Transcription
You're about to get a free Chat GPT replacement that connects to all your apps, answers all your questions, keeps everything local on your computer, and is so easy even your mom could use it. You're going to see a free way to make any website, even Facebook Marketplace, easy for your AI to interact with. You're going to get a free meeting notetaker from OpenAI that doesn't just take notes, it actually does the work on the call while you're talking. And you're going to see a way to turn Wi-Fi routers into motion sensors for free. All that and so much more.
We're going to go through the top 10 GitHub repos. It's all linked below so you can skip through whatever you want. Let's get into it.
The number one hottest repo of the week on GitHub is Open Human. Here's the deal. We are all installing. Even my mother is installing Chat GPT on her computer. People are more and more installing Claw Desktop on their computers. These things cost money, 20 bucks a month, which you wouldn't think is a lot of money, but I can't imagine how many people complain about freaking 20 bucks a month. And at the same time, here's something else that bothers people: these platforms are training on your information. And so this guy came up with something that is a free alternative. That's what Open Human is. It has been really interesting how it's blowing up. We got uh 25,000 stars. Adam, you've taught me to look at forks. So we're looking at 2,000 plus forks of this software. Here's what it looks like. In fact, I'm also going to say this is Product Hunt Product of the Day. So you can see that it's very consumable. And Stephen, the founder, has said, "I did this so my dad had something that he can use." And so that's what we're looking at here. I have some thoughts on it. What do you think, Adam?
>> I mean, I think you hit it there. It's something his dad could use. Like all these other tools, they're so annoying to set up. There's so much configuration and you've got to figure it out in the command line and the documentation isn't very good. Like what I love about this, click into that one there. The 118 plus apps.
>> Uh this one, yes, it also connects to all these different apps. Look, look at how easy it is to connect to all these different things. Like this is OpenClaw. If you've tried to set it up, it's a nightmare to get connected to everything. This one, this one makes it really, really easy. Nice, free, usable. I do have one thing, one cautionary tale. I like that this person here, Omar Diani, has actually tried it for 5 days and did his write-up. And here's the thing that he said. Look, he's giving it a 72 out of 100 because even though the architecture is compelling, the promise is ambitious. We're looking at beta stage execution. There's still problems with it. It's still not as flawless and as easy to use as let's say Claude. Having said that, for all the people who are cheaper than me, it is free. And for all the people who are looking for something more privacy forward, this is the future. I think Open Human is a really good repo to look at.
Let's look at the next one. Number two for the week is CodeGraph. It maps your codebase to make your AI assistant stop grepping the same files 50 times. It basically understands what's there to make it easier for uh Claude Code or whatever coding tool you're using to help you develop.
Help me understand the problem and why this is great.
>> Yeah, a good word for AI is when you start a new Claude Code session, it has amnesia. It forgets the code it was looking at. It forgets all of where your files are, where your folders are. And you know, if you're starting up a new project and you've got 50 files, that's pretty easy. Claude Code can find those files. It's going to look at the whole list. When you get to these bigger real applications that have thousands or tens of thousands of files, Claude Code will spend almost its entire effort budget on just trying to find the file that you're talking about. You say, "I want to go change this button," and it's going to search for minutes and minutes, burning tokens in the process. So when you burn tokens, you burn time and you burn money. So, CodeGraph is a way that you could say, well, instead of just forcing Claude to look at all the files and folders on my computer, I'm going to give it this tool that can help it search really effectively to find, oh, here's the one file, and that one's related to this one, and these are the 10 we should look at. Now, Claude Code gets to look at only a small set of files. It's very fast, which makes it more accurate, which makes it cheaper.
Okay, next. I'm going to go through this one quickly. Here's the deal. You're spending thousands, tens of thousands of dollars going to school, and then you have to write all these academic papers, and it's a big involved process. So maybe you use Chat GPT, maybe you use Claude, maybe you create your own skills. The idea behind this is they've got a set of tools, including these agents, that all work to help you do your deep research on your papers, to help you write, to help you review, and so on. It's interesting. It's a good set of uh skills. I think these are interesting, but not super compelling. I'll leave academic research skills to someone who's in school to try out and help them, but it's not going to get you through school.
>> Fair.
>> Yeah, I think that's fair.
>> I maintain I'm against AI-generated academic research, but I do think this is a great example of how to build out a cool skill pack. And so, it's a nice template. Um, I may be against the the...
>> Number three repo of the week. It's like how...
>> All right. Maybe they're...
>> What are you gonna do, Andrew?
>> I don't know.
>> It's impressive, though. They're sharing their skills.
>> Okay.
>> Zapier this week is sponsoring three extra players that did not make it into the top 10 list, but deserve a spot in your hearts and on your computers. I'm going to tell you why Zapier is doing this. Because they're supporting developers like you and me. And the beautiful thing about Zapier is they've got this MCP that will allow you to connect 10,000 tools into anything that you use, but it makes it so easy. Nice visual connection. You get to pick what you want access to. I love the example of, "I'm going to let my agent read my email and draft my email, but not send my email." That's the kind of power that you have with Zapier MCP. If you're not using it, if you're thinking of Zapier as like what it was 10 years ago, you're old-fashioned. Get up with it. Zapier is the tool to connect to right now.
All right, so here's what they're introducing us to this week. This is not on the top 10 list, and it's stupid that it's not. CLI Printing Press. It is from a well-known developer, Matt Van Horn. What's Garrett? What's the problem that Matt has identified?
>> Yeah, you know, Claude Code doesn't really know how to talk to APIs, which is the interface that almost every service is connected with on, you know, on the internet. And these APIs, they're often esoteric, they're complicated, there's pages and pages of documentation for how to talk to them. And we just talked about it before, Claude Code struggles. It can do it, but it struggles to gather all that context so that it knows how to talk to it. What does Claude actually want? Claude wants a simple command-line interface where it can run a command, or even better, it wants an MCP server, which is like the command line but built into the AI instead of it relying on this external system. So what Matt has built here is a tool that can print for you a new CLI. You say, "Oh man, I wish that I could, you know, interact with the ESPN website like this." You can go download Printing Press. It's a set of skills. It's a set of tools. You can create your own custom CLI and MCP that can look at whatever service you want. I would give an example like, I'm applying for grants in Missouri. No one's built an MCP server for that. Well, Printing Press could build one for me in probably 20 or 30 minutes.
>> I love also the example from one of my past guests here on the program. First of all, we should say that here is the GitHub repo, CLI Printing Press. And the thing that I like about it is when people are creating these, they can add it to this library here where other people can see it. My friend Katherine, past interviewee here, she has created a few. Here's one of hers for Shopify. Where's the one that I think she did for Katherine? I want to show this one because it's really good. Katherine, she's saying, "Look, I want to sell a bunch of stuff on Facebook Marketplace, but I can't stand all those annoying interactions from people who are saying, 'Is it still for sale?' Yes, it's still for sale. I don't want to deal with all this. I want my agent to deal with it, but my agent can't go into Facebook." Honestly, Andrew, I struggle to go into Facebook because it's just such an annoying collection of links. Just a mess. I want an easy way for my agent to interact with Facebook Marketplace and deal with all this. For me, it's not that hard to do if you can interact with it. So, that's what she built. She printed it. Everyone else can access it. This is the kind of use case that we're talking about and is a really good repo. People are talking about it. It should be top. It should be top 10.
Um, all right. Next one is I love this one. This is from OpenAI, Realtime Meeting Assistant. Here's the deal. Everybody's got these meeting assistants, right? They listen in on your calls and they transcribe. The people over at OpenAI said, "You know what? What if it could do more than that?" Here, I actually, if you get our report, you're going to get this link where you can see OpenAI's video explaining this.
>> [snorts]
>> They're essentially saying, "Look, imagine if you're in a meeting and instead of just having something take notes, what it can do is like create a Kanban board where you and your team can say something like, 'We have to um, we have to add a CLI to our site,' add it to the backlog." Actually, you know what? Let's move it in progress. Somebody just started. And then it gets moved in. So, you're in a meeting talking, and notes are not just being taken, but tasks and other activities are being done. This is one example that they've given. What do you think of this?
>> Yeah, I'd point out this one isn't a tool for people to use. This is a demo. They're saying, "Look at one of the cool ways you can use the OpenAI real-time API." And what you don't get enough of now is people building these custom applications that use AI in real time to, you know, bring up their game. Let me give you one example, which is in one of our businesses, we've started testing. What if on a sales call, we have this real-time API that's listening to the call? It knows the templates of our sales contracts. It knows the projects that we perform really well. And so live on the call, the tool is building the sales contract, it's giving you examples of outcomes. This is a recruiting business, so it's also saying, "Let me search our database of past talent that we've placed based on what the lead is saying." I could say, "Oh, these five people would all be really interesting." Well, the salesperson can't search all that stuff in real time on the call. They got to be engaged. Well, the agent is listening, and the agent can pull up those templates. The agent can pull up that real talent. And now the salesperson can point at it on the screen because they're just screen sharing this web page that's like building in real time. They say, "Oh, you want to hire this person? Oh, yeah. I would love that." It's so much more engaging. And and this demo, Andrew, this isn't a tool. This is a demo where I'd say, "Wait, if you have in your mind some custom agent, what if there was an AI agent that was on your call that wasn't just recording, it was actually proactive?" This is the demo where you would clone this to your computer. You'd say, "Claude, I want you to help me customize all of this to help me build my contract in real time or search my database or whatever the thing is that you want that's custom to your business. Take this demo. You could customize it very easily, and you've got now your own little custom, you know, get Fireflies out of there, get Granular out of there. Like you've got your own custom bot now."
>> Interesting. I could see that. You know, one interesting use is when someone tells me how their operation runs to be able to say, "Okay, now turn it into a skill right now." And then we take a look at a skill that does the thing that they just described to me. All right. By the way, I say he, uh, Adam runs Gateway X. If you are building an interesting project, you should be telling them about it and possibly being part of their venture studio. I know I am. So, go to gateway.xyz.
All right. Next. We have so many today. I've got to move faster here. This is such an interesting repo. This guy did the thing that they that we all think we should do. What he noticed was [snorts] that all these different uh companies offer a bunch of free tokens to get you to try them out. They give you a few million tokens a month, a few thousand requests a day, etc., etc. Each one has their own tier. It is a toy on its own. But what if you stack them together? What if you say, "Hey, you know what? Let's use Google's." And then when we're done with that, move over to Gro, and so on. You could just keep using all of these forever. It reminds me, Adam, of this guy who I met in San Francisco when all these food delivery companies were coming out and offering $100 a day. He was committed to not paying for food at all by switching from all these providers. He did this big blog post about it. That's what we're talking about here, right?
>> I mean, that. Yes. And it's genius. And how could you argue with saving a buck?
>> So fun. All right. That is Free LLM API. Next. Uh, by the way, of course, we got this full report. I'm trying to click on all the links that I put together for you. I spent hours on this. If you go click the report below, you're going to get it all and you can look at it in your own time and be able to do this.
Okay, [snorts] this we've seen a lot of text-to-speech tools that are free that are available. This one is really interesting because of how fast it works. It's called Supertonic. I've got an example here, and here it is on Hugging Face. Got the text up here. We're going to hit "Generate Speech."
"This text-to-speech system runs entirely in your browser, providing fast and private operation without sending any data to external servers."
>> Nice.
>> It's pretty good.
>> Yeah, text-to-speech is getting good. I want to show you one actually. Let me get your take on this, and then I'll show you an example of what somebody built with this, because these are not just standalone tools. These are like building blocks for things that you build. What do you think of this?
Well, man, you take this and you combine it with that OpenAI real-time uh chat API example, like, oh my gosh, now now someone's talking on your call. It's like, yeah, exactly. These these building blocks, and man, it's free. Like, this is not a thing where you're having to pay for tokens. You're not paying 11 Labs or whoever. This is free. And I think you can customize your own voice in this, right? You can clone yourself.
>> I think. So, um, here, let me find it. No, no, no, no, no. There is a way to clone it. It's not directly. I forget about the cloning part of this. Um, good question. I don't know. We'll have a link.
>> I think it's on their on their website. I think they have a link to do it.
>> I thought so, too. Okay, let's come back into this and see if it's on here. Here we go. That's where we saw it. [snorts]
>> There you go. Yeah. Amazing.
>> All right. Um, um...
>> Here's... Oh, this is what I wanted to show. This guy on Hacker News built an ebook reader. Is it going to... There it is. I didn't want it to load up the app itself. Right. This is this is nice. He took it, built a tool that's that's effective. Got it in the App Store. It's not it's not super huge, but he's been maintaining it really well. Somebody said, "This is promising. Any plans to implement Super Tonic 2?" He goes, "Yes, actually, Super Tonic 3 model." So...
>> Really interesting product. I love how people are actually building with us with all these great repos. [laughter]
>> This one is coming back. All right, I'll set it up like this. Every new chat that you've got with your AI feels like meeting a brilliant employee who totally forgot what you told them yesterday, the day before that, and a month before that. What this does is it gives your um agent a memory. I've got hands-on use case. Anything you want to add to this from what we said last week about it? Agent Memory is the name of the repo.
>> I mean, you you gotta if you haven't tried one yet, you have to try one. If you haven't plugged in some memory system into your building process, you need to try it. Once you try it, you're going to be hooked and you're say, "Oh my gosh, now this thing starts up and it doesn't have amnesia. Like, I don't have to waste my time redescribing everything I'm working on." Like, it changes your whole workflow.
The one that I like is Gbrain by Gary Tan. and it it's got a bunch of tools in it. We need to spend more time on Gbrain from Gary Tan and um I like it because I like the way that it lays out the information so that I can read it and use it on its own. Okay.
>> Yeah.
>> Next. Cloak Browser. I've been running into issues with browser a lot. You know what I was doing, Adam, is...
>> I want to move my wife's website from um Squarespace over to Cloudflare. Freaking Cloudflare makes it so difficult to touch anything with them. And I just want my AI agent to go just learn, just read it. And it keeps coming up with these with these "Click if you're not a robot" screens. And so I will manually click that I am not a robot on the the Gbrain browser or the whatever the GTools, whatever Gary Tan's thing is.
>> Like this thing, this thing has been my nemesis. And the idea here is we are going to get you around all of those. That's what Cloak Browser does. If you're doing anything with a browser on your computer, this is the tool to use.
Yeah, I I actually tested this one over the last week because we do a lot of scraping, and it would be nice to be able to do some more automations, not just scraping, but also clicking buttons, adding data, performing actions like on platforms like LinkedIn. And it, I found it to be pretty effective. It did get blocked once, but it was far more effective than any of the other tools that we're trying. And it was extremely easy to set up. Like they they say just swap the import three lines of code. It really is like 10 lines of code. It's a very easy thing uh to get plugged in. So, if you're doing anything where you're trying to interact with websites, whether that's scraping data out of LinkedIn or Facebook or wherever, or if you're trying to interact and go and update sites that don't have APIs, that don't have CLIs, I mean, this is this is definitely a great way to do it.
>> How do you get your uh username and password into the browser? Do you type it in yourself or do you give it in?
>> That's that's the sketchy part. You know, you program it into your into your script, and your script types it, you know, through the browser types it onto the screen for you. So, you know, you still all these things you got to be a little bit worried about security. You know, what are you giving up? What's the risk? Um, and then you've got to be a little worried, you know, when if LinkedIn figures out what you're doing, maybe they'll ban you. Probably they'll only ban you for 20 minutes. I've been banned many times for for 20 minutes. Uh, and I, I can handle it. Uh, but, but yeah, it's, you know, I got a job to do too, Andrew. I got to, I got to push through. And so something like this that's so easy to set up is just I appreciate it so much. This used to be hours and hours of work to get scraping to work.
>> Now it's you're looking at it's three to 10 lines of code.
>> So I installed it. Also, I didn't even read the instructions. What I did was I copied the URL from GitHub. I said to Claude, "Can you just make this work?" and it made it work, and it's nice.
All right, let's go on to the next one. Number seven for the week. I'm so excited to talk about this because this is creepy as hell, but also super useful. All right, I'm going to show you a YouTube video that I picked right at the spot. What this does is this. It says, "Look, you want a motion sensor for your house, for your office? Why go buy it? Why spend tons of money putting it in all these different systems when you already have this Wi-Fi router? The Wi-Fi router is already in every room in your place and it's directly connecting into your device. And the Wi-Fi router is using these waves to figure out like where your device is, how to connect to it. Well, the interesting thing is if you interrupt the waves, it could sense it, and so it knows that there's a human in there. It actually even says that if you have a heartbeat in the room, it could detect it. And that's the idea behind this. I saw a Hacker News post about this from like 2 years ago or so where people said, 'This doesn't work. It's fake. It's slop.' Whatever it is, it has not only been working, it has been picking up steam on uh GitHub. It now has over 60,000 stars. It is called RuView. What do you think of this?
>> I got questions.
>> I mean, it it I know that it works. I've used ESP32s for human detection before. Also, like most of the the motion detection sensors that are running today, they're all microwave sensors. They're exactly the same technology. It's no longer the little infrared. It shines light on you or sees your light moving. Uh, it's it's microwave. So yeah, this is just, hey, you've already got a microwave sensor. You've already got a sensor in in your room, and it's your it's your Wi-Fi router. So absolutely this this works. No, no argument there.
>> How does it how does a Wi-Fi router pick up? I thought Wi-Fi just...
>> Oh, no. It goes both ways. Of course. Of course, because...
>> So it goes both ways and it's and it's really, you know, it's on most like think of like radar is a really easy example, which is there's those radar domes outside an airport. Well, how do they work? Well, they send out radio waves, and those radio waves hit something and they bounce back, and they time how long it took something to bounce back. So, there's all these different ways where you can see, oh, a signal bounced. Oh, there must be something there. There's also a signal bent around something. There must be an object there. Or, you know, the heartbeat one, Andrew, it's like, well, actually what's happening is my skin is moving, my blood is moving, and that changes, you know, for radio waves, which is a kind of light, it changes how transparent I am in the moment to those radio waves. Well, if we monitor the transparency of that object that's in front of me now, we can detect the the radio waves there. And so it's all these little blips. It's very, very noisy, hard to detect signals, but as processing has gotten better, as these like software-defined radios and these radios in a thing like an ESP32, as they've gotten more precise, we're able to do some pretty awesome things with, look, a $9 board. Like, you could buy, when I buy ESP32s, I buy them in packs of 10. You know, these things are really cheap, and they're little Wi-Fi uh pro uh processors that you could program.
>> I like these use cases. Okay, fall detection. You put one in your parents' house. If there's a problem, you got an you got you're alerted to it. Uh, if there's somebody in your office, you want um a security device, you want the lights to turn on, you want them to turn off. This is really interesting.
>> And then you don't have to have those those people who start to wave their hands around when they haven't been in a room for like or when they've been in a room for a bit and light turns off because they haven't moved enough. Okay. All right. RuView. Really interesting.
>> I don't understand this one. This next one says, "A faster replacement for Node.js that bundles in every tool you'd normally install separately." Help me understand what this is.
>> Yeah, you know, JavaScript has been around a really long time. It's it's very old. It's it's oh man, I don't even know. It's probably at least 30 years old at this point, maybe 25 years. And that means that there's a lot of croft around it. And and even NodeJS, which is this package management system where it says, well, I can help you run tests over here and I can help you package your application over here. These are all like a bunch of individual tools that Node helps to manage. What Bun did is they said, "No, like we see that everyone's building the same set of tools. Everyone's buying these same five things. We're going to bundle that all into one application and we're going to build it here in, I think they started in 2024, 2025. We're going to build it in a way that's really, really efficient, that's using modern technology, and so it's fast, it's efficient, and all the tools that you need to run tests, to to build your package, to execute like the development mode, to deploy it, it's all built in. And it's so good that last year, Anthropic bought them. They bought them not necessarily because uh they, you know, loved Bun, although Claude Code is built with Bun. They're using Bun. They wanted it because man, the guys that built this, they're amazing. That team, they're geniuses. They they see JavaScript in a way that nobody else does, and we need Claude Code, and we need our software to run with that efficiency because they were wasting a lot of time in compute just building and building over and over again. So, they bought the Bun product and Bun team to get the the like very, very smart nerds on that team to build stuff for for Claude. Uh, and they paid a lot of money for it. It's a great acquisition, and they, and they really are very smart people.
All right, next. 12 Factor Agents, a free playbook for building AI agents that actually work in production. The hard-won lessons that usually make that usually take a year of mistakes. Let me show you the original Hacker News post about this. This was from about a year ago from the creator. He goes, "I've been building AI agents for a while. After trying every framework out there and talking to many founders building with AI, I've noticed something interesting. Most AI agents that make it in production aren't actually agentic. The best ones are mostly just well-engineered software with LLM sprinkled in at key points." So I set up document that I've learn I set out to document what I've learned about building production-grade AI systems. Help me understand this. What what are we looking at here?
>> Yeah, I mean, Andrew, you've used Claude Code. Most of the people listening to this have used Claude Code a lot. And what happens with the people I'm advising and even in our businesses is Claude Code can do anything, man. You can drop in new files. You can ask it to integrate with this service just by giving it a URL. And then we use Claude Code to build applications that are like no smarter than the application I could have built 5 or 10 years ago. It's like, oh, but now I can't drop in a random URL into this agent I made because my agent isn't all that flexible. It's not all that smart. It's nowhere near as smart as Claude Code. And it's what what this is trying to say is like it's not truly agentic. You're telling me that that Claude Code could integrate with a voice-to-text service, but if I'm using the app that I've built, I couldn't just like a random user couldn't like say, "Hey, I want to enable voice-to-text in this app." The app's like, "No, I'm sorry. I don't have voice-to-text capability." Well, if it's Claude Code, you'd say, "Awesome. Build it." If it's most apps, the app is going to say, "Nope, I can't do that. I don't have that. I'm sorry." And then you're okay, now I can't build it. So, how do we get that like power of Claude Code and the agenticness of Claude Code into the applications that we're actually deploying for our users? And these are 12 like rules, principles that you could follow that if you believe them, if you if you want this to be agentic, if you want that vision, like these are the, you know, your 12 commandments that you should have on the wall.
>> Okay. So, for example, I now have this tool that I built with Perplexity Computer, which is really cool, that will hunt down potential sponsors for my newsletter. It will put them into a combine board. If I like them, it will then go and find their contact information. Actually, they'll say, "Here's a person who's deciding on newsletter uh ads. Here's their contact information. Here's what I think you should say to them." What I would like is to be able to speak to it and say, "Move this card over here. What do you think of this?" And not just fill in the forms. This is what I would use to add that in.
>> These these are the design principles that you would use when building that tool that would allow the tool to be flexible. Because if you wanted to add that in today, you wouldn't go into your app to add that in, right? You'd have to go back to Claude Code. You'd have to go back to a real agent to build in this new feature. And a truly agentic application, you can ask it to build it. You could say, "Hey, I know that that I'm only one user, but I want you just for me to integrate with this with this other service for me, please." And and the app should very well be capable of integrating that service all on its own without relying on the developer to build out Claude Code. This is kind of a pipe dream. This is like a utopia sort of vision, but that's what principles are for, right? They're they're here to set the the bar really, really high so that we build applications that are that are capable of like genuinely useful and groundbreaking things.
All right, number 10 for the week. EasyVibe. All right, this understands that a lot of us are learning to vibe code. I had the benefit of working with you directly, Adam. You were teaching me all the principles. You were teaching me the software. You're walking me through it. Most people actually are either on YouTube learning it one step at a time, trying to piece things together, or they might spend hundreds of dollars on a course. They they're people who are spending now thousands of dollars on these courses. What this is is them saying, you know what, we are going to put together a free course that will teach you how to vibe code, teach you the principles, teach you the tools, and make it fun by gamifying it a little bit. That's the idea behind here. What do you think?
I love it. And and you're seeing the screenshots right now that I love the most here, which is is it's a bunch of little games, too. You know, there's content. You've got to learn some of these things, but it's got to be interactive. And a lot of the AI concepts, honestly, they're hard to wrap your head around, and they're not things we were trained on. No one went to university for this yet. And so, how do you learn it? Well, you learn it visually. You learn it by playing. And the people that I've seen that have learned it the best, they've screwed around. They build games. They build funny little side projects. They're not taking it too seriously. And a tool like this that helps walk you through a bunch of these experiential learning modules is exactly what it would take for most people. And this is free. You can download this. You can install it. Like you don't need to pay anybody. You don't need to find a teacher. Like there's kind of no excuse to not start learning this now. Go and clone this. The thing that I like is there are things that I don't even know to that I could do or could ask. The best example of that is when I learned from Pat Walls how to build a website in Claude Code. I watched him use it, and what he was doing was he took he took a screenshot of YouTube and he said, "Copy YouTube for me," and he pasted it into terminal. I go, "I didn't know you could just paste into terminal like and tell it to do that." Actually, it wasn't terminal. He was using Ghosty or something, but he pasted it in. He said, "Copy that," and it freaking did it. So, this tells you what you don't know and what you should be aware of.
All right, we got these top 10 for the week. I've got a set of 10 that you need to see. GitHub repos. If you watch this far, I've got a link on the screen for you to go watch and see 10 others that you need to install right now if you're into vibe coding, building with AI, or frankly just into seeing what's hot on GitHub. There's a link on your screen right now. Go get it.