Інструкція для початківців: як використовувати Claude Code
Відео пояснює, як використовувати Claude Code, інструмент на основі штучного інтелекту, навіть без попереднього досвіду кодування. Показано створення лендінгу та підкреслено можливості самокорекції інструменту.
Ключові тези
- Claude Code дозволяє створювати застосунки та аналізувати дані.
- Він може самостійно виправляти помилки під час робочого процесу.
- Інтеграція з GitHub дозволяє контролювати версії.
Простіше за Bubble, але вимагає базового розуміння HTML/CSS.
Самокорекція коду робить Claude Code більш надійним, але потребує контексту та чітких інструкцій. Без планування результат може бути непередбачуваним.
Опис відео▼
I'm going to explain as simple as possible as if you've never touched a piece of code in your entire life how to use claude code. And by the end of this video, you'll have a complete comprehensive idea of the two major reasons why you need to start using cloud code because trust me, it's not just for the nerds. You can genuinely use this in a lot of context. Let's jump in. Welcome back, y'all. Today's video, let's go over everything you need to know about Cloud Code. First thing you need to know about Claude of the Code is that don't think of this as just an app builder, but in alternatively, you can use code in a bunch of different context. So one very specific context is that if I come down here to the bottom left go to get apps and extensions. We can create excel sheets get analization based off the excel sheets and what is essentially happening here is that it writes a custom script just code and it analyzes for you or alternatively create powerpoints via cloud code. But let's go ahead and just create a demo app together so we can get comfortable with this interface and this nice little crab. I think it's a crab. So first things first, what is going on in this input bar here? Auto accept edits means that it can work continuously on a task that you give it. Ask for permission you probably don't want to do. This is like every single time it wants to do an action, it will ask you for permission. And then plan mode, which will do together, which is very important for you to understand what plan mode is. You have the ability to do standard things like attach CSVs, images, everything of this nature. The ability to choose models, the ability to talk to it. Oh, we're talking to AI now. And these two down here, don't worry for now. But local, think of it as literally on your computer. The old days with Microsoft Word where the file was on your computer and it was never in the cloud. That situation. Here's what you want to do when you use cloud code. I don't know what your use case is, but let's assume you want to create a landing page. You're going to go into plan mode. I want to create a landing page for my business. I'm not even going to use developer terminology, y'all, when it comes to creating a landing page to show you how far AI's gotten when it comes to coding. So, we're going to say I want to create a landing page for my business. I'm going to choose the highest level model of this 4.6. And it's going to then ask me where do we want to store everything that's about to be written right now. So, I'm going to hit enter here. As you can see, there's like a bunch of different like connections that we have. GitHub. Don't worry about that right now. But think of GitHub like you have your like local folder or just folder on your computer. This is where you would store it in the cloud. But what I suggest you to do, especially if you're early, is just say choose different folder and then just simply hit new folder and call it like claw test or whatever you want to call it. So I'm going to do that and open. This is just where we're storing everything. This is where the clawed AI can do its work. I'm going to hit enter. It's then going to ask for security reasons like can we access this folder? Yes, you can little claw. And it will get going. the entire process and why you've probably seen this in your feed a lot is because of the fact that we've gotten to the point now with these AI models where you could genuinely have zero idea of how to code but then code real things and the reason why and maybe in the past why it didn't work for you well is because now the AI models are so smart they actually self-correct and what you'll notice in these workflows as you see right here is it will self-correct bad or errors that happen within the workflow process so it needs context it needs to know what we're even building so we'll say SAS tech product and then it wants to know a thing called tech stack. Don't make this difficult. It's like when you make a cake, you know, you get eggs, milk, wheat, like all the good stuff. I want to eat some cake. You choose a tech stack. If you don't know which one to choose, you can talk to it. Ask like what's going on here. We can go something very simple like HTML, CSS,JS, and then what's going to be on the landing page. Now, a lot of this stuff might be like this is confusing Corbin. This is confusing Corbin. Then open up another chat claude. Ask what this means, but just click through. This is very experimental. And what we're going to do here is we're going to build a landing page here that would have alternatively taken you a couple days or alternatively costed you money and literally while it's creating this landing page that would have taken hours days. I'm just drinking coffee. Put me in a coffee world. Coffee. So TLDDR of what model to you choose for what when using clawed code of his 4.6 if you want to mitigate the amount of errors. But as you could probably already assume there is more usage around this like how often you can use it. But in reality, if you have no development experience, but you want to create really cool things, I would stick with 4.6 because it is very forgiving. Let's create this landing page. So, why is it just plan mode? And even if you have no clue what any of this means, is at least you can get like a higher level of okay, sticking header, navbar, like it's creating something here. And also, this is a big pro tip. It makes it so your outputs are way higher quality because think about it. If you just told make a landing page and it has no context, it's going to wild out. give it a plan like you would give a human approve clause plan and start coding. You simply click this and it'll proceed. So sometimes it'll ask you can I do a certain thing even with auto accept on. You just say always allow in the session. Now I want you to realize what is happening right now. And what you'll see is that it is quite literally taking screenshots of the landing page here confirming it looks good and it's going the extra mile here to make sure it works on your phone. As you see mobile responsiveness here it is taking screenshots and all live with no input. Therefore, as you can clearly tell, right here is where we're going to be able to put input change build. Right here is where we see our build happening live. And in the bottom right, don't worry that it's red. This is essentially what in software development is a terminal and where we get messages of what's happening in the system here. What's cool about this is that at any moment, you can simply copy this information here, put it here, and say what's happening or fix it. So now that you know officially how to make a landing page and the same steps and processes will incur and happen when you want to build other maybe tools on your computer to make a presentation or Excel sheet. Let's understand fundamentally what we did initially when we created that folder. Basically we store code in things called repositories and what you saw when we set up earlier in the video was in a local folder which means that if the computer that I build this on absolutely blows up, destroys fire, I lose it forever. Also, we don't have the ability to maybe say, "Hey, this is a good checkpoint. I like this." So that if any further iteration you do on the landing page, you break it. You don't have the ability to roll back because it's all local. This is where a thing called GitHub comes into play. And you should learn how to use GitHub. This is the cloud. This is what allows for what we call version control. For example, this is version one. We keep working on it. Work, work, work, work. We like it. Then we save another checkpoint called version two. Corbin, how do we do this? And what on earth is GitHub? Well, you're in luck. I built two videos, both over like 500,000 views, how to use GitHub for beginners and GitHub for experts. Essentially, check out this video's description down below, and you're going to learn how to do GitHub, where you can connect it to this cloud code workflow and make sure you can have versions of everything you build with cloud code. And cloud code is not where it stops. Check out techsniff.com. It gives you new stories all in one place every single day. It sources it from everything across the internet. Expost, Reddit post, forms, whatever it may be. Description down below, techniff.com. So therefore, that is cloud code. If you have zero to no development experience, it doesn't even matter anymore. You can build whatever you want. So make sure you leave a like. I'll see you in the next Wait, Corbin, I want to learn more. Check out this channel. I've been doing it for 3 years and there is literally hundreds upon hundreds upon hundreds of hours of me showing you how to use AI with no experience in coding and developing real software type of video.




