anything you can keep it as general as you want or as detailed as you want. 

And the example application that we're going to build today is a very simple to-do app. For this example, we're going to say a to-do app where users can add to-do items, check off to-do items to complete them, and delete and edit to-do items. 

And the name is going to be to-do. Then we just hit enter and here it goes. 

And now it's developing each of the team members based on the roles and what's happening is it's a bunch of AI agents working together. 

All right, it's finished. So it used 16,000 tokens and it gives you a bunch of other information as well. 

Within this Warehouse folder is where all of the projects are going to be located, the one that you just created as well as a bunch of other example applications. 

So double click into there and you can see all.

Of these different example applications we have another to-do list. We have a tic-tac-toe. We have MS paint password generator a bunch. But at the very top the one I just created is right here to-do default organizations. 

Okay but once it's done here's the part we can actually watch the replay and all the AI agents working with each other simulated. To do that we're going to type Python 3 online _ log slapp dopy and then we're going to hit enter. If you're on Windows you just type python not Python 3. 

Here we're going to grab this URL it spins up a server so we can actually watch it switching over to our browser. This is what we should see. So we click chat replay then we click file upload then we find this little log file right here and we click open. Then we click replay and there it is. 

Now we can actually watch watch it happen in real time and right here we can slow down the

Replay speed or speed it up and you can see here. Here's the CEO. Here's the chief product officer. The chief technology officer and so on. So you can actually watch them going back and forth working together. 

And another thing you can do is actually see the chat chain visualizer. So click on that. Then we're going to click choose file. And in the ChatDev folder go to company config default. Then we go to chat chain config.js. And now we can see all of the different prompts and the rolls chain together, which is pretty cool. 

And the last thing I want to show you is just one of the examples that were created by this app. So let's look at the Flappy Bird app. So we open it up in Visual Studio code. We go to main.py and then we click the little play button in the top right of Visual Studio code. 

And here we could see Flappy Bird. This is a really hard version of it. Let's try that one more.

Time and so here it is. It's obviously a very simplistic version. Sometimes the pipes overlap with each other, so it's certainly far from perfect. But it's really good for just typing a prompt out and getting everything that you need. And that's it. 

So now you know how to install and use ChatDev. Let me know what you create with it. If you liked this video, please consider sharing AI Agent Store with colleagues. And I'll see you in the next one.

Note, that there may be slight differences in text and video. Watch video for the original content.

ChatDev Video