Master the GitHub Copilot app: Installation to Full App Deployment

Today I'm going to show you everything you need to know to get started with the GitHub copilot app to build software and an agent first approach. Now, before we jump in to a full follow along workshop where you're going to build and publish your very first app, step by step with me, let's take a little step back. Okay? And let's ask ourselves, what is GitHub copilot? Well, I think of it as my command central for all things agent development. No matter where I'm building my applications today with a single GitHub copilot subscription, I get access to a diverse set of models from Google, Anthropic, OpenAI, Microsoft, Xai, and so many more. You can even bring your own running locally on your machine or through an API key. And it is where you work inside your favorite Ides, Visual Studio, VSCode, JetBrains, Xcode, and more. It's in your terminal with GitHub copilot CLI. It's on the web on GitHub dot com, and it's even on your mobile device inside the GitHub mobile app. Now let's talk about something that we're going to use today, which brings it all together, which is the GitHub copilot app for windows, Mac, and Linux.

It enables you to plan, build, review and ship in harmony with our AI agents across multiple sessions and multiple projects, all in a single app. And it orchestrates them all together. It really helps you go from issue to merge in one single application, working seamlessly across other places where GitHub copilot lives and you work today. Now let's talk about some of the key capabilities that we're going to step through with our workshop today. First, it's just delegating anything, handing it a task to any of the coding agents. Pick your model, pick an issue, write a quick prompt, look at an existing PR and flight. Really no setup required. Besides just installing. The app has built in reviews so you can look at the code, review the diffs, go ahead and create the PRS, and you can use Agent Merge, which we will today to check CI failures, comments and keep my code up to date. And it all runs all autonomously. You can create automated workflows that run daily, weekly, monthly. It does issue triage, accessibility reviews, refining of issues and getting tons of work done on your schedule.

It has advanced capabilities of canvases, which are two way communication channels with agents. Draw and build anything and you can run it inside the app, which will build our own Kanban board. Today I've built things like release management and actions, monitor design guidelines, manager, and even a game that I just boot up while I'm working. And it's not just for code. It's fully extensible. You can bring things in like work, IQ, stripe, MCP servers, canvases, skills, and so much more. So let's get into it. Let's build an entire application from start to finish. And this is a full coding workshop. It's really, truly your first steps with the GitHub copilot app. We'll start with installation and you can follow along at home. I'm going to put up a URL here and you can find it in the description below. And it's going to guide you all the way through start to finish of everything we're going to do today. So let's get into it. Okay. First things first is we're going to be walking through the first steps with GitHub copilot, which is a hands on lab workshop that I've done in person and is available to anyone watching right now, or just happens to find it on the internet.

It takes us through building and using GitHub copilot in a few different services and building out a space quiz, doing different prompting, testing, and actually publishing the project to GitHub pages. You'll be able to pick through using the app, the CLI or VS code, and all three of them highlight different features, but take you from start to finish. And building out this project. And as you're going through, you'll have prerequisites. You'll have little check boxes that you can go. Now I'm going to follow this one to one, but I'm not going to really go back to it at all. So just go ahead and make sure you click on that link in the show notes and go through it. And I'll keep it up to date as GitHub copilot continues to evolve. So if the UI changes at all, I'll update the lab too. I'll also point out here, when you go to GitHub dot com under platform, you'll see GitHub copilot. Which brings you to this page, which is the main page for GitHub copilot telling you all about the app, the different IDE integrations, CLI and the cloud coding agents that are available to. So it guides you through everything.

Different developers and customers using it and some of the core features. I'll also point out here the specific plans. GitHub copilot is based on a usage based billing based on the different tokens, consumption of the different models that you're using for these AI requests. And I'm not going to go through the specifics of that, but we'll take a look at it in the lab specifically of how much AI credits we're using. Now, everything that we're going to be doing should be completely doable on the free tier, unless you're out of AI credits on the free tier. But on the free tier, you get code completion. So if you're typing in an editor, but you also get access to some AI credits that you can use with specific models. And we'll be able to do almost all of the lab today with that in the paid plans of Pro, Pro Plus or Max, you do get access to different models and the cloud agents and code review agents. And we are going to use one of those. You'll get to see it, but you don't have to use it specifically. Now, in addition to a different features and models that are available, you'll also notice that you get different amounts of AI credits.

So the Pro, as of at least recording right now on ten, ten, twenty twenty six is ten dollars and you get fifteen dollars of AI credits. The Pro Plus is thirty nine dollars, you get seventy dollars of AI credits, and the max is one hundred dollars per month, and you get two hundred dollars of AI credits. One AI credit is one penny. So one hundred AI credits, one dollar USD. So you'll definitely want to click Get Started. And that's going to activate your GitHub copilot account. That's going to be the very first thing you want to do. If you go over to your profile and specifically here under the menu, go to Copilot settings, you'll want to make sure that your GitHub copilot Free Pro, Pro plus or max is activated. And you can see here I am on the max plan on my specific account. We can also take a look here of my included usage. So here I have the max plan. So twenty thousand AI credits. And so far I've used twenty four hundred AI credits. So about ten percent of my usage, eleven percent of my usage.

And then I also have additional usage enabled. You don't have to enable this, but I've enabled a ten dollars overage. So that way it continues to work. And I'll get notified as I go to sort of get close to my AI credits. You can also scroll down and take a look at additional features. Most of these are going to be enabled by default. But of course, if you're at an enterprise or business account, you may have to turn these on or go to your IT admin to turn on some of these features, including different models. So definitely take a look here. You'll be able to browse through that. Now I mentioned up top here on the main copilot page. If you scroll to the top you'll see desktop app, which is going to take you to the main GitHub copilot app page. You'll be able to see a bunch of different features and then be able to download it. So here for windows, Mac or Linux. I'm here on windows ARM. So I'm going to go ahead and tap on that. And it's going to go ahead and download it here. So I'm going to hit save. And this will simply download it locally on my machine. It's a simple exe that has everything bundled up inside. All right here we go.

So the GitHub copilot setup and we're just simply going to walk through the installation. And I'm just going to use the defaults. It's just a few button clicks. And it really only takes a few seconds to install. It's actually really snappy. So once the main executable is extracted in the installer, you're good to go. Now, of course, your installation experience on Linux and on Mac will vary a little bit, so check out the documentation or link specifically for installing the GitHub copilot app on those specific operating systems. But since I primarily use windows, that's what I'm going to be doing and in today. But the experience is exactly the same across the different environments. No matter where you're coding, it's going to look exactly the same. All right. Looks like we've completed here. I'm going to go ahead and run it and create our desktop shortcut. And we'll be good to go. All right. So the very first thing we need to do is actually sign in to the GitHub copilot app with your GitHub account. So like I said, you need a GitHub account. I've already activated my GitHub copilot account.

Of course it will guide you through if you haven't here. And you can sign in with your GitHub account. Or if you are in an enterprise cloud, you'd sign in with that account. So I'm going to go ahead and tap on that. And now it's going to give me and copy automatically to my clipboard this code that's going to have me log in. So I'm going to go ahead and paste that in and hit continue. And then it's going to go ahead and authorize sort of different access to my account. I'm going to say okay. And then I should be good to go once it's fully authenticated here. All right. So the first thing here is it's going to ask me to connect my work. Now I don't have any specific folders on my machine. But one thing I will point out here is that I can choose the specific folder where I want items to go. So I like to personally put mine inside of like a GitHub folder, and that's where everything is going to be cloned to. Now, if I had local folders, I could go ahead and add them here so it would pre-populate my projects. Or I could simply click on a project here that I have from my GitHub profile. Now I'm not going to select anything and I'm just going to hit continue.

But I do like to hit that clone location, kind of like where I want my work to be. Um, so that's up to you. I'm going to hit continue. Okay. Here I am inside the copilot app. The first thing I want to point out here is that we are on the new page. Now, I don't have any projects, but I can start a new interactive chat directly right here on the left hand side. I'll be able to take a look at pull requests, issues, automations and customizations. And we'll go ahead and walk through those a little bit here in the future. And we'll also see any projects. And again, I don't have any, but I can always add existing chats, GitHub repositories. I can clone a repository or open a folder here inside of our main chat. This is where you're going to spend the majority of your time here. For example, I can go ahead and add contacts like files or folders or other projects. I have different modes such as interactive plan and autopilot mode. I'm going to be able to use all of these today's. So I'm going to walk through them when I specifically need to, but also point out here the tools permission. And this is when it sort of wants to interact with writing files or making different commands and, you know, execution.

It will automatically ask you if you always ask or it'll approve all, which will just run all the tools without asking. Here are the model selection and reasoning. So here there's a default model that can be selected. But here I can see that I have access to a bunch of different models based specifically on my account. So your models may vary, but I have a lot of them here. And of course these change over time. For this one, we're actually going to toggle on auto mode. Auto is one of my favorite because it will automatically pick the best model based on efficiency, balance, or intelligence. And often you get a discount here for picking auto. So we're actually going to use for the entire lab auto or balance if you want. And one thing that I do like to use often under more models is GPT six Luna. It's one of my favorite ones. And you can go ahead and select that. And that's a great model to use if you do pick a model, I'll also point out that you can control the effort. For example, I love Luna on extra high. It takes a little bit longer, but it does a really great job and you can control the context window.

Now, if this is a lot just like, wow, let me just get into it. Just go ahead and select auto balance. And that's what we're going to do for the rest here, we can see that I'm going to be starting a new chat. And here if I had other projects, I could go ahead and add them here and we'll see them here in a little bit. I'll also point out that I'm able to do voice dictation, which will use a local model to do voice dictation so I can speak to it. And then I can also just send the prompt down here. I have some different project ideas, and we're actually going to build one of these today, which is a space exploration quiz. But let's go ahead and ask it something. I'm just going to go ahead and hit this microphone microphone button here to do voice dictation. So let's go ahead and open the voice settings. And here what we'll be able to see is it brings me directly in where I can select my input device, my microphone permissions, I can test the microphone shortcuts and I can download a transcription model. I'm going to go ahead and download this Negatron speech streaming here to my local device. This is really, really nice because specifically all of the voice transcription is happening locally on my device just using my microphone.

Okay, so now that it's set up, let's go ahead and do this. How does the GitHub copilot app use work trees? And now all I have to do is send it off. Now, this is just a chat. No folder, no code, no repository. Just a chat. And here we can see if I hover over it, that this is using GPT six Luna. And it has picked the reasoning effort automatically for me. And the best model for this question. And it describes exactly how work trees inside the GitHub copilot app work. Now I can follow up here and I could say, can you give me a breakdown of how work trees themselves work? And then I can go ahead and fire that off. Now, while it is thinking I could continue to type and I could go ahead and send or hold down control and then actually go ahead and queue them up too. And we'll go ahead and do some of that today. All right. This is great. This actually just walks everything here through how exactly to use work trees.

So what's great about chats is that they enable you to get access to real time AI to ask it about anything. You could ask about issues about sort of source control. You could go a deep dive and have it do research. You could have it spin up anything. Triage, backlogs, a whole bunch of things without having to have a project or anything like that, just like an AI chat. Now over here, we can see that I can right click specifically on this chat and I can go in, I can rename it, pin it. I could go ahead and share it. I could copy the session ID or archive it. So when you're done, you can archive it and you can go in and manage those later. All right. Let's head back up to new. I'll point out a few other things here in the settings before we start to build our application. So if I click on settings here, this is where you're going to check for updates. And it'll automatically check for updates. You can always change the storage location. You can adjust your theme and tons of other things like start on system, start showing system tray, which is really nice actually, because when you do that, you can right click and see all of your different sessions that are here and also get a notification.

And then you can see, you know, delete chats without confirmation and different sounds that will ping basically when things are done. If I go under accounts, we can go ahead and take a look that I can see my usage that twelve percent, and I can add additional accounts and toggle between them. For all the sessions we're going to be creating. You can manage sessions, you can see archive sessions and clean them up. You can have applied instructions. You can have different tool permission settings. You can basically customize every single piece here, including different branch names, work tree locations, if you want them, global insider repository or next to a repository. If you want to keep your machine awake. If you want to show GitHub copilot CLI sessions, um, how often you want to do session cleanup and so much more. We also have the themes. So I talked about here. You can of course go into dark. I have it here just under system. You can go ahead and pick any of the different ones. There's tons of themes built in automatically. So if you want different monochrome fox, I'm just going to use the default GitHub one here.

Tons of accessibility features built in. There is computer use which is in preview, which enables the agents to click around and use your computer just like you would. And then we'll go ahead and get into customize a little bit later, which is actually in the main toggle here on the left side. But I also want to talk about model providers here. I'm using obviously the GitHub copilot model provider, because I'm logged in with my GitHub account, with my GitHub copilot subscription. But I also want to point out here that I can add custom endpoints. So I can use things like any open AI compatible endpoint, open router. I could use an anthropic API key foundry. I could run things locally with foundry local or Olama, or connect to an OpenAI directly. If I have a subscription there, I can use my API key. And of course there's tons of experimental things too. So definitely check this out if you want to be on the leading edge. Okay, awesome. A lot packed in. Now let's go ahead and start building our application. And we're going to build a space quiz. So the first thing I'm going to do here is I'm going to go and click on this chat, because we're not just going to have a chat. We're going to work on a project and I'm going to say open folder.

Now here it opens my localhost or my local disk GitHub. And I'm going to say new folder. I'm going to call it space quiz. And I'm going to go ahead and select it. So now all of the work is going to be done. In this project. You can have as many projects as you want on the left hand side. So this is my space quiz right here. And we can see that really nothing is happening yet. It's just a new session now. It's brought over my settings of main interactive mode. And of course it still has my auto, uh, for the model selection set to balanced. Okay, so the first thing that we're going to want to do is start to build our application. So I'm going to copy the prompt here from the lab, and I'm going to ask it to create a space exploration quiz with ten questions. A progress bar score counter. A colorful animated feedback and different emojis. I want to center it in a narrow column and I want it to be just an HTML page. No server dependencies. Use nice fonts, different text body and different color schemes. And then I'm going to ask it specifically to open the integrated browser.

And the integrated browser is sort of this canvas that can interact with the agent back and forth. And we'll take a look here. So I'm going to go ahead and send this off. Now again auto is going to go ahead and select that model for me automatically of what it believes is the best model for this specific prompt. So the first thing it's going to start to do is start to view and understand what's in this folder. And in fact, there's nothing in it, but it is going to take a look for different canvas capabilities and understand what its capabilities are. And then it's going to start to think. So you can actually expand all of these and understand exactly what the model is thinking and what it's going to be doing along the way. Now, this might take anywhere from 30s a minute 90s, but we're going to go ahead and let it cook here, and I'll speed this up a little bit, but we'll see along exactly how it takes to create our space exploration quiz. Okay, beautiful. So it looks like it's created our application.

And I'm going to go ahead and minimize this on the left hand side since we're just working this one session. The first thing I want to say is I have hands off keyboard. What's great here is that the agent is taking a look at this interactive browser that is available to the agent. Since I asked it to open it up, it went ahead and opened it. And then it took a screenshot and it evaluated to make sure that the application was running and that we see the questions and the progress bar right there. If we browse up, we can see that it created that HTML page, and I could actually expand all of it to see it there. Okay, so let's go ahead and take a look at our space quiz. So I'm going to go ahead and maximize this panel. So I hit this little button here. And we can see that I'm able to toggle between light mode and dark mode. It looks like it did implement it. We can see that it's a file right here in the space quiz folder. And I can go ahead and scroll down here. So I like this orbit. How far can you explore who is the first human to travel into space? First human I think.

Buzz. Oh, no. Yuri. Okay. Wow. That's my totally bad. There. That's good. We can see some of the things here. We can go to the next question. It's Apollo thirteen, Apollo eleven. I'm going to be bad at space quiz. That's great. That's definitely Mars. Perfect. Okay, cool. I'm not going to go through all of them. But we can see here, which is really good. Okay, cool. So this is really great. I really like this. So, so many points I have. That's pretty fantastic. Okay, so now that we have our application sort of up and running, we can actually customize it a lot more. So one of the cool things about the integrated browser is that you can pick and polish. So this allows you to not just only take screenshots. I can shift click, but I can also pick and polish anything here. So I can tap on this and I can check any of the different elements here. So if I wanted to customize the title or the description here, I can go ahead and click on this. So let's say this area here, it will automatically send that selection there.

And again I could also send it a screenshot here. Why don't we just say and let me go ahead and talk to it. Make the selected element here feel a little bit more like mission control display, keep it accessible and preserve the existing light and dark themes. So this allows you to really work with the model and customize the user interface, either that you are dreaming up or that it's dreaming up together with you. So now it's going to go ahead and view that HTML and it does have the session in its context. So it knows and it understands what we have here. And now what it's going to do is update this in real time. And we can see the diffs that it's going and doing for us here. And we can see the HTML changes. So if I actually expand that here, we'll see that it's changing the CSS and updating it here too. This is pretty cool. It looks like it's automatically invoking navigating around. And we have sort of a newer, bigger area here, which looks really nice. And I could continue with it. I could go ahead and specifically ask it, give it like a cool hex background.

That's a little bit spacey. And again, because it kind of knows exactly what I'm doing, I'm just following up on my prompts, but I can check here to make sure that it did. Yep. Light and dark mode. That looks very nice. So here's going to add a subtle hex grid here. And the more specific I can be, of course, the better it's going to do. When I gave it sort of a generic make it get mission control. What does that mean? But now it has a cool hex background that's really, really neat. And it's investigating along the way and evaluating JavaScript and taking a look at those images, which is really nice. So a lot of things have happened in a very short amount of time. If I hover over here, we can see that we're using GPT six point one. So in my case, you might be using a different model here, but I'll point out that there's a lot of information that is here, but not just visible right away without clicking around a little bit. The first thing that I'll point out is that you can click on the title, and you can get information either about the current branch and the changes that are happening or the folder that you're in.

So here, for example, I can see the path. I can copy the path, show it in the explorer. I could go ahead and I'll get to remote control later to expose this out. To do a remote session, rename the session, copy the ID, or even share it as a secret guest. And that's really nice when you're looking to diagnose it or archive it or delete it when you're done. In fact, if you fly this out, almost all of the same items are in the right click menu of that specific session. I'll also point out here. Down here we have that interactive mode and our models, but I can also go ahead and click specifically on the AI usage. And here this is going to show me my total plan usage which is still pretty similar that twelve percent. I can see my session AI credit usage which right now is twenty four point nine. So about twenty five cents or a quarter. And then we can see that on my plan, there is additional usage enabled. And more importantly these two items here, which is tokens. So how many tokens were output or sent to the model and how many were cached.

And then how many were received back and how many were used for reasoning. I can also see my context, and I can go ahead and click on that and see where the context is today of my four hundred K token. Here's for the context. So I have system prompts, tools, MCP messages, lots of free space allocation and a buffer. So it's really nice to get all that information. I'll also point out here that the this area here on the right hand side that you can toggle in and out right here is this interactive panel that has many different panels available. So this opened the browser panel, but there's a terminal panel, a files panel, a side chat. So you can have chats while your main session is going insights and canvases, which we'll talk about in a little bit. If I tap on insights and expand that out, we can see exactly what the agent was doing. Was it reading, thinking? And I can see exactly the calls that it was making. Take a look at all of the different tools, specifically of what it was doing. You could dive in deeper too, which is really, really nice.

I could, of course open the files, which just has this one here. So if I go ahead and tap on files again, we can see that I have my HTML page, and I can click on this button to open it directly in the browser. And boom, there I am. So you have access to browse everything here. You could open it in the browser, copy the path, show an explorer, do anything like that, which is really, really cool. So a lot of information at your fingertips, which I absolutely love. Okay, so I have this running and it's been opening the browser and it has specifically been looking to see if the application is running, but it hasn't really tested the application. So I'm going to have it test the application now. So I'm going to go ahead and talk to it. And again this is in the prompt. Run a browser level smoke test for the quiz in the integrated browser here. Check for keyboard navigation score updates correct and incorrect feedback and the result screen. Fix any failures or problems here based on the design that we wanted, and then report what is past and what is failed.

So this is really cool. Again, hands off the keyboard. Hands off the mouse. I'm asking the agent to go ahead and access this. So in this case, it's going to go ahead and launch it. And we can see, because I opened it, it wasn't shared specifically. So it opened a new instance where it was shared. So it access. So it's doing it on my behalf. And now it's going to go ahead and start testing the application to make sure that one all different parts of the application are working, but also that it is accessible. And also if there's anything it needs to do, it'll fix it up. So it looks like here. Yeah, it's fixing up some different, um, headers. It looks like here. So that's really, really nice to see. And we have some low contrast light theme text. And I've asked that specifically to do that pass. So I actually love just kind of sitting back having it testing and dogfooding the application. And more importantly, as we add features, we can ask it to go deeper and go ahead and test those features too.

So let's just go ahead and let it cook here. Fantastic. So it gave me a full report. I love this so smoke test has passed for the quiz, a keyboard keystrokes remain unverified, so I wasn't able to access that. But it did do a full check here of everything. So all the results Focusable answers. And it did three hundred and seventy six flow assertions along the way. That's awesome. Now we'll point out here anything that we may have to do manually, such as screen reader, but that's really, really cool. I also want to point here that you can go ahead and hover, and you could actually quote text this, or you could add it to a side chat or copy copy this text to and I love it. I'll also point out here that you can actually go back in time and just scroll up to the different prompts that we were sending. I love these little bars here. So, so nice. All right. Now at this point, we have a full application, a pretty gorgeous design, to be honest with you. And our first part of our space quiz. Now we could continue to add features, but as I continue to evolve this application over time, we're going to want to make sure that our agent or any new agent or anyone else that's working has a base set of rules for the application.

So here inside of the copilot app, I can use a slash command. These are tons of built in commands. And if you've used the GitHub copilot CLI, or if you do it in the GitHub copilot CLI, you'll see tons of different commands that allow you to do different actions. Many of these are available in the UI, but it's nice to have a nice little slash to get to it. So here I can ask it to review changes for the current session. Restart the session rename check my permissions. I can go into debug mode. I can open the terminal. I can go ahead and ask for a security review. I can do by the way, which is like a side chat. Inside the chat I can look for Chronicles, which is viewing the session history and so much more. There's tons of different slash commands built in. The one I'm going to use is init. And specifically we are going to ask it to initialize our project with a simple set of rules. So I'm going to say init. Create a simple set of rules for the project.

Now this init is actually a skill or a specific prompt that will be used to understand what I'm saying. So it's going to say initialize instructions and it's going to go off and use this skill to understand what that means. And what this is going to do is analyze our project. Probably take a look at our coding standards, take a look at our color themes and anything that we've asked in the chat, like accessibility. And it's going to go ahead and create a concise document that is going to go ahead and understand when we make new sessions, to go ahead and use these as the default every time. So every new session will include these here. Now I am in interactive mode. So what we're seeing here is that it is asking me questions about if it should add anything else. So it's asking me if I would like to adjust the instructions at all. So let me go ahead and open that folder view here. And I'm going to look in the GitHub folder and copilot instructions.

So it went ahead and created this for us. And I'll also point out that while these are GitHub copilot instructions, GitHub copilot can also use agents dot MD file, which is a kind of a standard for agent instructions and rules. But the copilot instructions allow a little bit more customization when using with copilot. And here we can see exactly how to run the application, how to do validation on it, the main architecture, any UI and accessibility conventions that we have so far. So this is really cool, right? So it says keep progress with Aria value now and value text and how to update these along the way. So I could decide is this too verbose? I want it to be more concrete. And it's going to ask me, hey, do I want to like add playwright, for example, and make sure that it knows how to do that? Do I need to adjust the instructions? I'm just going to keep them written as is, and this interactive mode will prompt me when needed. So really, really nice here to actually just get that up and running. Okay, so we have our project, but as I want to collaborate and I want to publish it, I'm going to need to put this up on GitHub.

So there's a few ways of doing that. Of course, I could go to GitHub, I could initialize it, I could push it, I could do a bunch of things, or I could just go ahead and ask the GitHub copilot app to do it. So I'm going to go ahead and copy in a prompt here that's from the lab, and it's going to ask it first to initialize the folder as a git repository, which I could do myself manually. But I always forget those commands. I'm asking it to create initial commit and create a new public repository name space quiz, and push the current branch and set it as the default branch and then refresh this project. I'm actually going to modify this and say also create a Readme because why not? And I'm going to go ahead and send this off. Now, this is really great. And I want to point out here that the tools permissions I have on by default here are approval. So it may prompt you and it has maybe been prompting you along the way. If always ask is the default. So up to you on your sense of control that you want to have. If you want to go ahead and add those there. So it's going to tell me exactly what it's going to do, which is add that Readme going to commit the current files to main and publish it to my account as space quiz as a public repository.

And then it's going to go ahead and make sure everything is linked up here. Now again, I can go through and see exactly the git init commands it was running. I could go ahead and see everything else that's going to do. And now, because I've initialized it as a repository, the copilot app is now saying, hey, listen, this directory change, do you trust this repository including the agent instructions and everything like that? And I'm going gonna say, yes, I trust this repository. So just as if I was adding it myself. Now that we've initialized it, we're good to go. Cool. So we have our space exploration, and we can see now that I'm on main here and I'm totally good to go. If I tap on top, I can see that I have a branch which is just main, and that was added here, which wasn't there before because it was just a folder. If I go and tap on this public repository, this is going to bring me into my GitHub account and boom, there it is, my Readme, everything like that along the way. So it tells me everything that I have right there. Okay, so so far we've just been an auto.

It seems to be using the same model here that we have. And at this point we can go a little bit further. Now, I could probably think of, you know, fifteen or sixteen different features that I might want to add to this app, but I'm going to go ahead and see if GitHub copilot can come up with some new ideas for me here. So this is another prompt. I'm going to go ahead and copy and paste in from the lab. And here I'm just going to ask it to review the space quiz. And I'm going to ask, actually ask it to suggest five focus feature ideas that could be implemented in a short session. Now what's really cool here is because it's deeply integrated into GitHub. Obviously, we just saw it create the GitHub repo for me. I'm going to have it create GitHub issues for this repository with clear titles, sort of user focused description and acceptance criteria. And of course I'm going to ask it. Don't implement it yet. So I'm going to go ahead and send this off. And it's going to understand what my application is and come up with some new ideas for me to implement.

Now, if you have an idea, you can also just go and say, hey, I have this idea, create the issue for me. So that's really nice too. So let's go ahead and see what it does and comes up with. Okay, perfect. It looks like we have five different issues that were created, and I can go ahead and tap on them and they'll open right here. So the first one is add a post quiz, answer review. That's pretty cool. Offer a shuffled replay, show the current and best corrected answer streak. That's kind of cool. Add discoverable number key answer shortcuts. So support number key one, two three, four. That's really nice and accessible at a Copyable mission results summary. That'd be pretty cool too. And it gives me a full acceptance criteria here. So that's super duper nice now because I'm going to be starting some new work here. I'm going to actually work on these issues like I would normally do on a given day. I'm going to sort of start a new piece of work or a new session.

And the first place that I'm going to go is over into my issues. And here this is going to show me all of the issues that are currently assigned to me, created by me mentioning me, or they're already complete. I can go ahead and create a new view, and I could go ahead and sort these by open and close, and I could add any filter that I may want in here too, which is really, really nice. Tons of filters that are available. I could go ahead and save that. I'm going to go ahead over to the created by me. And we can also see that I can create the new issue that allows me to create the new issue here. And if I had multiple projects, it would ask me what project I want to be working on, or show me all of my projects inside of my app. That's actually something that's really, really important. It's showing you what's inside of my application today and the specific projects that are inside the application. So if I had ten here, it would show me any of the issues from those ten projects, not all of my issues on GitHub, for example. And I absolutely love that. Okay, let's do this first one, which is a post quiz answer review.

So it says after finishing a mission, explore should be able to revisit and see what they got right and wrong. So from here I could go ahead and start to implement it. I could come over into a new session here just like this, and use the pound sign or hash tag to pull up my issues. But from the issues page I can just simply tap New session. This will automatically link that specific issue with a title and create a new session for me automatically. So from here I can use this issue fix that is built in as another prompt, and then I can give it additional fixing instructions. So for example, here I'll go ahead and send our prompt over that we had, which is to go ahead and implement the issue completely, keep the single file dependency, test the behavior in the integrated browser and summarize the changes when finished. Now here we can see that I have multiple sessions open, so I could go back to the other one and do some work.

And I'm still using my default interactive mode and I'm using my default auto two. But I will say that this has this attached issue. And I of course could attach issue here or attach different projects. And now I'm going to go ahead and send this off. And this is now going to go ahead and get to work on this issue and report back to me. Now, while this is going and working, let's go over to the other issues, and I'm going to go over to the second one, which is show the current and best correct answer streak. So this has a small streak readout, gives explorers an extra sense of momentum. That's pretty cool and track consecutive answers. So I'm going to go ahead and create a new session for this one. But instead of just fixing this issue, we're going to go ahead and go into a plan mode. So this enables me to work on multiple things all at once. And I'll point out here is when I create a new session, a work tree was created for me automatically, and a work tree enables me to work on multiple things on a single repository at once without conflicts.

It kind of creates a shallow copy of my repository on my machine. And I'll take a look at that here in a second. So what I'm going to do here is minimize this down. And we're going to change from interactive to plan. And plan mode is not going to implement anything yet, but it's going to ask me questions along the way. If it's not clearly defined and have me agree on the plan before implementation. So I'm going to go here and instead of saying issue fix, I'm going to say plan how to implement the issue. Investigating the existing quiz list of files you might want to change. Call out risks and stop before making any edits. Just to be one hundred percent sure. I'm going to go ahead and send that out. Now here I can go up and see that this is a branch from main to this specific Jamont show current best answer that generic name that it gave. I can go ahead and show this in the explorer. And what I'll point out here is that under my GitHub? I've told it to put all of my different work trees into the copilot work trees here.

And here's all of the different work trees that I have been doing for this specific project so far. Okay, so it looks like it has some questions for me. So here it's asking me from the results screen should the best streak appear as a third item in the existing stats panel or as a separate line below? It even gave me a recommendation as the third stat item. So let's go with that. So this is really really nice. So it is going to follow up ask me questions. But it has the context of all the issues here. So I love that. And then what it's going to do is come up with a plan and it's going to make sure that I'm able to review it before starting work. So I can tap on this plan and open it up. Or actually it opened it up for me automatically. And we can see show current and best corrected answer streak. So here, here's the problem. The files to change the notes in any risks that I might have. And it's going to say, do you want to approve and implement this plan? Exit. And I'll just prompt myself so I could follow up.

I could save this for later or I could suggest changes. We're going to go ahead and approve and implement this plan. And the first thing that we're going to see is that it's changed automatically to autopilot, which is end to end execution without interruption. So what's interesting about that compared to interactive is that autopilot is just going to go ahead, make the tool calls, and then what it's going to do is go ahead and specifically answer any questions automatically for me on my behalf. I'll also point out that specifically inside of the autopilot mode, it's sort of like a goal. It'll continue to work specifically on it until it's done. So it's going to execute, explore and more. I got a little ding here and it looks like the add post quiz answer review is complete. So let's go ahead and take a look here. So let's go ahead and open this up. and I'm going to go ahead and click around. Now that I know that it is Yuri Apollo eleven Mars. And I'm just going to click around here. I'm going to get some right. I'm going to get some wrong.

I'm not even looking. Oh, Venus. No Mars. Uh, this one perfect telescope. Hubble. Oh. James Webb okay, perfect. I know these I'm learning along the way. Perfect. Nope. Valentina. Wow. My space, you know, isn't maybe the best. I'll say that. So I'm just clicking around. Okay, perfect. We get a nice little review here of my answers, and boom, I can see everything here. Got it correct. Got it wrong. And a little bit of information. Wow, that's really nice. I love it. Super duper cool. So we've just added that here in seconds. Now if I hover over this prompt here is using GPT six Luna compared to my main one that we had earlier, which was using GPT six point one sol. So so far, this one here, if I go ahead and pull this up, looks like it's about forty seven credits, so about fifty cents to add this feature here. If I go ahead and toggle this here, this was one point four eight credits. So one and a half pennies. And then here on this one, which is looks like my commander explorer rank and best streak.

Also use GPT six Luna for this. And that one here was one point five eight credits. So I'm implementing features in near record time, which is super awesome. Okay, let's take a second here and pause. So the first thing that we did is we went off, created our base project, put it on GitHub, and initialized git with a full Readme, including our instructions. If I go ahead and open these up here, we can go ahead and see them all. I'll also point out that over on my issues that we've created, they're all here directly on my issues. We went ahead and started different sessions and both of them have been implemented, one specifically just with the simple fix command, and the second one with a full plan that was executed first and integrated here. And we can see those check box along the way. So that's super duper nice. Okay, so now at this point we have three different sessions up and running.

Now we want to go ahead and merge these pull requests or well, actually we don't want to just merge pull requests. We want to create pull requests and then merge them in. Now we've been working in different branches or different work trees, so we're not working on main. So we're going to need to create the PR. So the first thing I'm going to do here is go up to this big green button. And I can go ahead and create a PR, create a draft PR or agent merge, which we'll talk about here in a little bit. I'm just going to create the default PR. This is going to go ahead and invoke another skill or prompt that is going to go off and understand everything that it takes to go to create a pull request on this repository, and it is going to look for pull request templates. So if you're using these in your projects, it will first make sure and understand what a valid pull request looks like for your project, which is really nice. It's then going to go ahead and add the file changes and then push any file changes up and then create that PR. I can then click on it and get detailed information about what it was doing, and it's automatically linked to that issue for me.

So here we can see that it's issue number two. And it's the same one that's linked here our plan and our changes. So really, really nice. Now I don't have any checks or any CI set up here. Maybe I might want to add linting for example here. But right now it just says hey it's totally ready to merge. And if I click on it, we can see that it's ready to merge. Now I'm going to go ahead and sort of simulate something that I would do in real life, which is if I go ahead and click on this, it's going to open the pull request here and I can see all the exact same information directly on GitHub dot com. There's my issues and my pull requests. Now I'm going to go ahead and assign it to a code review agent. And I'm going to request that from copilot. And this is going to look at these changes and give me a recommendation if there's changes that are needed. And specifically if there's any other recommendations that it may want to implement. This is going to spin up a new copilot session in the cloud automatically. Now while that's running, I'm going to go over to our other session. I'm going to create a PR for that one.

And here it's going to run that same exact open pull request command. And in this case it's going to go ahead and see again if we have any pull request templates or it's going to understand what else it needs to do, push the feature branch up, and then it's going to go ahead and create that pull request for us. And additionally, it's going to not only just create that pull request, but it's going to go ahead and link it again to that first issue because we've linked it here. So having that link is really, really nice. So this is going to walk through everything. So now in this case what we're going to do is we're going to see what happens when multiple people or myself start to work on features and we get some merge conflicts or when other people comment back for changes. So on this one, I'm going to go ahead and click on the ready to merge. And I'm just going to merge it in. I know it's super duper good to go. And what I'll point out here is that it's automatically merge that PR for us.

And it is here on the green says merged. And then additionally, if I open that issue up, we can see that it is now marked as closed for us again, because they're all linked together in this system. Now I'm going to go back over here to the other pull request. And the first thing we're going to notice is that there are conflicts that have arisen here. And if I tap on that, we can see that the merge is blocked. And that's because there's a merge conflict, because we just pushed some changes from the other session that was going on. Now, if I wanted to get deeper and I want to resolve the conflicts myself, or even just look at the changes, I'll point out here that on any session, you can open it directly in your favorite code editor or in the CLI. So if I click on the VS code button, it's going to go ahead and ask me, do I have permission to open it up? I'm going to say yes. We're going to zoom in a little bit here. And what I want to point out is that inside of VS code, let me go ahead and trust this folder. It is automatically opened the specific work tree for me.

So down here it says Jamont show current and best answers. That's that work tree. So it's not the main folder, but it's the work tree folder. I can then go ahead and I could have taken a look at those specific, you know, source control changes here, and I could go ahead and tap on this and see the changes specifically that took place right here. And this item or that item. Wherever I'm at, I could commit changes from here. I could merge directly into it. I could also go into that session. And this session specifically is the same exact session that was from the GitHub copilot app. They're all running the same exact agent harness here. So I could follow up with it here. Or I could go ahead back into the app and I can navigate back and forth. So it's really nice that everything really is in sync with my same exact GitHub copilot subscription. Of course, I need to be logged in here inside of VS code, which I am, but let's go ahead and take a look and see what else has happened here. So the first thing is that we're going to see, well it looks like our code review is good to go. So that's great. So no issues there.

So now I can go ahead and tap on this conflicts. And I can ask it to resolve this conflict. Or what I personally like to do is toggle on agent merge and what agent merge will do. And we can see here in the settings is automatically have the agent address any review feedback. So if the copilot code review did have suggested changes, it will go ahead and implement those. It will automatically fix any CI failures and it will also automatically resolve conflicts for us. Now there is a fourth option which specifically here is to merge pull requests automatically when everything else is green. Now if you've been noticing, we can see that it actually ran the agent merge check in, it loaded the agent merge, and it's gone ahead and it's updating this pull request to figure out all of the different changes that need to be made to resolve that merge conflict. So not only is it merging, but we can see that in addition to that, it's invoking some canvas actions.

And we can see that there's even a conflict. So it's resolving things that have been happening here in real time. So now it's going to take a look at the pull request comments. And it's going to understand what it needs to integrate here to make sure that our feature and functionality is correct. And again we can tap on those changes. And more importantly we can again open it directly in VS code, which we have here. And we can follow along to see if there are additional changes that are being made. Anything else that's being merged in and we can follow along there. So that's really important. I could also come over to this new and see the changes here. And I can see the committed uncommitted last commit anything like that that's coming in in real time. You have this fine grained control right here. Okay. It looks like it resolved the merge conflict, retaining both the streak feature and the answer review feature. That's super nice. And now it's mergeable. So we have all this ready to go. So let's go ahead and merge it in boom. Just like that in mere minutes we've integrated new features, resolve merge conflicts and have the agent handle it for us all automatically, which is super duper awesome.

I absolutely love this Agent Merge feature, and having you be able to subscribe to these events on many of my projects, I of course have CI being run automatically across Linux, Mac, and windows on many of my projects. If any of those fail or a test fails, it will automatically pick that up and resolve it. It's the same thing. I could be working on ten to fifteen different features and bug fixes at the same time, and the agent merge will subscribe and make sure that my PRS are green at all times automatically for me. Super duper nice. Okay, so so far we've taken a look at our issues. Now, I also want to point out here that you can also go into pull requests. So here I can go ahead and see my done pull requests, which are those two. And just like issues pull requests, you can see if anything is assigned to you, involves you or requires a review at any time. You can click on one of those pull S, and you could actually review it and actually go ahead and leave comments or approve it. And you could also create a new session that's super nice.

If you've started on another machine or in the cloud or from your phone, and then you come back over and you need to open up that pull request, just simply create a new session from that PR and it'll automatically be toggled on. So that's really nice to have right there. So we've taken a look at new, we've taken a look at pull requests. We've taken a look at issues. Let's take a look now at an awesome feature that's built into the copilot app, which is automations. Automations enable you to handle recurring work on a cadence that you choose or things that are triggered by events. What's great here is that it gives you a bunch of templates like issue triage, a change log draft, a repo audit, telemetry review, performance improvements, or accessibility audit. So for example, here I can go ahead and tap on this issue triage. And I could then say, yeah, issue triage for space quiz. And then I can say win. And I'm going to say weekly, for example, and I'll say every Monday, that sounds good.

I can then select a project and then I can decide how I want it to run. I like my automations to run in autopilot. And I, you almost always select auto here for that prompt. I can say give me a grid breakdown of all remaining and also all closed issues in the last week with a summary. And now we can create this. And this will be run every week. Or I can create and run it right now. And here is our automation running. And it has a brand new session. We can see that they can be run locally or also in the cloud for private repositories. And I can search my automations. If you have a lot, you can always get to those templates, and you can come back here and just create a new automation. So if I wanted to trigger one here for an accessibility audit every week, I can do it again on the space quiz and it will go ahead and review. The PRS merged this last week and summarize any accessibility issues.

You can have it create issues for you and can do much more. So again I'm going to say create and run. Here. We can see that the issue triage has finished and it looks like it reviewed all of them and also closed within the last seven days. Those two specific items. And it says here I was accountable owner for both of these specific issues. And you can be much more verbose with your different automations. I have quite big ones there too. Let's take a look at the accessibility audit. And we can see here it's now going to go look at those two pairs that were made in the last seven days. And this is handled all automatically for me. So when I walk into the office on Monday, I have many of my automations boot up right around six or seven a m so when I'm in the office by seven thirty or eight with my cup of coffee, they've all run and I can review all of their outputs. So it's really, really nice to see. And here we can see it's all running in autopilot mode, and it's actually taking a look and reviewing the different capabilities and making sure it's good to go.

All right. So now from here I'm going to start work on one more issue while my automations run. And we can see them running right here. I'm going to say offer shuffled replay for a fresh mission. I'm going to start a new session. Now from here I'm just going to go ahead and say issue fix. After that work tree has been created and fire it off. Now at this point I may have to step away, but I don't want to lose the work that's being done locally here. So what I can do is come up to the top and enable remote control. This is going to enable my session to be controlled remotely. It's running on my machine, but completely remotely. I can click open with a QR code or simply open it directly inside the browser. It will also be visible from the GitHub mobile application too. And here we can see all of my sessions and this one specifically here that I can interact with. So for example, if I scroll up to the top, we can see here's the issue, fix the name of it, how many AI credits, and then I can go ahead and see that this is exactly where I enabled remote control and see all of this here. Now if you try to enter this URL, you won't be able to see it, but from here I'll be able to access it because I have the authentication and authorization to interact with the session running on my machine.

So I can also go ahead and change the mode. And of course I can stop it if things aren't to my liking. And I can also follow up. I can say when done, please Give me a full overview of the changes in a table grid. Per line. Just to really get verbose here. And then go ahead and send this off. And then we can go ahead back into the copilot app. And we can see that that message is pending here as it continues to work. So you can stack up additional items. So, um, do a quick review and then I can go ahead and queue that up or steer it. And that's going to add it here to the specific one if I want. And you can queue up as many messages as you want. So this is really cool. Look at this. I've just asked that for a full review, and I can go ahead and get this on any other browser on my phone or anywhere else right there, which is really, really nice.

Now, this is not delegation, which for example, would be delegated to a cloud agent. So for example, if I go over to this project and say agents, I can go and create a new session directly in the cloud. I can see all my other sessions here. But this session in the cloud specifically would be running on a remote machine in the cloud for me automatically. And I have my different models that are available, just like I have inside of the GitHub copilot app. So that is nice. And a pull request would be open on my behalf if I wanted to start implementing work there. And it's very similar as going into an issue, for example, and then assigning it to an agent. So when I assign it to an agent, it will go and ask me what agent I want to assign it to, and it will create an agent session for me. So I have that features and capabilities to do it on my phone, on the web, in the copilot app, or anywhere else, or GitHub copilot is available and decide, do I want them running sort of in this remote environment, or do I want them running locally on my machine, running locally on my machine?

Obviously gives me the advantage of being able to use my local tools that are available. So if you're doing Node.js development dot net development or Maui development or anything like that, you'll have your local environment available to you, which is super duper cool. All right. So so far, what have we done? Well, quite a bit. We've gone ahead and implemented several features. So let me go ahead and create the PR for this. And so far we've been working with the issues tab here and the pull request tab here to take a look at all the different issues. I want to point out here that we've really just been scratching the surface of the capabilities, because there's a vast amount of customizations that are built in. There's a base set of tools that are built in the copilot app, like we're seeing with the browser, for example, or the vitals, but also you can extend the GitHub copilot app. Here we can see that there are a bunch of different extensions and customizations that are available. So things like MCP servers, plug ins and skills.

If I scroll down here, we can see we have Azure DevOps, impeccable design, Microsoft Foundry, but I can scroll further and take a look at all of these different servers that are available, from Atlassian here to Awesome copilot to AI, foundry, DevOps, Kubernetes, MCP servers box. And it's not just for coding like we talked about using things such as stripe are all directly built in. So you can easily add them here. You can go up and add an MCP server, a plugin, a skill or anything directly from the add button. And if you want to take a look at data, you could filter on data and scroll down and plug in fabric or Elasticsearch or Neon, which is really cool, or super bass, which is one of my favorites. We have all these capabilities. I often tie in agent three sixty five to be able to send emails or work IQ to grab things on my work account. Now, I also want to point here that if we go over to Awesome Copilot, I'm going to open a browser.

I'm going to go to awesome copilot dot GitHub dot com and there's a dash in there. So awesome dash copilot dot GitHub dot com. This is the main community library for GitHub copilot. This is where you can go ahead and install custom agents, instructions, skills, plugins, and even a full learning hub here to the one I love is the extensions which are interactive canvases. Now I've mentioned a few times, but if we take a look here that we have this browser. This is sort of like an interactive canvas. It's a surface that you can create. In this case it's a browser. But I also have the agent interact with. From here you can go and actually implement more and integrate more canvases or create your own. Now there are a whole bunch that are available through the customize. But I also want to point out here that on Awesome Copilot, there are a bunch of community ones. So for example, there's some arcades.

There is an artifact explorer for navigating GitHub actions. There's Azure Functions, Azure Resource Query, and there's eight pages here just full of good things. ESRI agent. This one's a really cool one that I built, which is a backlog swipe triage, for example. Um, there's different integrations for stats and chat cards and a lot more. So one that we're going to look for if we go into search is we're going to look for a Kanban board and we're going to look for extensions and the repository issue Kanban. Now what this is, is it gives us a different way of working with the repositories issues. If you're used to working in a Kanban style, you can go ahead and create them for a sprint. You can look at your entire backlog and a lot more. So you get installation instructions to run directly from the terminal. Or you can come up top and say, open in the copilot app and click on open. And when you do, you'll be asked to install the plugin from Awesome Copilot.

I'm going to hit allow here. And then it's going to make sure that it's able to install it. I'm going to say install. And what we'll then see is that our canvas has been added automatically to the copilot app. So if you browse through and see different features and capabilities, you're able to easily integrate them and turn them on and off right here. I can also look at installed and here's our plugins. I could go ahead and say all types and we can see things that are built in to. So for example, here's that agent merge, create canvas, orchestrate PR stack. I can browse extensions. There's other canvases built in which are all super duper cool. Now from here I can also say, yeah, let's create a new session here for space Quiz. And what this will do is create a brand new session for me to do some work on, and then open my Kanban board. Here we go. It's going to load our canvas. And what we'll see is that I have my backlog plan ready and implement.

Now what's unique here is it is bidirectional, which means when I drag and drop this backlog item to plan, if I come back into my agent, I scroll out a little bit. Here we can see that the canvas has sent over to the agent instructions of what to do. So it's saying, hey, listen, I want you to plan out this feature and update it and then update the Kanban board. So it's bidirectional. We can see it's planning implementation. So it's then going to go ahead and read the details and then update that plan concretely here. And then it will automatically update that Kanban board again, which is super duper cool. So it's a really nice way of working in this case in Kanban style. But I'll also point out here, if I go into like one of my favorites, which is like actions, I can look at the let me see.

Yeah. Repo actions. We can see that here. I've built an entire canvas to help me manage all of my GitHub actions directly from the copilot app. So you can get access to all of these right here. So really, really nice. All right. It looks like it says issue five is now ready. It moved it automatically to the ready state. And now I can drag and drop it into implement. So I can go and have that do it automatically for me. And if you don't like this feature and capability, you can go ahead and say, hey, listen, update the Kanban board, modify it, change it, or create your own version. In fact, if you go in and say slash create, you can create a canvas with any idea that you may have, and it will create it and open it for you automatically. Okay, I'm going to do one more thing here. And this actually isn't in the first steps with GitHub copilot, but I did mention in the intro that we're going to see this live. So let's create one more new session here and let's go ahead and talk to it. I want you to go ahead and set up a repository with GitHub pages.

And I want you to create the actions to automatically deploy to GitHub pages and turn it on for us so we can see our application live, then report back once it's all pushed live with the full URL. And here, because we've logged in and because we're asking it to create and set up GitHub pages for us, it's going to go ahead and figure out how to do that and set up our continuous integration. So every time we make changes, it will go ahead and update. It says pages isn't available yet, so it's going to add a deployment workflow, enable pages action source and then merge into main. So let's go ahead and check it out here. So far it looks like it's updated my Readme. So it's going to figure out exactly where it's at. And it's able to figure it out just based on the normal schema here it says pages are enabled and it's committing this here. Now I did it on a work tree. So it is in a work tree, and I've told it to go ahead and make sure it's running. So it's going to try to merge the PR check for file deployment and more.

But if I go back over into my space quiz, we can go into our settings. And I want to point out here under pages, it just enabled it for us automatically and set it to GitHub actions. We can take a look and see that it's gone ahead and automatically merged that GitHub pages, because I asked it to set it up for us. And now if I come back over to my space quiz, we'll see that we've updated our live site. And if I go ahead and tap on it, Boom. There it is, right there for me. My full space quiz. Ready to go. And yep, it's still Yuri here. We can see that it's gone ahead and told me the workflow is created, the pages were created and so much more. So now if I come over to this pull request and I go ahead and merge it in, which was one that we were working on earlier to go ahead and add shuffled replay capabilities. Since there are no merge conflicts, we can come back directly into our GitHub repo and we can take a look at our actions, which are now firing off to redeploy our GitHub page in real time, which is super cool.

So from start to finish, we not only installed, we took a tour, we created, we build, polished it, tested added instructions, created a GitHub repository, worked with issues and pull requests, went into plan mode, reviewed automated went remote, added canvases. And of course there's so much more. I'm honestly only tipping the very, very top and scratching the surface of the GitHub copilot app. If I mark this as complete, it's good to go. And I'll point out here that there's tons of more resources. Check out the GitHub YouTube for so much more on GitHub copilot app for deeper dives on it. And if you're interested, you can do the same exact one in the CLI or in VS code and see how you can use GitHub copilot there. But we went from idea to backlog to implementation to deployment in record time, all using the GitHub copilot app. And the real power where this shines is as we start to bring in more projects here. So if I add other GitHub repositories like my snake game, I can start to add sessions.

If I go into here and add another GitHub repository like my tiny tool repo here, I can go ahead and create brand new ones. It's going to clone these repositories automatically and start them up here. I can work with sessions across everything. And this is the real power of the GitHub copilot app, not just working on one project, but working on a project with multiple sessions and working on multiple projects with multiple sessions across all of them, and enabling our agents to do even more. All right, there you go. That is the full from start to finish first steps with the GitHub copilot app. We just sat down and we came up with an idea for our space quiz, and we went all the way to full production on GitHub pages, all in under an hour, working on multiple issues, backlog, triage, merge, conflict resolution, and so much more. This is really, like I've said before, really just the tip of the iceberg here of the GitHub copilot app. There is so much more built in, of course, link to all the documentation, of course, to the workshop here and to other videos such as the beginner's series on GitHub, YouTube, that enables you to go even further using the GitHub copilot app.

But get in there, start exploring simply. What I like to do is just ask it to do something, and it probably can. Here's a good one. Go in and say, hey, listen, I have five issues in my backlog. Assuming you have five issues, create five sub sessions and orchestrate them all together and understand if there's any dependencies. The GitHub copilot app will spin up five sub sessions under that chat and orchestrate them all for you automatically. Give it a try. It's astonishing. I hope that you enjoyed this style. I really love teaching. I really love this type of workshop. If you do really appreciate it. If you've made it this far to give the video a thumbs up. And of course, subscribe to the channel for more deep dive long form videos like this and other things along the way as GitHub copilot and Agentic development continue to evolve. And as always, happy coding and thank you so much.