0:00 If you use cloud code, then you know that the plug-in and add-on ecosystem is actually insane right now. Whether it's new MCPs, new CLIs, new skills, new frameworks, there is something new coming out every single day. And it feels like it's impossible to separate the wheat from the chaff. What should you actually be using to improve your productivity? So, in this video, I'm going to do my best to help you out by showing you my top 10 cloud code plugins. And this ranges from skills to CLI tools. And these are things I actually use and I believe they're a true value ad. Now, for all 10 of these tools, I'm going to give you four things. I'm going to tell you why you should care about this tool, how to use it, how to install it, and an actual use case. So, some of these tools may already be familiar to you, and a couple of them we've already done full deep
0:46 dive videos on, but my hope is that at least a few of them are brand new. And if you add at least one new tool to your stack, then I think this video is going to be worth your time. So to kick us off, tool number one is the Superbase MCP server. Now why should you care about the Superbase MCP server? There are two reasons. Number one is databases and number two is authentication. At some point in your building journey, you're going to create a project that requires databases and requires authentication. Superbase can handle both of those things for us and they have a very generous free tier. So you can actually go pretty far without paying a penny. The other cool thing about Superbase is it's built on open-source tools. Superbase is just a nice wrapper around it that's easy for us to interact with, but we can also self-host and develop locally this
1:32 entire like authentication and database framework. Now, to install the Superbase MCP is actually very simple. It is just this one line of code. If you're looking to where I found this, this is inside the cloud code documentation under MCPS. Now, as for use cases, I kind of already alluded to this. anytime you need a database even if it's a basic one like a just a standard Postgress database or we're trying to do some vector database like rag we can use this tool also like I said authentication like are you going to have users visit your site do they need to build a profile do we need to keep their data separate again perfect use case for superbase now to actually use this inside of cloud code again just like any other MCP after you run that installation command we're just going to do for/mcp and you will see your superbase mcp show up down here inside of projects you will need to authenticate at first which requires you
2:18 to hop into Superbase and get some API keys. But again, as always, Cloud Code will walk you through that process step by step. And once connected, when it comes to creating the databases or editing the databases, we no longer have to hop into Superbase and like do SQL code. We just use plain language. So, it's great. Now, I'm going to be real with you for a second. You should actually never use the Superbase MCP server in reality. There's no point. There's no point because we should be using the Superbase CLI instead. Well, Chase, why did you just talk about the MCP server for a minute and a half and waste all our time? Well, two reasons. One, to tell you this story. I myself, someone who was looking at these tools all day, all the time, had been using the Superbase MCP up until very, very recently. In fact, it wasn't until this
3:04 comment above that was calling me out in another video did I realize that the Superbase CLI actually existed, which is kind of wild because I use Superbase all the time. I understand that CLI tools are in general superior to MCP tools. Yet me who again this is like my profession didn't even realize that. And I just wanted to show you that like as someone who is very deep in the space even I struggle to keep up with these tools even when it comes to tools I use consistently. So, if you feel like you're overwhelmed, I promise you're okay. The second reason I wanted to do this is because I feel like it's worth having a discussion about CLI versus MCPs because in the next nine tools, we do have a couple MCPS, but for the most part, we always, always, always try to use CLIs instead of MCPS. Now, MCPs, as
3:49 you know, these came out in 2024. These are ananthropic creation, right? It was a standard for us to connect our AI assistants to different systems where our data lives, like Superbase. However, today in the big 26, our AI assistants, they live inside the terminal with us. So, they have the ability to use these CLI systems. And we like the CLI more than MCP cuz MCP is pretty much just trying to ape what a CLI tool does. But because it's not in the terminal, there's some additional overhead that goes along with it. So, all things being equal, if you are looking at something and there's a CLI and there's an MCP version, go with the CLI. It's essentially purpose-built for AI coding agents like Claude code. Now, when it comes to the CLI, the Y is the same, authentication databases. The how is
4:35 pretty much the same as well. The biggest difference is just the installation. So, I'm inside of the Superbase docs right now. And the installation just changes depending on what operating system you're using. However, Cloud Code can again do this all for you. Just say, "Hey, download and install the Superbase CLI tool." And it will download and install that dependency. And it now has access to the Superbase CLI. Now, the one caveat when it comes to using CLI tools versus MCPs is that when we give Claude Code some sort of brand new CLI tool, we usually want to include the addition of some sort of appropriate skill, right? We want to tell Claude Code how to use that specific CLI tool in a specific way. And the perfect way to do that is with skills. So, anytime you install a new CLI, make sure you check their GitHub or
5:20 check their documentation or check somewhere to see, hey, do they already have purpose purpose-built skills for this tool and Superbase has that for us right here, Claude Code plugin? We can just look it up inside of the marketplace. So, that's tool number one, Superbase CLI, not Superbase MCP. Sorry for the bait and switch, but I wanted to be able to have that discussion about CLI versus MCPS because I think it's an important one moving forward. Now, before we move on to tool number two, a quick word from our sponsor, which is me. I just released my Claude Code Masterass inside of Chase AI Plus. There is a link to that in the pin comments. And if you're looking to go from zero to AI dev, no matter your technical background or lack thereof, that's the place for you. I also have a free Chase AI community that is in the description. Tons of free resources, which is perfect if you're just getting started. So, make
6:07 sure to check it out. Now, tool number two is, in my opinion, the most powerful out of all 10 we're going over today, and that is the skill creator skill from Anthropic. We did a full deep dive on this guy a few days ago. I'll put a link above, but this one is amazing. Not just because it can create skills, but because we can modify and improve existing skills and measure skill performance. Now, Enthropic put out an entire blog talking about the improvements it made to this particular skill. And I really can't emphasize enough how big of a deal this is because first of all, we can create custom skills that we know are working. Before this skill, it was kind of just random, right? Did we know the skill was actually making things better? Maybe. But did you have any actual data? Well, if we look at stuff like this, we now do. I can do AB tests with the skill,
6:53 without the skill. I can do AB tests to existing skills, right? Does my current incarnation of this skill actually perform better than the previous version before I made the modifications? Right? Before you were kind of in the dark or it was kind of wishy-washy and it was kind of just based on gut feeling. Not the case anymore. And that is extremely important when it comes to improving cla. Now to use the skill creator, we first need to install it, which is very easy. You're just going to do for/plugin and you are going to find it inside of the plugins. You can just search for skill creator, hit install, and then you'll see it here. It should say skill-creator. Once installed, there's two ways to use it. Either we can just do forward slashkillcreator and then I know it's going to be invoked. Or you can just use natural language and say, "Hey, cloud code, use the skill creator and I want to either a
7:40 create a new skill or b I want to go ahead and do some eval on this current skill." It's that easy. And like any skill under the hood, it's just a text prompt. So if you go to Anthropics GitHub and Claude Code and look at skills, you can see the entire thing in all its glory right here. Now tool number three is actually a framework for a cloud code and it's GSD the get done framework. This is essentially an orchestration layer that sits on top of cloud code and sort of just changes how cloud code operates when it comes to creating new projects. And essentially what it does is it just gives us more guard rails in terms of specd driven development which makes it perfect if you're trying to create a brand new project from scratch and you want it broken down phase by phase, feature by feature. And on top of that, it also does a really good job at context rot
8:27 management, context window management. And it does that by ensuring every time we're executing a new portion of the plan, we're doing a fresh context window. We're using sub agents. Everything is done in a way to make sure we don't get into the back half of our window and get worse outputs because of that. So, if you're looking for something that's going to create your projects from scratch in an extremely regimented way, I can't recommend GSD enough. And to install it is really simple. Just search up GSD or get done on GitHub and it's simply this single command. And once you install it to start a new project, it's all through slash command. So you just do forward/gsd new project and then what it will kick off is essentially a plan mode on steroids where it really really really tries to nail down what it is you're actually trying to build which
9:13 you know if you've built anything you know how important it is to essentially like plan 300 times with cloud code once to get quality outputs. Now tool number four is the backbone of my research workflows inside of cloud code these days and that is the notebook LM-PI skill/ CLI tool and that is the tool that allows us to connect notebook LM in claude code. Everything I can do inside of Notebook LM I can now do via the Claude Code terminal even though there is no official API for Notebook LM which means research, analysis, creation of deliverables like videos, infographics, slide decks, flashcards, podcasts, whatever is offloaded to notebook LM for free pretty much because of this tool. So it's a wild value play. Now to install this is very simple. We're just going to head to the notebook LM-PI
9:58 GitHub, copy these commands, paste them into our terminal, and install the dependencies. Once you install those dependencies, then we need to log in. So, you'll just run notebook LM in your terminal. It will pop up a browser, and then you just log into your notebook LM account. Now, like we talked about before, this is a CLI tool, which means what? We need to add a skill. The skill install is here as well. It's just notebook LM skill install, and that teaches Cloud Code how to use all of this. And these are the actual commands Claude Code will be using, but you just talk to Cloud Code in plain language. And again, if you can do it in Notebook LM, you can do it in Claude Code via the terminal. Just use natural language. So, this one is one of my favorites, definitely add this to your stack. Now, tool number five is perfect if you use Claude Code as a personal assistant in any capacity, and that is Obsidian. Now,
10:45 Obsidian is great for you, the end user, because it lets you get more insight into your text files by seeing how they connect, and it just in general keeps you infinitely more organized, which is why I specifically said that the use case here with Cloud Code in Obsidian is in a personal assistant type context. Somewhere, like you see over here on the left, you have tons of markdown files. This isn't a huge value ad if you're doing this in a typical coding project because that sort of context doesn't make sense. But in a in a personal assistant world where you have vast sprawling and accumulating context in the form of text documents, markdown files, Claude Code plus Obsidian is great. And there's no CLI we need to install. You don't even really need a skill. You just need to download Obsidian, create a folder that is designated as the vault, and then you just open up Claude Code inside the terminal in that particular folder. And
11:30 for me, that folder is quite literally called the vault. And then after that, all you have to do is tell Claude Code, hey, when we create markdown files, just follow Obsidian conventions. That's it. Again, the value ad for this connection is in specific use cases. This isn't a one-sizefits-all. Now, tool number six is all about making it easier for us to manage our deployments, and that is the Verscell CLI. Now, I love using Verscell for my deployments because it's simple to use and they have a great free tier. So, what more do you need to install the CLI? Very simple. We're just going to give it this command. And also remember for all these install things, you can also just ask Cloud Code to, hey, go ahead and install the Versel CLI for me. And you know the drill at this point. We're adding a CLI tool. So, what do we need? We need that skill. Verscell gives it to us on their actual website. And in fact, they actually have a ton of skills available if you ever want to take a look. Some of them are pretty
12:16 interesting. But if we just click that, it will bring us to the skill. And this is the command we want to put in the terminal to install it. And this is the sort of tool that's great to use in conjunction with like the brand new agent loops. You know, you can have a loop that calls the Versell CLI skill to like check deployment statuses if you're constantly putting up new deployments as you make changes to a particular project. Now, tool number seven is all about enhancing Claude Code's browser automation ability. It's ability to actually go out on the web and do things like even something as stupid as like, I want you to go out on Amazon, fill up my shopping cart, and then buy it, right? We do that using the Playright CLI. Now, Playright is an open- source tool from Microsoft. It's very popular. I've really enjoyed using it when it comes to browser automations and cloud. And the install is very easy. We're just going to head to the Playright CLI GitHub and
13:02 we're just going to copy this command, paste it in our terminal, and it's going to load the dependencies. Installing the skills is also very simple. Playright CLI install-skills. That's it. Installs done. Now, as for use cases and demos, we did a full deep dive on Playright in Cloud Code. So, make sure you check out that one. Just came out a few days ago. But anything you could essentially normally do in a browser, you could do it using Playright. And it's also great for things like testing UI design or testing like form submissions, anything like you know that's kind of laborious when you're changing something inside your project or your app and you need to manually test it. Well, Playright can do that for you. And just like all these CLI tools, we just use plain language to execute the task. Now, one thing I didn't mention in that deep dive is the Playright CLI show command. And this actually gives you a visual dashboard that shows what all of your browser
13:48 agents are doing, even if they're running headless. So, really cool thing to use. Now, tool number eight works with a platform that should be near and dear to your heart, and that is the GitHub CLI. So, with the GitHub CLI, we can essentially do everything you were doing manually on GitHub via the CLI. This is obviously great to use in tandem with the Versell CLI tool. And so, we can essentially do everything from coding to pushing to GitHub to deployment, again, all from one place using natural language. Now, as for installation, obviously, we can lean on Cloud Code to walk us through it. But if we go to this GitHub page, it breaks it down by operating system. So there shouldn't really be any confusion. Now coding agents like cloud code are so spun up on Git and GitHub that you really don't need a skill to use this to its full potential. But if you feel so inclined, I would just use the skill creator to create one yourself. But
14:35 honestly, just having the tool installed and talking to cloud code in plain language is enough. Now tool number nine is all about web scraping. And for that we are going to be using firecrawl, both its CLI and its appropriate skill. Now to install the Firecrol CLI, it's super easy. Just search up Firecrawl CLI tool. you'll get to this page. It's this one line. This will also include the appropriate skill install. Now, if you've never used Firecrawl, the pitch for Firecrawl is that it's web scraping, but the data it brings back is tailor made for AI agents. And there's really four different commands when it comes to firecrawl. It's scrape, crawl, map, and search. Now, if you've never used Firecrawl before and you're kind of new to web scraping in general, then these four core commands kind of might seem like they're all doing the same thing. But luckily for us, you know who knows how to use the right command for the right job? Cloud code. Hence the skill
15:21 and fire call is also nice enough to give us some of the standard use cases things like competitor research documentation site change monitoring or deep research with sources. So I really love firecol. It's definitely a step above than just asking cloud code to use its web search tool. And last but not least tool number 10 is one of my personal favorites and that is the excaladraw diagram skill. I just recently started using this and shout out to its creator Cole Med. You've probably seen his channel on YouTube. One of the greats in this space. love the work he does, but I love his skill even more. So, what does this do? This allows me to create diagrams through natural language with Claude code. If you watched my last video on the six levels of claude code, you saw these diagrams. Like all of this, I didn't do any of this manually. All of these things were created using cla code,
16:07 which saves me a ton of time and actually gave me something to speak to. Like this is actually an awesome thing if you ever do any sort of presentations or you've ever worked in Excal before because again doing this by hand brutal would be totally brutal but this skill gets rid of all that manual work. Now to use this you're just going to head to the Excal diagram skill GitHub and then you're going to clone the repo and then you're going to copy it into your project skill directory. And again just like everything else to use it we just talk to Claude code in natural language. The best way to use this I have found is again using these diagrams is sort of an add-on to a presentation or something but point claw code it's some sort of you know directory of knowledge right whatever you're doing for in this case it was like a whole write up on what I was going to talk about my six levels of
16:53 cloud code and I was like hey take all that information break down each level into some sort of chart people can you know see and then do it right obviously I put in plan mode had some back and forth so I had some idea what I was going to come up with but This thing is awesome. I I can't praise this um directory enough. So yeah, shout out Cole. So that does it for my list of my top 10 favorite Claude Code plugins. I hope at least one of those was new for you and at least one of those can be added to your stack and kind of help you out in your day-to-day because there is so much out there to improve your Claude code quality of life. But again, we don't want to fall into the trap of thinking increased capability equals increased performance, right? There is a sweet spot somewhere. there is that middle ground that we're always searching for and hopefully we're
17:39 getting you closer to that place. So, as always, let me know what you thought. Make sure to check out Chaseai Plus.
