0:00 If you've been paying attention to the cloud code scene at all lately, then you've noticed a dramatic shift from MCP tools to command line interfaces. And the reason for that is pretty simple. It's a huge performance increase. Cloud code already lives in the terminal. The CLI lives in the terminal. There's no middleman, no overhead, no config, nothing stopping your agent from communicating with that program. So if it's clear that this is the future we are marching towards with Agentic coding agents, our next question should be, is there any way for us to speed this up? Is there a way for us to create CLI tools that interact with software we are already using? Well, luckily for us, as of a few days ago, the answer is yes, we can. And that is thanks to a brand new open source tool called CLI Anything. Now, this thing was created by the data
0:46 and intelligent labs at Hong Kong University. If you don't know who those guys are, they're the makers of Lightra, Rag Anything, Nanobot, and a bunch of other open source tools. These are one of the titans of the AI open source community. So, when they build something like this, it's worth our time to take a look. So, long story short, what does CLI anything do for us? Well, it allows us to point CLI anything at any open-source software or something you have built and create a CLI tool for that that something like cloud code can use. And that's echoed right here. Every software becomes instantly agent controllable through structured CLI. Agents control any applications without APIs, GUI, rebuilding or complex rappers. And ultimately, we're transforming human design software into
1:32 agent native tools with one command. Now, again, what does that really mean? That means we can have cloud code control software that normally you as the human would be controlling through some sort of graphical interface. Remember the Google Workspace CLI we just covered yesterday. Link to that video above. Same exact idea. Normally with Google Workspace, you know, I'm talking or Gmail, you would go to Gmail, you would click through the things, you would manually do these things. But since there's a CLI, you can now do all that through the terminal. This is allowing us to do that for anything, right? Not just Google, not just whatever. If it's open source, we can point this at that software and create a CLI for it. That's a massive, massive productivity boost. So, what's going to happen now is I'm going to show you how CLI anything is working under the hood. And then we're going to go through a demo of actually pointing this tool at
2:19 some open-source software, hint, it's draw.io, and creating a CLI tool for that application and then controlling it via Cloud Code. But before we do that, a quick word from our sponsor, me. So, I just released my Claude Code master class inside of Chase AI Plus, where I take you from zero to AI dev, no matter your technical background or lack thereof. There's a link to that in the pinned comment. And if you're someone who's just getting started with their AI journey, isn't quite ready for a master class, I have a free Chase AAI community in the description, so there's something for everyone. So, make sure you check that out. All right, back to CLI Anything. How is this thing actually working? Well, it essentially has a sevenstep process. Once we install the CLI, and I'll show you how to do that, and we point it at whatever software we want to create a CLI tool for, it's
3:06 going to run this pipeline of seven steps. And it does this all automatically. You don't have to touch anything. First, it analyzes the code. So, this is why it needs to be open source. Like, it needs to actually see the guts of whatever application we're looking at. Next, it does design, right? Takes a look at the output format, implementation, then it plans, tests, writes tests, documents it all, and then publishes it. And then after all that, we have the step to actually refine and improve the CLI if after the first pass it isn't exactly what we were looking for. So they give us a lot of things related to use cases in this repo. And I'll put a link to that down below as well. But a cool thing is is they've already done this for a number of large open- source applications. So Blender, Inkspace, Audacity, on and on they go. So they actually have already done this for real. And you can kind of
3:52 see the actual CLI tool. And when you download this CLI anything tool, it comes with all these CLI commands essentially like pre-loaded. And you can also see how many tests they ran for each. So for example, something big like what Blender was 28 tests. So it's pretty robust. And they've obviously proven this model on some of the big players in the open source community. And for today's demo, we're essentially going to redo what they did with draw.io. Again, they have the pre-built CLI ready to go, but we're going to do it from scratch to sort of prove how this works. Now, to get this started is pretty easy. We just need Python 3.1 and the target software installed. Since we are going to be using draw.io, what we're going to do is we're going to install anything CLI, that command line interface, and then we're going to clone the open-source GitHub repo of draw.io.
4:38 And then from there, we point the CLI at the cloned repo, and it creates the CLI tool. Okay, so now what we want to do is we want to head into our terminal, create a new directory for this. So I'm in the CLI- anything-demo directory. get Cloud Code up and running and we're going to install the CLI everything plugin. Now, the commands are right here inside of the GitHub. So, first you need to add the CLI anything uh tool to the marketplace and then we're going to install it. So, just copy and paste these two commands. You'll know you've done that correctly if you do /plugin and you go to installed and you see the CLI anything plugin here and enabled. Also, remember when you install the plugin, you need to cycle cloud code to get it running or do for/reload plugins. Next, we want to build the CLI. But to build the CLI, we need to point it at
5:23 something. So, what we're going to do now is we're going to clone that draw repo. So, I told Cloud Code, go ahead and get clone this repo. I'll put a link to this repo down below. It's just the open source repo. It's going to go ahead and copy that entire repository and put it in your directory. Once you do that, we're just going to have one more command. It's just CLI anything. CLI anything. And then we point to wherever that repo is. And so, for us, our command will look exactly like this. You just hit enter and it's going to start building the CLI tool. And you can expect it to take about 20 minutes or so for something of this size. All right. Once it's complete, you should see a message that says something like all nine success criteria are met. It'll give you the architecture breakdown as well as the test results and finally the key capabilities. Now remember, you aren't stuck here. If you want to continue to iterate and add more
6:09 features, super easy to do. So let's actually demo this now. So the prompt I'm giving Claude code is let's test out the new draw CLI tool. Create me a thorough yet clear diagram explaining a typical SAS backend architecture. Make it visually striking. Pull up the draw diagram for me at the end. So you can see it's taking a look at the draw IO package and you will also see it referencing this agent harness. That's the nomenclature it often uses for this CLI wrapper or CLI harness. So here's what it built us this diagram for SAS backend architecture, right? Pretty solid actually. There's a lot going on here. Um, even did like cool little like little shading behind this thing. Now, the point of this video isn't like, oh, look how cool this draw IO thing is. The point of this video is that we just created this CLI tool that did not exist, you know, 10 minutes ago and had
6:55 it create this fairly complicated and welllaid out diagram with a single text prompt. So, think about what sort of workflows you're doing in your day-to-day with open source projects or really projects that you have access to. their back end in full that we could point CLI everything at and create a CLI tool for that software. The implications of something like this is pretty crazy. And as you saw, like that wasn't hard. It was like one line of code. Like we clone the repo, ran the code, done. This is as easy as it gets, guys. So, I highly suggest you try this out yourself. Download it, install it, point it at some other open source project, and see what kind of results you get. Like I said, this is so easy to do, and this is where the field is going. So, as always, let me know what you thought in
7:41 the comments. Make sure to check out Chase AI Plus if you want to get your hands on that Cloud Code master class.
