This video will teach you Claude code.

I'll go over everything and assume no prior knowledge.

I'll walk you through the setup and installation.

Step by step.

I'll show you how to utilize the tool, the best practices, multiple features,

and by the end of the video, you'll be comfortable using this tool to generate

some pretty insane outputs and awesome coding projects.

So with that said, let's get into it.

So without getting into too many details, cloud Code is effectively a command line

or terminal application that allows you to generate code,

debug projects, make websites, create applications, whatever it is that you want.

It is effectively anthropic version of, you know, cursor or something, right?

Where you can generate a bunch of code using the terminal.

Now this is prompt based coding, which means you don't need to actually write

any of the code yourself.

You simply type to an AI agent, you tell it what your desired output is

and then goes ahead and starts coding.

Now, oftentimes cloud code will be used in combination

with other developer tools, and I'll show you a few of them in this video.

But generally speaking, you do not need to be an expert to use this.

And even as a complete beginner, you can create some really cool projects.

So with that said, let's look at how we set it up.

Now, in order to get cloud code working, you are going to need a cloud subscription

which is paid.

So you're going to need a Pro Max teams or enterprise subscription or an API key

with some tokens or credits from the cloud console.

Now, I would suggest that

if you're just looking to mess around with this, get the cheapest plan from cloud.

You will run out of credits pretty quickly if you are using this for very heavy work.

But of course, test it first before you start going and buying a bunch of different credits.

It also matters what model you're using, which we'll talk about later,

which will determine what the overall price for all of the prompts that you send is going to be.

Now, generally, you're not going to be overbuild.

You'll just run out of credits, then you would have to buy more.

So don't worry about that.

Okay.

Now to install this is going to require that we work directly inside of the terminal.

So what we're going to do is we're going to open up a terminal or a command prompt or a PowerShell

instance, depending on the operating system that you're working on.

Now, if you are on windows, then I'm going to suggest that you open up PowerShell

so you can go to the windows search bar, search for PowerShell, and then open up a terminal like this.

And I'll just put it on the right hand side of my screen.

Now if you're on Mac or Linux then open up a normal terminal.

So just type terminal in the spotlight search.

And then you'll be in an environment where you can start using cloud code.

Now in order to install this I'm going to leave this documentation in the description

which gives you the commands that you can simply copy depending on your operating system.

So if you were on Mac or Linux then you can copy this curl command right here.

Just paste it into your terminal and run it.

If you are on windows and you're using PowerShell as I suggested, then you can copy this IRM command.

You can paste it, hit enter, and it should install cloud code for you.

If you're using the command prompt, maybe you're on an older version of windows.

Then you can copy this version right here.

Paste it in the command prompt.

Again, it will install the tool.

Cloud code is not a desktop application.

It actually runs directly inside of your shell or your terminal.

So in order to use this in the future, you will need to open up PowerShell again or terminal again,

or use this from some kind of tool in order to actually interact with cloud and generate code.

Now, once

it says it's installed, the way that you can test this is by simply typing Claude in your terminal.

So if you type Claude, you should get some kind of output here.

Now when you first run this, it will ask you to authenticate with your anthropic or cloud account.

So just do that.

Go into your browser, sign into your account and then you'll be good to go.

Now also whenever you run this for the first time and you type that command,

it's going to ask you to trust the current folder that you're inside of.

So we're going to go ahead and press on. Yes I trust the folder.

And then it's going to bring up the Claude Code UI.

Now at this point you should have got cloud code installed and just remember,

if you were to close the terminal or the PowerShell you would have to reopen this.

Right?

I'm just going to zoom in and then again type Claude and then you can press enter.

And again you get back into that user interface.

So don't worry if you lose it.

Now for some of you that are complete beginners and already this is a little bit intimidating.

Anthropic actually did create an easier way to write code using Claude.

It is not the same thing as Claude code, but it is very similar

and you can access that by downloading the desktop app for Claude.

So if you have an account with them, you can just download the desktop app.

If you do that, the app will look something like this.

From here, what you'll be able to do is switch between this chat and this code mode.

And if you go into the coding mode here, you'll be able to select a project.

You can write a prompt directly inside of here you can change the mode like coding,

asking, planning, right. You can add different files.

You can connect it to GitHub, you can choose your model and it's a lot simpler and easier to use.

So again, if you're a beginner you can just use this.

But if you want to get any serious work done, the CLI or the command

line tool which I'm showing you is what you're going to want to set up and have on your machine.

Now, if you've been programing for a while, you probably know the job market is tough recently

and you may feel like traditional resumes just don't show who

you really are as a creator or a technologist.

Now that's where dinked me comes in.

It's a brand new AI native career network built

for the real era of real work, real projects, real impact.

Not just job titles.

Now, instead of a static resume, Dink gives you a living identity card

your Dink card that automatically showcases your code, papers,

citations, and measurable impact across platforms like GitHub and Google Scholar.

Recruiters, professors, and startups can discover what you've actually done,

not just what you wrote on paper.

Now that's what dink does for you, but for companies.

Dink is an AI powered talent engine that analyzes and compares researchers and engineers in seconds.

It helps teams spot the best fits faster than traditional recruiting tools.

Now, the best part of all of this is that you can claim your Dink card for free.

Start building your personal identity in minutes and let your work speak for itself.

In a world where resumes get screened for six seconds before a decision.

Now here you can even see my dink card.

Right here has all of my details, information that you wouldn't see on a resume,

my whole career trajectory, my GitHub, the number of stars, my YouTube channel, Instagram,

all of that stuff is here.

As well as networks career trajectory, Twitter, and anything else that you want to include.

Now you can get started with Dink today for free by clicking the link in the description,

and a massive thanks to them for sponsoring today's video.

Let's get back into it.

Now, with that in mind, before you just go crazy here with Claude,

what I'm going to suggest that you do is installed git on your computer.

Now, git is a tool that's going to allow you to essentially kind of checkpoint

what Claude is doing and revert back to a previous version of your code base

in case it makes any mistakes.

You won't have it by default on your machines.

You do need to install.

Now. You can do that by going to the link that I'll put in the description

where you can download this for windows, Mac or Linux.

However, because we have Claude code, we actually can just ask Claude code,

can you install git for me?

Now if you do that, it should actually think about the commands

that it's going to need to run to install this for you.

It should likely ask you for permission.

So you can see here it's asking for permission to run a command.

I'm going to go. Yes.

And don't ask for this command again.

And then it's going to actually run the command to install that tool for you.

Now for some reason that doesn't work.

Then what you can do is just install it directly from the website right here.

Okay.

So in my case, it actually gave me an error that said that I need to be running

PowerShell as an administrator in order to actually run this command.

So if you get an error like that, no worries.

What you can do is you can just open PowerShell by pressing right click and then run as administrator.

Now if you're running this in Mac, it should be good to go by default.

But here what I'm going to do is just type Claude.

Okay I'm going to go yes I trust this folder.

And then same thing install git for me and let's see if it's able to do it.

Now that I gave it the permissions.

All right. So I just got installed. I'm not going to go through the whole process.

And the next thing that I do recommend that you do is create a GitHub account.

The reason why will want to make a GitHub account is we're going to connect Claude to GitHub,

so that it's able to actually create new repositories for us and save all of the code

that we're writing from the very beginning, so that we don't lose anything.

Now, GitHub is essentially a website or kind of a remote location where we can store different code.

So rather than just having it on our own computer, we can store it with GitHub

where we can access it from another machine, we can share it with another developer.

And more importantly, if we lose access to maybe our own computer,

or maybe we delete the folder or Claude messes up or something,

we can always get the repository or the code again by going to this website.

So go to GitHub, make an account.

It is completely free and I'm going to just quickly show you how to connect

your GitHub account to your terminal here locally.

So you're able to use it kind of with Claude.

All right.

So in order to connect git, which is running on our local computer to GitHub,

which is running in the web, what we need to do is run the following command which is g off log in.

Now notice that I didn't know how to do this.

So I just asked cloud code and it told me exactly what to run.

So what I'm going to do is type g auth log in like so.

And then it should run this command for me.

And allow me to connect to GitHub. Okay.

So you can see that it's going to ask to run the command.

So let's go ahead and type on. Yes.

And in this case G was not found.

So it should go ahead and install that for me.

So I'm just going to type option C so we can install that tool for me.

Okay. So it looks like it was able to install that.

And then I'm going to go yes so that it can run that command.

All right.

So notice it's giving me some output here where it says

it wants me to go to this URL and then type in our one time code.

So what we're going to do here is go we're going to select the account.

So I'm going to go with tech with Tim.

And then I'm going to copy this code that it's giving me here.

So just copy that and paste it and press continue.

And that should then authorize GitHub here in our terminal.

Now in order to do that I need just auth with mobile. So let me do that.

All right. So it looks like we are all authenticated.

Now you can see that I was logged in as tech with Tim.

And now we have GitHub set up.

I know it is a little bit of a pain to do this at the beginning, but trust me

it is absolutely worth it because now anytime we use cloud, it can automatically save the code for us.

A lot of other tutorials don't cover this.

That's why you're watching mine,

because I want to go through it step by step, and I want you to get set up in one single video.

Now, I promise we can move on to some more features.

Okay, so let me close my browser.

Let's open up cloud code a little bit larger.

And I want to show you how most developers actually work with cloud code

now, rather than just using the terminal here.

A lot of times you're going to use this with a development environment.

What I mean by that is you're likely going to use an editor like VSCode

or a cursor or JetBrains IDE, or something along those lines.

So you can actually see what cloud code is doing and not just have this terminal view.

So for example, a very popular editor is Visual Studio Code.

A recent version of that which is kind of AI enabled is cursor.

So for example, you could open up cursor.

I'm just going to open up an editor right here.

And if you want a recommendation

I would suggest just download cursor because this is the best one to use for right now.

That also has its own AI features, but we can augment that by just using cloud code.

So you can see that I have, what is it?

Kind of like a repository set up now when you open up one of these applications.

So let's say you download Curser or you download VS code or you get any code editor

on your computer, what you can do is you can go to file, you can go to open folder,

you can go to your desktop or any location that you're going to be able to find in the future.

And we could do something like clawed code test.

And I'm just going to make a new folder.

How do I already have a folder called Code test?

Okay. Well, that was not what I wanted.

So let's go.

Cloud code test two.

All right. So I'm going to make a new folder.

The reason I'm making a new folder is this is where I want my coding project to live.

I'm going to select that and then I'm going to open it.

So again I've just open my code editor.

Doesn't matter which one you're using.

You can ignore all of the crazy features from cursor here.

And we can close some of these windows.

And kind of just hide them.

And what we can do is we can open up the terminal directly inside of here.

So if you're working inside of cursor or VSCode or windsurf, so something that is like

a vs code fork, because a lot of these editors are based on Visual Studio Code.

What you can do to open the terminal is you can try to find kind of like a drag bar

at the bottom of the screen, but you can see I'm having trouble finding it here.

Instead, you can hit control and then backtick on your keyboard

or command, and then backtick on your keyboard, and it should open up

a terminal directly inside of this folder that you opened for you.

Now, if that's difficult for you, you can hit Ctrl, shift P,

or Command Shift P if you're on Mac and you can type terminal

and then create new terminal, and then just choose one of the instances.

So command prompt or PowerShell.

If you're on Mac or Linux again you just choose probably the first one that pops up.

So let's say I do that and you can see it makes a new terminal for me.

And I'll be able to see it on my screen.

Now if you want to see the left hand side bar, which is the file viewer,

you can press this button right here.

Again this is in cursor, but you can do this in another editor as well.

And you're building.

Now see all the files that cloud code will generate for us in one minute.

Now again if you can't find this button you can hit

Ctrl B or command B on your keyboard

and that will toggle the file view on the left hand side.

Now I know it seems complicated, but most of the time you're going to want to work like this

because you want to see the files

and actually know what cloud code is doing, especially if you're unfamiliar with the terminal.

I would highly suggest opening this in some kind of user interface

like we've done here, so you actually know what's going on.

Okay, so now we're inside of a terminal.

This is the exact same environment I was just showing you before, except it's now in this code editor.

Right. And again, this code editor is called cursor.

In case anyone is confused, if you just go you can type cursor

download on the internet and you can download it super easily.

Just open it up and literally all you have to do is just go file, open

folder, open a new location and you can start writing code.

So from this terminal we're going to type plot.

And when we type clod it's going to open up cloud code just like it did before.

But now in this location.

So same thing I'm going to press the folder.

And now I have cloud code in my terminal and I can start getting it

to actually generate some stuff.

So now we're going to actually start using it.

We're going to start building a project.

And I'm going to start walking through a lot of the features that it has.

Now you'll actually notice that because I'm using it in this editor now it's detecting that.

And it's telling me that I can actually have some special features

here in Cloud code because I'm working in cursor.

So what I'm going to do is just go ahead and press enter.

And now we can just start working.

But again it's going to be able to reference the files that we actually looking at in cursor.

It's going to show us the changes that it's making.

It's very powerful.

All right.

So without further ado, let's get this to do something.

And let's say that we just want to create a simple web based game.

Let's go with something really, really simple for right now.

Like maybe tic tac toe and just ask it to make that game.

So I'm going to say make tic tac toe

that I can play on the web.

Okay.

Now I'm going to start going through all of the other configurations and settings.

But for now, let's just run this

so we can see some kind of output and know that we're actually getting some result.

Okay.

So you can see that it's going to start doing this and generating some output.

Now, something worth noting here is that if you wanted to,

you actually could run cloud code multiple times in parallel.

So while I was creating tic tac toe, for example, I could create another game.

I could audit the code base.

I could, you know, change something, whatever.

And that's one of the things that makes this quite powerful.

I'm not going to do that right now because again, we're still at Kenner level.

But you get the idea.

Okay. So you can see that it's generated some code for me here.

It's kind of giving it right in the terminal, and it's asking me to allow it to create tic tac toe.

Now, what I'm going to do is just go here and I'm going to go, yes, allow all edits during the session.

And notice I'm using my arrow keys here to navigate.

So I'm going to press enter.

So I can just go ahead and start writing this without asking me for permission okay.

So you can see that it says it has created this for me.

And it's telling me that the way that I can run this is to start

the Tic TAC token, HTML or double click the file in the file Explorer.

So I'm just going to say start tic tac toe dot HTML.

I'm not sure if that's actually the correct command,

but let's just run that and see if that's going to do anything for me.

It says open it in the default browser. Yes.

And don't ask again for this.

And let's see if it can open this up for me.

And I can run it okay.

And you can see this is my browser.

And we have the tac toe which has now been created, which looks pretty good.

We can generate a new game. You get the idea.

Now, the one thing that I want to bring your attention to here is that as soon as we started

running this, a Doc Claude folder was automatically created for us

in the directory where we're working, as well as the code that was generated.

Right. So you can see tic tac toe now has been written.

So this is pretty cool.

And this is the reason why you would do this inside of a code editor.

So you're able to actually see the changes in the code that's being generated in real time.

Again, you don't need to do that.

But that's what many people like to do when they're using a tool like this.

Okay. So we've got something created

now we need to start going through a lot of the different configuration and setups.

Now keep in mind if you don't know how to do something, just ask Claude to help you.

It can help you.

It can change configurations. It can set stuff up for you.

But of course it's useful to know the basics.

So inside of here, as you already saw, you can just type human language and you will get some output.

However you can also type slash help.

Let's start with that.

And this is going to give you a general kind of shortcut or list of commands.

And things you can do in Claude.

So you can see that I can toggle between a few different things here.

And you'll notice that if I put an exclamation point, this is going to put me in bash mode.

If I do a slash, this is going to allow me to type different commands.

If I do an ad, this is going to allow me to reference various files or paths.

If I do an ampersand, this is going to be for background

tasks or keybinds or whatever that says I think.

So for background new line.

Not exactly sure what that means.

Actually, I've never used the ampersand.

Continuing we go to commands and if I scroll down

here, you can see there is a ton of different commands that I can run.

We cannot possibly go over all of them in this video, but I will show you a bunch of them

that are useful. So if you're ever like, how do I do something?

Or I wonder what's possible type slash commands and you can see a bunch of stuff pops up

and you're probably going to learn something just by looking through this menu and then custom commands.

Of course, you can make your own commands.

We're not going to go over that right now.

Okay. So first thing that you want to be aware of

is the different modes that you can use inside of cloud code.

You can have a planning mode, you can have an asking mode and you can have a coding mode.

Now the ask mode is the default mode where it's going to ask you

before it does something so you can see it asked before it generated a command.

It asked before it wrote code.

Write the code mode. It's just going to code by default.

It's not going to ask you for permission, and the planning mode is going to create a detailed plan

that Claude will follow.

So that it has a little bit more structure.

In order to change the mode, you can hit shift and then tab on your keyboard.

So shift and Tab, you can see it showing shift tab to cycle we have plan mode.

We have the normal mode right which is this.

And then we have the coding mode which is kind of except at its on.

So here planning mode right.

Accept edits.

It's just going to do the thing

and then the ask mode is the default mode where it's not showing you anything there.

Again to do that is shift and then tap.

Now if you want to see the shortcuts, you can type again the question mark

and you can see the different shortcuts that you can use.

And it will start showing you all of the different ones that you can put directly inside of here.

Okay. So that is how you change the mode.

So the first mode that I want to change into is the planning mode.

And I want to show you a few other shortcuts.

Let's say you write something super long here and you don't want to have to just like erase all of it.

Well, what you can do is hit escape twice on your keyboard

to escape escape, and that's going to clear whatever is currently inside of this window.

So escape. Escape.

Now, if you want to go back to the previous thing that you typed, you can use the up arrow key

so you can actually toggle through the previous commands

you ran by using the up and down arrow keys in your terminal.

There's actually more shortcuts that exists there, but if you just want to cycle

through the previous commands in case you're repeating something, you don't need to type it again.

You can just cycle with the up and the down arrow key.

Now, another thing worth noting is that sometimes you want to actually create a line break in the prompt.

So for example I have like hello.

And then I want to go down to the next line. Right.

I don't want to press enter because if I press enter it's going to submit the command.

But I want to move actually down and have like kind of multiple lines.

So in order to do that you hit the alt or the option key and then you just press enter

so you can see that what I did on my keyboard here is I just press alt.

So I'm holding Alt on Windows.

If you're on Mac it might be option.

And then I press enter

and I can actually go down to the next line so that I can have things on multiple lines.

And then again I can clear that by hitting escape twice.

So again that's alt or option.

You hold that down and then press enter okay.

So we're in the planning mode. Now we've learned the commands.

We have a general kind of lay of the land when it comes to using this.

We have to first decide what it is that we actually want to create.

So the most important part of AI coding is actually knowing what you want.

If you don't know what you want and you look the agent steer you,

you're going to get horrible, crazy results that take you forever to debug.

And change.

But if you're very clear and specific on what you want and you have it do one

small task at a time, you can get fantastic results and create amazing stuff.

So this is just a tutorial.

I don't have anything crazy that I want to build, but I definitely still want to come up with a plan

for what it is that I want to do.

So I always start in the plan mode, and what I'm going to do is actually stream

my thoughts into Claude and ask it to come up with a plan to create this project.

I don't need to know the code base, I don't need to know the languages.

I don't need to have done research.

It's obviously helpful if you do those things, but if you just want to start completely from scratch,

we can do that.

So let's just stream our thoughts.

In order to do that, I'm actually going to use a dictation tool on my computer called Whisper Flow.

So rather than typing, I'm going to speak into my mic.

I want to create a simple top down shooter game that is runnable in the browser.

I want to have some sprites, or at least cool.

Kind of like character animations.

I want my character to have a gun.

I want enemies to kind of be coming at him from different angles.

I want to be able to move the arrow keys

and then use my mouse to kind of point and click at the different enemies and shoot them.

We should have levels, we should progress.

Clearly, we should have a menu screen.

We should keep it relatively simple and kind of like 2D retro style.

Okay, so that is my thought process.

So I've just streamed that into Claude and then same thing I'm going to hold alt and go down.

I'm just going to put kind of a little separator and I'm going to say

create a plan for this game, and then we'll start executing and coding it.

So I'm just instructing it even though it's in the plan mode.

You know, I want to plan this out before I proceed.

And you can use this to plan features not just entire apps like I'm doing right here.

Okay. So I'm going to go ahead and press on enter here.

And it should start generating the plan for me.

Let's give that a second to make it okay.

Now just while this is running I want to point your attention to a few things here.

So first you can actually see how long it's taking to run, as well

as the number of tokens that are being used when it's doing this generation.

So if you really care about the cost, obviously something to pay attention to.

Now you also notice that this is able to use tools.

There's some tools that are built into Claude. There's other tools

that you can actually install yourself if you want to make this more capable.

So for example, if I hit control O, which it's showing me right here, it's

going to expand this so that it's actually showing me, you know, exactly what's happening.

And I can see what it is that's kind of going on here.

Right.

So it says okay, I'm thinking the user wants to create a game.

All right.

This is a fun project okay.

Here's what I want to do. Blah blah blah blah blah whatever. Right.

And then if I want to go back, I can do control.

Oh, and it brings me back here to where I can kind of see,

you know, what it's doing in the plan that it's generating.

So if you want to open it in like verbose mode where you see the logs, it's control up.

Now you also can have this trigger to run in the background.

I'm not going to do that right now.

But if you hit control B or command B, depending on your operating system, you can trigger this to run

as a background session and then that way you can actually just use cloud code

for something else in the same terminal view while this is going on in the background.

And again, you can have as many background tasks as you want running at the same time.

Okay, so the plan was created.

However, it's asking me a few questions here that I need to answer.

So I'm going to go with no sounds for now.

So let's answer that.

And then it's also asking me about the game size and I'm just going to go full screen actually

you know what. Let's go 800 by 600 just to make it easier.

And I'm going to submit my answers to Claude.

So another cool thing that it does is it clarifies with you before

it actually starts going and generating all of the tasks.

Now while this is running, I will show you that you can change the model that's being used here.

So by default, I believe this is going to use the most powerful model.

Or actually, let me just answer this first and then we'll go into the model.

So Claude has written a plan and it's ready to execute.

Would you like to proceed?

Yes. Auto accept.

Yes. Manually approve edits.

Type here to tell Claude what to change.

Yes, I'm going to go. Yes.

Clear the context and auto accept edits.

Now, when you clear the context, what that's going to do

is essentially remove all of the previous stuff that you've message Claude with.

So it's no longer being included

in the API request to Claude, and it's just going to go based off the plan.

This is good to save tokens, but if you wanted to keep the context,

then you could do that right by auto accepting the edits, etc..

So I'm going to go yes, clear the context and auto accept this and you can see that

it is going to start actually generating this and going through the plan.

So this is the plan to implement.

And there you go. It's going to start building.

Now while it does that I did want to talk about the models.

So it can be quite expensive to run this.

And you can very quickly run through your usage.

So if you want to make this more affordable for yourself, you can change the model

that's being used here.

So if you type slash model here and Claude is not currently doing something,

it will bring you into an interactive window that will allow you to change the model.

So let's actually just do this from another Claude code here.

So I'm going to go slash model.

And when I do that it's going to give us three options okay.

So opus sonnet and haiku or however you pronounce that now you can see the three different models here.

So opus is going to be your most powerful model.

And as it says it's for more complex work.

This will be the most expensive but also just the best.

Generally speaking, sonnet is kind of a middle ground model.

So if you're doing some tasks that don't have extreme complexity,

don't require a bunch of tool calling, you can select this one by just pressing enter.

And if you want a faster model that's going to give you quick answers

and be significantly less expensive, then you can switch over to haiku.

Cool.

So now we've gone back to our first kind of cloud code instance here, and you can see that

it says it's implemented these features in the shooter's HTML file.

So what I'm going to do now is just say run the shooter file, okay.

And I'm just going to have it run this and open it in the browser for me

so I can actually see what it looks like.

And then of course if we have revisions let's go.

Yes. And don't ask for star commands.

Oh. It says it cannot find that for some reason.

Okay. Let's see if we can run it again.

Okay. And sorry, I've got this running now.

For some reason, I made a mistake and you can see we have this kind of retro shooter,

which I think looks pretty cool in my opinion.

And I can kind of move around, I can shoot.

We have, like, this wave of enemies that's coming.

Okay, cool.

And that's I mean, like literally exactly what I asked it

to build, which is pretty good.

I would say for a first iteration.

So we also have health, right?

We have like the menu bar, a bunch of that stuff.

We could obviously add add sound effects.

Cool. Okay.

So let's close that.

And if we wanted to make revisions and same thing, we could just do something like, you know,

add power ups

to this app and then it would go ahead and be able to do that.

Now before we do that though, I want to talk about git and GitHub.

So if you're going to be working on a large project, I highly recommend that you instruct cloud code

to create a new GitHub repository and to make commits and save all of the changes that it's making.

So the way to do that would be something like this.

From now

on, I would like you to make a new git repository for this project.

I would like you then to make clean commit messages, save them locally

and then push them to GitHub so that we always have a saved version of the project.

And it's easy for us to revert back in case we make any changes.

So set up a GitHub repository,

configure everything, and just use getting GitHub for the rest of the project.

So we're going to tell it that let's go ahead and press on enter.

And it will just start running all of the commands that's needed to set that up.

So you don't really need to do anything here.

It should just know what to do. So same thing.

We're just going to say yes and allow this to go and it will start setting it up okay.

So it took a second here but it did work.

And actually if I open up this URL now

in my browser, we should be able to see the git repository that it created.

Again. Notice it automatically created it for us.

And you can see now we have us as well as Claude.

They're adding commits to this repository.

So now it is saved and all of the work is going to be done

automatically with, what do you call it, git and GitHub.

Okay.

Now let's go through a few other things that are worth noting.

So I just told it right.

Like always use git and GitHub for this project and automatically make the commits.

Now sometimes it will do that just automatically.

But the thing with Claude is that as soon as I exit this so I go like slash exit, right?

And if I just go Claude again and I start with a new session,

I've just lost all of the context and the history of what was happening inside of here.

So if I ask it like, what did you do last or something like that,

it can look at the git history, but it's not going to really know what happened.

You can see this is the start of our conversation.

So I haven't done anything yet. Right.

Because we started a new session.

So that's an important thing to understand about Claude Sessions is that you

don't have this kind of like unlimited memory like you might have in some other tools.

And that's where we are going to get into talking about the Claude and D file.

Now, the Claude MDF file is something that will allow you to have persistent memory.

So Claude kind of knows the rules of your code base.

It knows what it's done before, what it should and what it shouldn't do.

And you can set it up to be really complex in advance so that you don't need to keep instructing it

to do the same thing.

Because again, if I start working right now, it's not going to automatically

commit this to get in GitHub like I want it to do.

So in order to create a Claude MD file, what you can do is type this slash init command.

When you type slash in it, what it's going to do is automatically create

Claude Dot, MD for you and in the claw MD,

it will actually put in different information based on the analysis that it has of your code base.

Then you of course can modify this Claude, the MD file, and I'll show you how to do that in one second.

But this is effectively, again, the persistent memory for Claude when it's working in your code base.

And we'll give it a second here to generate that MD file.

So let's go.

Yes. Go ahead and create that for us okay. Yes.

Go ahead and make that.

So that will have that file there.

And it will always reference it when we create a new session. Okay.

So you can see that it is showing me what the Claude MD file will look like that it's

going to create here where it goes over like project overview running

the game development set up, you know, whatever, all of this kind of stuff.

So I'm going to go yes.

And allow edits during this session.

And then it's going to create this Claude dot markdown file for me.

Now if we go through here okay let's see if it has anything else.

Design pattern color scheme get workflow okay.

Cool.

Awesome.

And then it explains kind of all of this stuff.

So now if there's some specific things that I always want Claude to do in the project,

I can just tell it that.

So I'm going to say update the Claude MD file to indicate that as you do work,

you need to commit things to git and push them to GitHub regularly with clean commit messages.

So we never lose the status or work that we've done.

Okay. So let's go ahead and press on.

Enter here and you can see I said, you know, Claude MD

whisper actually picked up that this was a file and it automatically tagged it for me.

And now you can see it's making changes, right.

And then updating that in Claude DMD.

And if we scroll down here it explains what it needs to do.

So now if we were to restart this.

So let's go slash exit to get out of that okay.

And go Claude I'm just going to say what do you know

about this project.

Okay.

And it should be able to look in this claude.md file and give me some information okay.

And you can see it reads the cloud.md file because it knows to go there.

And it gets us all of that persistent info.

All right.

So now that we've gone over that I want to talk about tasks.

Now if you run these tasks command, this will show you any background tasks that are running.

Now. Background tasks are typically going to be things like web servers that are running

continuously, and that you don't want to block the execution of your cloud code instance.

So right now, if I want to get out of this, I'm just going to press, escape to get out of that there.

And you could see that I've kind of got out of the dialog.

Now, tasks might be something like a web server,

a front end, I don't know, like react website that's running whatever.

There can be all kinds of background tasks.

And by default

Claude will put some of them into the background, but you can also put them in the background yourself.

So what I want to show you is if I do something like, you know, run a web server

to host my two sites or host my two games, what's going to happen

here is Claude is going to detect that this is a task that would be blocking.

Blocking means that if we just ran it

in this terminal, that it's going to block us being able to do anything else.

So you can see that we now see one bash, which means that there's a task

that's happening and it says it's running this in the background.

So if we want to see that we can hit control T, which will show the different

tasks, we can go to the bash and hit enter to view it.

If we go back, we also can type slash tasks.

And then it's going to show the task that's running and how long it's been running for.

Now, from here, if we want to kill this task.

So I no longer want the web browser to be running, for example, I can press okay

and that's going to kill it for me.

And then I can go back just by hitting enter. Okay.

And you can see background command was killed right now for some reason

it's going to go and generate this again I think.

But if I wanted to stop this then what I could do is press escape.

So for example,

if I do something like hello and then I press escape super fast, you'll see it's interrupts that command.

So if you want to stop the execution of something in lifetime press escape.

I probably should have mentioned that earlier, but I honestly just forgot.

But that's something to know.

Okay, so those are tasks.

Again, these are things that are kind of like

running in the background that are typically these long running operations.

Now however, if you do want to create kind of parallel

clawed bots that are like writing code or doing different operations for you,

whether they're touching files or modifying things and that's where we get into agents.

So if you go to slash agents, you can create agents,

and these agents can perform special tasks and you can delegate tasks to them.

So for example, you could have an agent that is handling or doing a styling right.

Or like changing the user interface, an agent responsible for the back end.

Okay, I'm not going to get into all the details

of how to set this up, because it does get a little bit complex and kind of beyond

the scope of this tutorial, but I want to show you things like that do exist.

And we also have things like MCP if you want to connect to, for example, Notion

or Gmail or Google Drive, or you want to connect to a bunch of other services, you can do that

right from an MC server inside of Claude, which really expands its capabilities.

Again, let's get out of that. Let's go through slash.

You can see there's a bunch of other stuff that you can do here.

You can set up hooks, for example, that will automatically run code for you or different tasks.

You can set up Keybinds memory, you know, all of this kind of stuff.

So just go through it, have a look.

And another thing that's useful is skills.

So you can actually create skills that are essentially repeatable tasks

that Claude does all of the time.

So rather than creating an agent, you can create a skill and then you can run

a particular skill by just saying, hey, you know, run this skill that I've taught you how to do.

It will go look up the skill and then be able to run it continually.

So you can see how powerful this can get if you really go through the full configuration.

But at this point, if you follow along, you know how to use cloud code better than 99% of people.

I showed you all of the core functionality and features, and you can go out and start

coding effectively whatever you want, because now you know how to navigate this tool.

I know this was long.

I know it's a bit complex, but I want you to get the proper setup.

So you didn't need to watch a bunch of other videos.

If you guys have other questions or you want more tutorials on cloud code,

then definitely let me know and I will see you in another video.