Transcription
Andre Carpathy just tweeted out a brand new second brain system that you can use with Obsidian to create a massive repository of different documents, knowledge bases, and pretty much anything you can think of and essentially build a true artificial intelligence-built second brain that you can house literally any piece of info that you can think of and build a massive repository of info that you can easily search through using something like odd code.
And in this video, I'm going to go ahead and show you guys how to build something like this, how this works, and show you the reason that this is better than just building a simple second brain using the traditional folder structure. And we're also going to show you how to use some really cool tools with Obsidian, which are going to help you pull info from the internet at lightning quick speeds.
So, first things first, if you don't know who Andre Carpathy is, he has been doing some really amazing work lately with a bunch of different large language models. And that work specifically has led him to a few very interesting developments specifically in how you can use LLMs for really advanced data entry, data analysis, and more specifically storage of that knowledge and being able to interface with it and basically have conversations with it.
And one of his tweets that went viral recently, just a couple of days ago, was this tweet right here, which is basically saying that you can take any existing LLM, and this doesn't have to be Cloud Code, by the way. This could be built on codecs. This could be built on cursor. This could be built with any type of LLM out there. And essentially, you can build a knowledge base using the LLM. But you can do it in a way where it's easier to ingest things like massive amounts of documents or information and basically use an IDE or independent development environment as the front end, which in this case we'll be using Obsidian.
Now, the first thing I need to show you guys how to do is what is Obsidian and how to download it, which is pretty easy. All you have to do to download Obsidian is head over to obsidian.md, which is going to be this URL right here. And then, as you can see, you can get it completely for free by simply clicking on this button, which will download to your local system.
Once you've gone ahead and done that, you're officially ready to open it up inside of your computer and you should see more than likely not something like this because you'll have a brand new version of Obsidian. So, what you need to do is head up to file and then simply go over to either new note, new note to the right, new tab, new window, or basically just open a vault. We're going to be clicking on the open vault because this is likely the screen that you're going to be met with first. And you're going to create a new vault by clicking on create, which should give you this screen. Obviously, once again, without all this info, but this is where we're going to start building our knowledge base that we're going to be using a basically better version of the RAG system that we would otherwise have to use.
So once we have this existing Obsidian Vault, the next thing we need to do is we need to start building out our folder structure, which is these three core folders that you're seeing right here. One output folder, one raw folder, and one wiki folder. These are pretty easy to build, and we can actually go ahead and use Claude Code to get most of this project pretty much built out completely for us.
So the first thing we need to do is we're going to open up a new terminal window, which should look something like this. If you don't know how to open your terminal, you can either search for it on your Mac OS or search for PowerShell if you're using Windows. I'm personally using Mac OS, so it's going to be obviously a little different, but many of you out there I know already use it, so this should be easy.
Once we're inside of this terminal, we need to do what's called CD, which basically means that we're going to open up the specific Obsidian Vault directory that we need to be working in. I'm going to type in CD. Then what I'm going to do is I'm going to open up my Finder window. And then inside of my Finder window, I'm going to head over to documents, which is this little part right here. Now, this might be different depending on the system that you're on and your preferences. But me personally, when I go to create my Obsidian Vaults, they appear in my document section. As you can see right here, this is my Obsidian Vault. So once we have that Obsidian Vault inside of our documents, we'll simply take that folder path, drag it into our terminal, just like that. And then I'm going to hit the enter button.
Now that we've typed that out, CD space with the path to that folder, we're ready to start up our Claude Code instance. So I'm going to simply type in claude to invoke Claude Code. And then you'll see Cloud Code right here. Okay. So now we are officially working inside of our Obsidian Vault. There's a couple things you need to do to get these three folders that we were just talking about that we need to set up so that way this works properly.
Now, lucky for you guys, instead of having to kind of reverse engineer this and everything, I essentially built a really easy to copy and paste system inside of our school community, which is going to be that first link down in the description. I'd highly recommend joining because one of the most important parts of this is going to be something called the Claude.md file, which is basically like the brain of this project that makes it work properly. And we have a copy and paste version of that inside of our community that you basically can take and start using literally the second you hit that join button. Uh, but you got to join in order to see that empty file to get that all set up as well.
The first thing though you're going to want to do is you're going to head over to this part right here where you're going to make your folder structure. And I have a very simple little prompt that you can copy and paste here, which essentially just says to create this folder structure for my knowledge base and for source material. It tells you all the folders you need to make and all that fun stuff. We'll basically copy that, open our terminal back up, and then inside of our terminal, we'll paste that into our terminal just like this. You'll hit the enter button, and then once you hit that enter button, we'll open Obsidian, and you'll see these three folders get created for you with nothing in them. Now, obviously, I've been doing some work inside of mine, so they obviously currently have some information inside of them, which brings us to essentially the next part of this video, where we're going to show you how to get information, documents, different web pages, images, and all kinds of stuff into your Obsidian by using a very special and also really useful browser extension.
So, let's go ahead and pull up my Google Chrome right here. And then I'm going to head back over to the Obsidian Web Clipper. And then inside of this, this is going to be a new URL. So, we're not on the original Obsidian website anymore. We're on the Obsidian Web Clipper browser extension, which is basically found at this URL right here. And then you're going to go ahead and add this to your Chrome browser. Also, this works on things like Mozilla, Firefox, and I'm sure there is probably other versions of this extension if you're not using this browser. Uh, I would just recommend using Chrome just because it's going to be a little bit easier. But you'll click on add to Chrome. It'll basically give you this little extension right up here. And then inside of this extension, you'll see that anytime you click on this in your Chrome extension, uh, depending on the web page that you're on, if there is information in front of you, you'll see that it has a bunch of info here. And this info is essentially what's going to get pushed directly into your Obsidian Vault. You can see here, it even converts it into a markdown file, which is really helpful and also very, very useful. And then you'll see down here where it says raw, this is specifically the folder that this info is going to get pushed into.
So just to show you an example of this, I have Andre Carpathy's tweet pulled up here, but I think I already put that one in. So I'll just open up a different tweet of his and we'll just look for something that might look kind of interesting. I don't know any of these will work. Let's go ahead and see this Project Hail Mary tweet that he has. Perfect. So we have this tweet pulled up. I'm going to go ahead and pull up my Obsidian Clipper, which will then pull up all the data and info that we have on this particular page. And then I'm going to go ahead and make sure that we're uploading it to the raw file or the raw folder. And then I'm going to click on add to Obsidian, which immediately takes me into Obsidian with that exact tweet. And it even uploads the images, which is really, really cool and also very helpful as well.
Um, one of the things you need to do in order to get these images, because this won't actually happen for you unless you have a special community plugin that you'll need to download, which is really easy to get. You'll go up to Obsidian, head over to settings. Once you're in settings, you'll go over to community plugins where it says right here. Then it should say something like you'll need to enable community plugins. You'll go ahead and say yes and enable them. And then once you enable them, you'll go ahead and click on the browse button. Once you click on that browse button, there are going to be a ton of different images and plugins here. We're going to look for images plus. I believe it's called images plus or live image. Local images plus is the name of the plugin that you are going to be installing. You go ahead, hit the install button and then hit the enable button. And then this will make it so that way anytime you pull in new info, it will also pull in any images that are available on that web page as well, which is going to be helpful for pulling in data and just also contextually will make it easier for that as well.
But once we do that, there's one more thing that you might be wondering too because more than likely if you're trying to do stuff with this browser extension, more than likely you're getting instead of raw here, it probably just says something like clippings for example. We can make a change to that so that way our workflow is a little faster. You're going to right-click on the extension, head over to options. Once you're inside of options, you're going to scroll all the way down or to default, basically in the templates area. And then once you're inside of default, the only thing you need to change here is what's called your note location, which is basically the folder path that the note or the clipping that you're saving is going to. And we can change that from clippings over to raw, which is where obviously we would want it to go. That way, every single time we activate this browser extension, it will go to raw instead of clippings, which just makes our workflow 10 times faster.
All right, but let's hop back into Obsidian here and let's start actually talking about how you can use Claude Code with this basically interface or index or knowledge base, if you will. So, we have a bunch of data. Let's just say that you went through and just a bunch of different documents just basically kind of like I did right here where I have almost 20 plus different documents about cloud code plugins agents. I just have a bunch of information that I did a bunch of research on inside of the documentation which gives me a lot of info about how cloud code works and so on and so forth. But here's the thing. I don't want to have to sift through every single one of these articles to look for a specific piece of info or to extract some kind of takeaway or specific knowledge. Boy, that's too much work. And the whole reason we built this is to make our lives easier and to obviously be able to interface with this data a hell of a lot easier.
So what I'll do is inside of my Glad Code instance that we were in, I'm also still within that same Obsidian Vault. That means I'm still working. I CD into that same folder and everything. So nothing's changed here. We're still able to, you know, fully interface with all of this knowledge and all the different documents and everything. But essentially, I can pull up this Glad Code terminal window that I'm deleting all this text in right now, as you can see, and trying to buy some time because I just need to delete a bunch of it. And also fixing my hair cuz it's crazy today. And what I'm [clears throat] going to do is now that we have a clear terminal here, I can tell it to do things. And one of the easiest ways or one of the most interesting things I can do is I can create what's called wikis, which is the wiki section that you see right here.
Now, I currently have a wiki section that's existing that's for Claude code skills. And it's just basically a complete wiki on anything and everything you could ever want to know about Claude code skills. And you can see there are little subviews, sub-index parts, and all these little kind of subcategories of info that I was able to pull from the raw data here. And it gives me kind of more actionable insights and interesting info from these deeper raw pieces of info that we have in this system. And this is something that you couldn't do in a regular folder because a regular folder if you try to interface with Claude code, the connections just don't really work as well. It's just a whole lot more clunky. It's a whole lot more difficult to utilize and it's just a headache and a half.
So let's show you guys an example. We'll pull that terminal back up and I'll say instead of Claude code skills, let's say we want to create a wiki about plugins. So I'll say exactly that. Can you please create a wiki about Claude code plugins? Hit the enter button and then we'll go ahead and let it start working where it'll basically look through all of our raw data that we have here and look for anything related to Cloud Code plugins on how to build them, what they are, how to use them, things that we might not even think about like, you know, advanced ways to use plugins, new ideas for plugins. There's so many different ways that it can go with this. But essentially, it'll go ahead and start building out that wiki for us. And this might take a couple seconds just because these wikis are pretty expansive. This was a Glad Code skill wiki that I built. And as you can see, there are almost like 12 different files here that we can kind of sift through. Just so you guys are aware, cuz this is a pretty massive directory. But just in the meantime, in the meantime, while this is working, I do want to make it very clear that if you guys are interested in building a second brain system, this is probably my best prescription for that type of use case.
Oh, and then there is our cloud codes plugin index. This is the index file which basically gives us direct links to each of the info that it's going to be talking about. Oh, and then more than likely, let's see if it's still working here. It's going to build a bunch of subfiles below that. Yep, here's our overview file where it basically tells us what a plugin is, blah blah blah. But you can see there's a bunch of stuff here. Plugin structure, creating plugins, marketplaces, installing, managing, team distribution, troubleshooting, pretty much every single thing you could ever want to know about plugins is being pulled directly from our raw inputs. That's from the documentation. Um, and this is just a basic example, right? Let's say I wanted to make, you know, short-form content about Claude code or about plugins. I could be like, hey, um, I'm really looking to create 10 short-form scripts and I could type this into Claude code directly. Can you help me pull info about Claude code and figure out 10 short-form scripts to write? And then it would give me some really interesting direct insight into this raw data that we have simply by using this setup that we have right here. And you can see too just in the meantime, we have a ton of those other pages that are currently being generated and also are very detailed and very in-depth pulling directly from that raw data.
That's pretty much it for this system though, guys. You could use this for anything. You could use this for your personal business, for your personal projects. And you don't even have to have one singular Obsidian vault. You can create multiple Obsidian vaults. You can open, you know, this is my first Obsidian Vault, but as you create more, more of them will pop up and populate into this little left sidebar here. And then you could have one for one business, one for another business, one for your taxes, one for your bookkeeping. You can do this pretty much infinitely for any type of project that you might be working on. This could also be really helpful too if you run like a consulting business as well as another interesting use case. But that is pretty much it.
Once again guys, if you guys are interested in getting this set up for yourself and doing so a whole lot easier and not having to go and try and figure out how to reverse engineer this, I would highly recommend joining my school community with that first link down in the description. We teach you how to automate your business, literally every single part of it. We give you the agents, the tools, the templates, the downloads, everything you can think of. And that's going to be that first link down in the description.