Transcript: Every Claude Code User NEEDS To Watch This

Source video ID: 8kWONfT_-H8

Transcript

• 0:00 — You need to be tool agnostic or you’re going to get screwed. A perfect illustration of this just occurred a few days ago when Anthropic came out and said they’re going to change how they bill programmatic Claude code usage. And if you’re all in on a single vendor, changes like this can have significant negative effects to your wallet. So in order to prevent you from becoming a prisoner to a single tool, we need to learn how to diversify. And that’s exactly what we’re going to do today as I show you how to introduce Codex into
• 0:32 — your stack. And the best part about using Codex isn’t even the fact that GPT 5.5 is really strong or that the desktop app is extremely intuitive. It’s the fact that Claude code and Codex, the Venn diagram of these two tools, is essentially a circle. Once you’ve mastered one of them, you’ve pretty much mastered the other. So if you’re an experienced Claude code user, you’re in a great position to start leveraging Codex. And in this video, I’m going to give you a speed run of what you need to know so you can do it as efficiently as possible. So we got two versions of Codex to think about. We have the Codex
• 1:03 — desktop app and we have the Codex CLI. If you’re someone who wants to use the Codex CLI purely through the terminal, it’s extremely easy to set up. We’re just running a few lines of code. And then from there, I’m not really joking, you just treat it like you would Claude code. That whole Venn diagram being a circle totally applies here. Now, there are some subtle differences. It’s not completely the same. But in this video, we are going to be taking a look at those differences through the lens of the Codex desktop application. Now, why the desktop application? Because I like it. Super intuitive, it’s nice to use,
• 1:34 — and it’s easy to sort of explain all the differences through that lens. Also, I say that as someone who uses Claude code purely through the terminal. I like using Codex on the desktop app. I kind of just like mixing it up. And as you’ll see, you can also use the terminal inside of the Codex application, which is a great way to use both Claude code and Codex. So understand everything we talk about, even though it’s on the desktop app, also applies to the CLI. Now, before we install and get it running, let’s talk about Codex pricing very quick. Cuz this is a place where it wins versus Claude
• 2:04 — Code, the OpenAI sort of philosophy on usage is much less draconian than Anthropic’s. So, you’re going to want to be on the plus or pro. The difference is the $20 a month plan versus the $100 a month plan, and they also have a 20X pro plan similar to Claude Max. The big difference is if I’m on the pro plan at $100 versus the plus plan at 20, I get access to GPT 5.5 Pro. So, there’s GPT 5.5 and there’s GPT 5.5 Pro. 5.5 Pro is
• 2:36 — better, obviously. The base level 5.5 is still really good. So, if you’re someone who’s already paying $200 a month to Anthropic and you’re like, I kind of just want to try Codex, I don’t want to go nuts with it off the bat, just do plus. Just do the 20 bucks a month on 5.5. Your usage is still really good because the usage at $20 a month on OpenAIs better than $20 a month in Anthropic. There’s not like a great way to do like a one-for-one test, but 5.5 also just uses less tokens in general. So, that’s
• 3:06 — my take on it. You’re going to get more usage, more bang for your buck. So, to get it started, you’re just going to go to openai.com/codex, download and install it. When you open up the Codex desktop app, it’s going to look something like this. Very similar to ChatGPT and also pretty similar to the Claude Code desktop app. We have our prompt window. We have the ability to add photos and files, plugins, plan mode. We have sort of our permissions set up similar to how we set permissions inside of Claude Code. I can set my model, the speed, and the intelligence.
• 3:36 — And we’re pretty much ready to go. From here, I can also select where I want my project to live when I’m working on. Now, before we demo it in action, let’s take a look at some of the settings. So, if I come down to the bottom left and I head to settings, so the first tab is general. You’re going to want to be on coding to get more technical responses, and then it’s going to ask you about permissions. So, this is similar to sort of the permissions with Cloud Code, what it can edit and what it can edit without actually asking you. For me, I enable all of these because I run on full access the entire time. Beyond that, it’s sort of just asking you stuff about
• 4:06 — your environment and what sort of speed and all that you want. So, for speed, understand if you go to fast, it’s charging you 1.5 times the tokens. Another interesting thing is follow-up behavior, queue versus steer. So, you’re going to want to leave it in queue for default. That’s saying, “Hey, if Codex is in the middle of some sort of tool call, it’s doing something, and I give it another prompt, if it’s in queue mode, it’s going to wait to finish everything before it hits my next prompt. If it’s on steer mode, then whenever I give it a follow-on
• 4:36 — prompt while it’s doing something, after it finishes its tool call in the middle of it. So, maybe it’s doing 20 tool calls. At tool call five, I add another prompt. It’s then going to inject that prompt. So, sometimes you want that, and as you’ll see, we can manually press steer whenever we give a follow-on prompt, but that probably isn’t how you want it to default. At least I don’t. So, that’s what that is. Next up, we got appearance, so you can kind of adjust this how you want. You also have pets. Now, I kind of like pets. Like, I made my own like cool little custom pet. The way this works is if I’m working inside
• 5:07 — of Codex and I give it a prompt, this acts is like a little hook, and it shows me what it’s working on. It gives me a little like visual indicator that it’s still working on something, so I can tab out, go do different things, and this little like little pet guy lets me know when it’s done as well. So, you’ll see that in action. Some people don’t care for it. I think it’s kind of cool. Then we have configuration. So, user config is similar to sort of the permissions we talked about earlier. When does it need approval to do things? This is you sort of setting the global values for that. You’re also going to want to enable Codex dependencies, and
• 5:38 — we have the user config. So, if I open config.toml, what we’re looking at here is just an internal Codex settings file that says some of the like plugins and things you have installed. One thing you probably will want to add to this is the goals feature. So at the very bottom if you add a line and you manually put in feature and then goals equals true, that will enable the goals feature inside of Codex. I have a whole video on that, I’ll link that above, but it’s a it’s a nice way to use Codex as a long-running agent to harness. Next, we have personalization. This isn’t an agents.md, this isn’t a claude.md thing.
• 6:08 — This is sort of just giving it a quote-unquote personality. How do you want it to actually talk to you? And then you have memories as well. Memory inside of Codex, same way as memory inside of Claude code. If I say, “Hey, every Tuesday I go to the gym.” It’s going to be like, “Oh, yeah, let’s write that down.” So it’s kind of just like, you know, I’m not a huge fan of these like goofy memory things, but if you want it, you can add it. Beyond that, the rest of this has to do with just like hooks and get and environment. It’s relatively self-explanatory. Certain things like computer use, um and you might have another thing called
• 6:40 — I believe it’s called like archiving or something to that effect, are Mac only. So there’s certain things that you can only do when you are using the Mac OS. Now let’s talk about actually using this thing. You kind of have two options. You can either be in a chat or you can be in a project. If I’m in a chat, it’s kind of like I’m inside the Claude desktop app and I’m just on like the Claude chatbot section. Like this isn’t really saving anything into a project. So I can say something like, “Hey, what’s the difference between Claude code and Codex?”
• 7:11 — You know, it’s going to do its thing, but it isn’t working inside a specific that I’ve set. So if you want to that, let’s say you’re actually working on a real coding project, you’re going to want to be in projects. So if I click on project, I can start from scratch or use an existing folder. I almost always do use an existing folder, so I have a little more, you know, granular control of where I’m putting this folder, but let’s do a new folder and we’ll call it CC to Codex.
• 7:43 — Select that folder. And I now have a project over here. And you also see this is saying, “Hey, do you want to migrate some settings?” So, if I hit continue, what it’s doing is it’s importing all my settings from Claude Code into Codex. All my skills, all my plugins, all that stuff is now, just like that, inside of Codex. It’s that easy. And so now, if I’m here, what I can do is I’m now working inside my Claude Code to Codex project. And this is what you want to do if you’re really doing anything of substance. You want to be working in a project.
• 8:13 — So, I can set the folder, I can set where I’m working, whether that’s locally or in a new work tree, or inside of Codex Web. And then I can actually select the branch. So, let’s continue on this. Let’s say, “Hey, can you create me an Excel document that lists out the top five differences between Claude Code and Codex.” So, we wanted to create this Excel document. Well, I want it to also use the Excel plugin. So, if I do the at symbol, you can see up here, I have a number of plugins, and they’re spreadsheets. So,
• 8:45 — if I do that, it’s going to get to work. Now, what are plugins? What are spreadsheets? What are all these things? So, plugins are just like plugins inside of Claude Code. So, if I come up here to the top left, I have a series of plugins. And what’s the difference between plugins and skills? Kind of a blurry line. Plugins tend to just be like kind of official skills that have a little more to them. So, you can see some of the ones I’ve already enabled up here, things like spreadsheets, and presentations, and Chrome. But you can add a number of things. So, I’ve added Vercel, I’ve
• 9:15 — added Supabase. And if you want to add any of these things, literally all you do is just hit the plus button, and it installs it. You usually have to log in to that system. It’s very, very simple. Up here, we also have skills. So, here we are inside of skills. If I click on one of the skills that are installed, like the plugin creator, it brings up exactly what’s going on inside there. I can enable it, disable it, try it, uninstall it, all that. If I come up the right, I can easily create a skill or a plugin, so it brings up the skill creator skill. And so, it’s pretty intuitive to sort of
• 9:45 — control all these things in a visual format. Now, taking a look at our project, you can see our chat over here that we were working on earlier. I can pin it, I can rename it, delete it, all that, or I can click back into it. So, I’m back inside our Claude Code to Codex chat, and we can see it working right here. And while this is working, let’s talk about automations. Automations, just like with skills, relatively easy to set up. It has a bunch of templates you can work with, but I can also just hit new automation
• 10:15 — up top. I can describe whatever automation I want to do, give it a title, auto one, put the work tree, the project, and when I want it to run. I can also set what model I want to use, as well as the reasoning. That easy. And like you kind of saw me do earlier, if I ever want to use a particular skill or plugin, Codex is smart enough to know when I say, “Hey, can you create me an Excel document?” to use the spreadsheets plugin, I just explicitly call it. So, if I want to explicitly call a skill or a plugin, I
• 10:47 — just use the at sign. So, you can see all the plugins. I can also do forward slash stuff, so if I do like forward slash front end design, you know, you can see the front end design skill there. So, again, very, very similar to how you would use Claude Code in the terminal. So, finished up its Excel write up, and I can actually view the Excel thing inside of Codex itself. And then, if I actually want to see it inside the folder, I can just click right here, go to open in folder, and here’s the actual Excel file opened up in Excel itself.
• 11:18 — So, again, really intuitive. Then, we can do something like, let’s create a landing page for a website that talks about exactly what’s in this Excel file. Go ahead and use the front end design tool to help you create that. So, now it’s going to use this Excel file as sort of its inspiration to create a landing page that talks about the differences between Claude Code and Codex. And I will show you how we can also have it populate inside this sort of like sidebar, and we can leave comments and things like that. That again makes it very easy to do
• 11:48 — things like front-end design from the Codex desktop app. And one of the thing, like you see here, you can very quickly see what sort of changes it makes with the code. You can undo it, you can review it. So, if you review it, it actually brings it up here in the side panel. So, again, so easy to use. Now, remember how I also talked about steer versus queue. So, if I give it a prompt while it’s working saying, “Can we make this in a neo-brutalist style?” You see right now it’s queued, so it’s going to wait till it’s done. If I hit steer right here, it now essentially injects
• 12:18 — that into the current working tool calls that Codex is using. So, you know, it’s doing something. I pretty much tapped it on the shoulder and said, “Hey, by the way, before you finish, make sure you do this.” So, it’s a nice little thing to have. Also, while we’re here, other things to talk about, context. So, you can see right here if I mouse over this window, you have the context window. Uh it’s only 258K tokens on 5.5 Pro versus 1 million on Opus. I would argue that’s not actually a downside when we think about things like context
• 12:48 — rot. And because it’s so small, 258, when it auto compacts, it’s less less likely to have sort of that auto compacting drift. Now, you still can get that if we just let it auto compact again and again and again and again and again, it kind of starts to struggle, right? Like any system does. So, you know, if you’re someone like me who’s very aggressive with their slash clears, there is no way to slash clear here. All you would do is you would just start a new chat inside the same project, right? That’s
• 13:19 — the equivalent of doing a forward slash clear. Again, very easy because all the new chats will show up here inside of the project section. So, overall, in terms of context management, I would argue it’s actually easier than Claude code. So, here’s a look at what it built for us. So, I can go ahead and expand this panel as well. Looks like it took the Excel document and was used that as like a background visual for the website itself. It took sort of this neo-brutalist style and then goes through the top five differences between the two. So, we can see Claude code versus CodeX
• 13:49 — on and on and on. Now, let’s say there were certain things I wanted to change about this. So, what I can do is I can like highlight something and then I can leave a comment, say something like, “Can we change the font color to all black?” And you can see over here, it’s now an annotation. And I can continue to do that over and over and over again. So, once you do one comment, up here this gets changed to annotating and I can go ahead and select any sort of
• 14:20 — component of this webpage and continue to leave more and more comments. Again, compared to working purely inside the terminal, whether that’s with the CodeX CLI or with Claude code, little harder to do this kind of stuff. You can, but requires other IDEs or additional skills and plugins. This is all out of the box. And another cool out of the box feature with CodeX is because it’s part of the OpenAI ecosystem, it has stuff like GPT images, too. So, it can create images for you. We don’t have to hook up anything like a Higgs field CLI like that. So, if I give it a prompt like,
• 14:50 — “Hey, I want you to create images that kind of go along with all the top five differences and have that be like a backdrop.” It could do that. So, it’s pretty flexible. But, for the most part, everything we’ve covered, even though that was relatively quick, like that’s that’s kind of like 99% of it, guys. Like, everything else is just like, you know, AI coding AI coding in general. Um but, if you’re someone coming from Claude code moving to CodeX, you just watched this video, you pretty much got like 95% of it. Um I’m sure I missed one or two little things, but like you see, like it’s not
• 15:21 — a hard transition. And the one other thing I didn’t really mention was like, “Hey, we can also add the terminal here.” So, over here on the top right, I can toggle the terminal and, you know, I can also have Claude Code open in here as well. So, like I talked about, you know, big picture, we don’t want to be just pigeonholed into one tool. Like, I don’t want to just use Code X, just like I don’t want to use Claude Code. It’s really easy to use these two things in tandem with a setup like this, where I have like all the benefits of the Code X desktop with all the benefits of the Claude Code CLI. So,
• 15:52 — highly suggest you guys try this out. Just do it on the $20 a month plan. You don’t use you need to use the 100. Just give it a whirl. If you like it, you like it. If you don’t, hey, no harm, no foul. Like you saw, it’s not some crazy lift, where it’s like you’re learning a who a whole new coding language. Like, you’re not. It’s pretty much the same thing. So, this is where I’m going to leave you guys. Let me know what you thought about in the comments. Uh I also have a link to Chase AI Plus in the pinned comment. That includes not only a Claude Code Masterclass, but a Code X Masterclass, as well. So, if you really want to dive deep into this material, I have stuff
• 16:23 — for that. So, definitely check it out. Other than that, I’ll see you around.