How to Become a Claude Code Master

Over the last few months, Claude Code has added a ton of new features that aren't talked about or documented anywhere. In this video, I'm going to go over five of those features that will immediately start giving you value. They'll help you save money, get better performance, and create way better applications. If you stick with me until the end, you'll be using Claude Code better than 99% of people. Let's get into it.

All right, so let's go over a really simple one. that's going to bring you value immediately and that is the new model picker you can use that will automatically choose the right model based on what you're doing. As you know right now you choose your model by doing slashmod and then typing in opus or sonnet or whichever one you want to use. If you choose opus yes that is the best model but it's going to eat a ton and ton of credits from you. If you use sonnet that's also a really good one and you're going to save money but it's not going to be as good at certain tasks. So get this. You can now make it so it automatically switches based on what you're doing. You're going to type in /model and hit enter. Then now you're going to use opus plan mode. That's this brand new mode right here. Number four. You're going to scroll up to it and hit enter. And basically what that's going to do now is whenever you're in plan mode, it's going to use opus. And whenever you're in execution mode, it's going to use sonnet. If you're doing your plan modes with Opus, you're going to get incredibly good, detailed plans that will make it so you only have to use Sonnet for when you do execution. This is going to save you a ton of credits and get you the exact same results you were getting before. If you use Opus for everything right now and you're not on the $200 a month plan, you're going to eat up your credits so unbelievably quick. But if you use Opus just for planning and sign it for execution, you'll get the same results as using Opus for everything, but you'll save a ton of credits and you won't have to do any extra work. So do that. Do slashmodel, hit enter, and then move to Opus plan mode. This will get you great results and save you tons of money, and you won't have to think about it ever again. The second tip I'm going to show you might be my favorite of the bunch because it's going to give you a whole bunch of cool new information right in your Claude Code UI. Check this out.

So, see this down here? This is a status line in Clawed Code. And this is brand new. And what you can see here is a few things. One, the weather of where I'm currently at. Two, the time Three, the model I'm currently using, so I can always stay on top of what model's being used, so I know I'm not using too many credits. Next, I can see what branch I'm on in GitHub, so I can always make sure I'm committing to the right branches and I'm writing code in the right branches. And the last thing here is a tomato that says done. What this actually is a pomodoro timer. I can just start at any time I want so I can stay focused. Let me show you. So if I type in / pomo and hit enter, it's actually going to start the pomodoro timer right here. Watch this. Start a 25minut work session. Your 25minut work session has started. And now there's a timer in my status line that's counting down from 25 minutes. So I can make sure I stay focused during those 25 minutes. I don't know about you, but I get constantly distracted when using Claude Code. It's so good and it works so independently. I am constantly distracted and doom scrolling as it's working. This Pomodoro timer is going to make sure you stay focused. So, how do you set this up for yourself? Well, it's really simple. In Claude Code, if you type in /status line and then type in what you want in your status line, you could put anything. You can completely customize this. So you can say show the weather in San Francisco. Show the current model I'm using. Show which branch um on. I'd say those two are probably the most important ones you definitely want to add, which is which model you're using and which branch you're on. Being able to see which branch I'm on at all times has been extremely helpful for me. But you literally can add anything else you want. So you can also say things like how many messages I've sent this session. So you can track how long your session's been. And then if you want to set up the Pomodoro timer I showed you here, then you can say also a Pomodoro timer that I can start and stop through slash commands. And then if you hit enter on that, Claude Code will set up all those status lines for you so that it sits under your chat bar at all times. I love the Pomodoro timer. It's kept me so focused. But you can do anything you want. I've even seen someone say also set up a Tamagotchi and it actually put like an AI buddy under the status bar that you had to constantly feed and interact with also, which might be a little distracting, but it just completely changes your experience with Cloud Code. It makes it so much more enjoyable and making your cloud code experience enjoyable is going to be something that gets you to use it way more, which is going to get you to build a lot cooler apps. Definitely set up your status line today. Feel free to add the things I showed you, but also add anything else you want. If you want my command and how I use status lines, I'll put that down below so you can just copy and if you want to set the exact same thing I did, but feel free to customize it any way you want. The next tip is probably the simplest one, but it might be the most important. It's going to be a really awesome way to make sure your code stays secure in Cloud Code.

So, take a look at this. We're going to go back in here. We're going to clear this message out. I'm going to show you a brand new slash command they just added. So, check out this brand new slash command. If you come in here and you do slash security review, you'll see this brand new default slash command that's in here called security review. You hit enter on that and what it's going to do is spin up a sub agent that now will do a full security review of your entire application by default. And Claude set this up so you know it's going to be the highest quality sub agent with the best commands in it. Now it's going to go through and it's going to search through my entire codebase and check all the changes in my GitHub as well and see if there's any security issues at all that I might not have thought about. This has been a major issue for a lot of people that have been vibe coding for their first time is they don't understand best practices when it comes to security for building applications. And there are so many ways you can screw up security in applications. You can lose API keys. You can put vulnerabilities all over the front end. There's so many different things you can do that will degrade your security. And so what Claude has done, which I think this is awesome, is built in this sub agent, the security review sub agent that will now by default check over your application and find any sort of security breaches. So this is interesting. It actually has found vulnerabilities in the app I'm building and is now going back to validate which one of these were false positives. So it's doing its it's reviewing its own work to find those. So here's the deal. Here's how I'd use this. I wouldn't use this after every single prompt you do, but I would use this before anytime you push code to production. You use a security review sub agent that's now built in. You make sure everything in your code is kosher, then you can push it to GitHub and push it to production. But this is really awesome that they built this in like this. The next hidden trick I'm going to give you is awesome for giving Claude context that's going to make it way, way more powerful, and that's a new way you can use Claude rules. Check this out.

So right now obviously if you create a claude. md file in your main directory it'll create a claude rules file that gives context to claude on everything going on in your application which is good. But say for a second you wanted to go more in depth and you wanted to give even deeper context about how specific parts of the application work. If you put it all in your normal claw. md file, you're adding a ton of context that will make your prompt slower, that might make your output worse. So, here's a new way to give really in-depth details to Claude without clogging up your entire claude. md file. What you actually can do now is go in here and you actually can add multiple Claude MD files to different directories. So, for instance, I'm building out a 3D simulator right now. So, I have a virtual room in this 3D simulator. And say I wanted to add new context and rules just to this virtual 3D room that Claude only knows about when it works on this specific component. I can rightclick on virtual room, click new file, say claude. md like I did in other parts of my application. And now the rules I put in here will only be read when Claude does work on the virtual room component. And this is huge, right? This is huge because before you just have to put all your rules in your normal claw. md file that's at the top of your directory. And that adds tons of needless context that can actually make your outputs way worse because there's just too much context and it's going to use your window up too fast. But now instead of putting all of your rules in one file at the top of your directory, you can go in here now and create a claw. md file for every single component. So maybe you have a backend directory that has your database in it. Now you can go in and you can have a rules file just for your database and just for specific components which frees up your context so Claude doesn't use up as much of your context window. But two make sure Claude understands every single component in your app so it never loses that context. Really easy way to set this up is you can just go in instead of having to manually go one by one and set up those clawed rules. You can just go in and say, "I just created a cloud. MD file in the virtual room directory. Please go in and fill it out for me and hit enter. " And instead of you having to go and make all your own rules for you, you just have your AI agent go in and do it for you and save you a ton of work, right? And so now it's going to go in read all the code inside of virtual room and create the entire rules directory for me. So Claude always has that context. This is going to make it so you don't use up as much of your Claude context window. And that's going to make it so you get way better results. And see, boom, there you go. It has all the rules uh on the virtual room in here. Look, it has uh about a 100 lines of rules that I got done in about 3 seconds. So I asked Claw to do itself. Amazing. I got another one for you. So just to recap what we did so far, we made it so that the model picking is dynamic. you're always using the right model and you're saving on cost. We set up a custom status line for you that always gives you the information you need when you need it. We went over an automated security review that's going to keep your code super secure at all times. And we went over custom clawed rules files that will make it so you don't use up your context window quite as fast so you get way better results from claude code. But let's go over one last pro tip here, which is going to be background tasks. You now can run background tasks in Claude Code that will make it so you can run multiple processes at once and makes Claude code so much smarter. Watch this.

So I'm back in Claude and I'm going to ask Claude to start the development server. So we're going to go in say please start the development server. And typically what would happen is Claude will start the development server and then it'll get tied up. You can't really do anything with claude code while it's running the server through claude code. So I'm going to hit enter on this. But now what's going to happen is I'm going to be able to hit controlB and run it in the background. So if I hit controlB here, this is now running in the background. I can now do anything I want while it's running and Claude's going to have full context of everything going on in the background. So, here we are in my 3D room operating system simulator where I can go in here. I have like my tasks on the wall. I have this computer I can use here. If I want to use this operating system, say I'm in here, I'm using around maybe I'm creating a new note and something crashes. So, here. I hit save and something crashes. What I'm able to do is I can actually go back, go back into Claude, come in here and say, "Hey, please debug what just happened. " and I can hit enter. And now Claude will be able to read all the logs and understand what errors occurred. Before you'd have to go into the logs yourself, highlight, copy and paste a 100 different lines, paste it in, say, "Hey, what's going on here? " But now Claude can work since it's running your server in the background and it can read the logs as it goes, detect crashes as they happen, and read the logs for you. So, no more copying and pasting all the errors as you go. It's just hooked into everything that's happening already. This is going to save you so much time. Any process you typically ran through the terminal before that you had to wait for and watch, you can now send to the background through clawed code and it will monitor it for you, run it for you, and you won't have to worry about again. You can keep your workflow moving seamlessly. This is really, really helpful. Will save you a ton of time. So, we went through five new features that Claude Code just released that they don't really document anywhere. To be honest, I think the status line one's my favorite because that Pomodoro timer has kept me so focused since I started using. I really recommend everyone go in and use it. All the prompts I used are down below so you can steal them, implement them yourself. If you found this helpful, make sure to leave a like. Let me know in the replies which of these were the most helpful for you. Make sure to also subscribe. All I do is create awesome videos about AI that make it way easier to vibe code. and I'll see you in the next