Claude Code + Obsidian = UNSTOPPABLE

CChase AI
컴퓨터/소프트웨어경영/리더십재택/원격 근무AI/미래기술

Transcript

00:00:00If you're using Clawed Code and its lack of memory
00:00:03makes it feel like you're starting every single session
00:00:05from scratch, then you gotta check out this one tool
00:00:08that can fix all your problems.
00:00:10And that tool is Obsidian.
00:00:12It's free, it's simple to use,
00:00:14and it's gonna help us unlock persistent memory
00:00:16for Clawed Code all while boosting our performance
00:00:19in the process.
00:00:20This allows us to do things that other tools
00:00:22like OpenClaw claim to do yet fail completely at,
00:00:25like actually remember things about us
00:00:28and act in a way that a real human personal assistant would.
00:00:31Now, Obsidian is one of those rare tools
00:00:33that really does feel like a value add across the board.
00:00:36So I'm really excited for this lesson.
00:00:38So let's just hop into it.
00:00:40So let's first talk about what Obsidian actually is.
00:00:42Obsidian is an orchestration layer, an organization layer
00:00:46on top of your markdown files
00:00:48in a specific folder known as a vault.
00:00:51Obsidian is free, it is not open source,
00:00:54but everything we create and use with Obsidian is ours, right?
00:00:58We own it, Obsidian does it, there's no vendor lock.
00:01:00This isn't like Notion.
00:01:01And the real value add for Obsidian from our point of view
00:01:05is its ability for us to see how our different notes
00:01:09and files and folders connect.
00:01:11So if I have something open right here,
00:01:13you can see how this connects
00:01:15to other related notes and files, right?
00:01:18And you can see over here on the right,
00:01:19we also have these like visual graphs
00:01:21that show us how different documents
00:01:24and projects relate to one another.
00:01:25Obsidian also has a huge plugin community.
00:01:28So everything that you'll see today
00:01:30is just on the base level Obsidian,
00:01:32but there are 2,736 ways to spice it up if you want to.
00:01:37But why should you care about any of that?
00:01:38How does Obsidian actually relate to you
00:01:42and Claude code and your productivity?
00:01:44Well, to answer that question,
00:01:45you need to understand the symbiotic relationship
00:01:48between these two things and you.
00:01:51So first and foremost, we have Obsidian and you.
00:01:53What do you get out of this relationship?
00:01:55Well, you get what I just saw, right?
00:01:56The ability to look at files and click through it
00:01:59and see how they connect and have like visual insight
00:02:02between all of your folders and notes.
00:02:04The problem is for me to actually do this
00:02:07and get the full value from Obsidian as a user,
00:02:09I have to actually draft up all these markdown files
00:02:13and I have to draft them up in a specific way
00:02:15that they link, right?
00:02:16I have to like make sure, hey, I do the double brackets,
00:02:18I find the note I want, I link it, right?
00:02:20Pain in the butt, you're never gonna do this in reality,
00:02:23right, no one's manually doing this in today's day and age.
00:02:25Enter, obviously, Claude code.
00:02:28Claude code will do all of that for us.
00:02:30Claude code essentially supercharges
00:02:33the Obsidian experience for you
00:02:35because whatever you tell Claude code,
00:02:38whether that's through text or prompts or braindubs
00:02:40or just, you know, verbal diarrhea, right?
00:02:42Whatever you give Claude code,
00:02:44it's going to turn that into a proper markdown file
00:02:48in the Obsidian format, link it all together
00:02:50so you the end user can get better insights into your notes
00:02:55and your projects and your thoughts.
00:02:56But if that's all it did,
00:02:57we still really wouldn't care, right?
00:02:59Neat, I can see more of my thoughts.
00:03:00I already know my thoughts, they're in my brain.
00:03:02How does this affect Claude code?
00:03:03Well, by virtue of creating an organized hierarchy
00:03:08of your notes and actually connecting them all,
00:03:11Obsidian improves Claude code.
00:03:14Now think of all this in terms of Claude code
00:03:16acting as a personal assistant for you
00:03:18and I use that demonstration on purpose
00:03:20because that's the sort of use case
00:03:23where you will get the most value add
00:03:25in this sorts of situation
00:03:26because when it comes to personal context,
00:03:28it's very wide in terms of breadth
00:03:31and the things you're talking about.
00:03:32Think of like daily notes
00:03:33and random projects you're working on,
00:03:34things you're just throwing Claude codes away, right?
00:03:36Huge breadth and it compounds over time, right?
00:03:39So if we're just talking about daily notes
00:03:41for a couple of weeks, easy enough.
00:03:43How about daily notes for a couple of years, right?
00:03:45These things add up.
00:03:46And so by having an organized system of these notes
00:03:50and having them linked together
00:03:51and they actually make sense when you look at them,
00:03:53this helps Claude code, right?
00:03:56Because when you ask it to do stuff
00:03:58and gain insights on things
00:03:59and essentially find relationships between documents,
00:04:02it helps to have that organization.
00:04:04Because that helps Claude code, that helps you, right?
00:04:08Whatever gives Claude code better performance and insight,
00:04:10ultimately that rolls downhill.
00:04:12And hopefully by getting better insights from Claude code
00:04:16based on your interactions with Claude code,
00:04:18that will improve interactions down the line,
00:04:21which will then become files that are properly notated
00:04:25and organized inside of Obsidian,
00:04:26which then come to you who can actually see them.
00:04:29And this whole thing isn't just a black box.
00:04:31That's the value add here with Obsidian.
00:04:34Obsidian is supercharged by Claude code
00:04:36by virtue of organizing your files
00:04:38that improves Claude codes performance,
00:04:40which ultimately makes you happy.
00:04:43That's the relationship.
00:04:44That's what this is all about.
00:04:45Now, you might be thinking that's all well and good Chase,
00:04:50but like, can't I just have a giant file system
00:04:53and throw markdown files in there forever and ever and ever
00:04:56and won't Claude code being as effective as it is,
00:04:59still be able to handle all those files?
00:05:01The answer is yes, to a certain degree.
00:05:03And this is kind of like a spectrum, right?
00:05:05Like on one hand, you can be the average user
00:05:08over here on the left, who's very lazy
00:05:09and uses Claude code as a personal assistant.
00:05:11You just throw crap at it.
00:05:13If you think of it as like being in a warehouse
00:05:15with all of your files and data,
00:05:16like you just have papers on the ground, right?
00:05:19They're just thrown everywhere.
00:05:20Claude code is good enough to handle this.
00:05:22Although that disorganization does come at a cost.
00:05:25The other cost is to you, right?
00:05:27This then becomes a black box.
00:05:29You as the human cannot possibly sift
00:05:30through all that information, right?
00:05:32So you aren't operating at full capacity
00:05:34and neither is Claude code.
00:05:36On the other end of the spectrum, right?
00:05:39We could go away from just markdown files
00:05:42in a folder system and we can instead go nuts
00:05:44and create some sort of like graph rag system, right?
00:05:46This nuclear bomb approach that's gonna have embeddings
00:05:48and analysis and all this stuff.
00:05:50The problem is if we do the rag approach,
00:05:53for most people, that's too much.
00:05:55And it's also too much to expect from them
00:05:56to be able to build and maintain that, right?
00:05:58If over here, you know, we have papers thrown on the ground,
00:06:01this is the library of Congress, right?
00:06:04You don't need that.
00:06:05What we need is a happy medium, right?
00:06:07And what is that happy medium in most cases?
00:06:09Well, it's actually obsidian because instead of papers
00:06:11on the ground or the library of Congress,
00:06:13I just want like a filing cabinet
00:06:15that's kind of organized for me.
00:06:17And then the offer here is,
00:06:18obsidian is gonna be the filing cabinet
00:06:20and it's a filing cabinet that is free,
00:06:22both in terms of actual cost
00:06:24and it's not like this is costing us tokens.
00:06:26This isn't some sort of heavy framework on top of Claude code
00:06:28changing how we interact with it.
00:06:30This is kind of honestly free value, you know?
00:06:35And it's not free value in terms of like,
00:06:37oh, if you use obsidian after this video,
00:06:39your life's gonna change
00:06:40and you're gonna be able to do projects you couldn't.
00:06:41Not really, but it's one of these on the margin plays
00:06:45that I think are kind of rare
00:06:46and that like it's almost just like a straight plus
00:06:49with no negatives, right?
00:06:50Someone's walking down the street,
00:06:51they're just gonna hand you 20 bucks.
00:06:53Are you gonna say no because it's not a hundred bucks?
00:06:55Absolutely not.
00:06:56You're gonna take it.
00:06:57So this video is about how do we take that $20 and use it.
00:07:01Now to use it, we first need to download it.
00:07:03So you're just gonna head to obsidian.md
00:07:05or just Google obsidian, go to the download page
00:07:08and download and run through the installer.
00:07:10It's gonna ask you a couple of questions.
00:07:12It's gonna ask you, where do you want your vault
00:07:14and what do you want to call it?
00:07:16So the vault is just a folder where essentially
00:07:19the obsidian orchestration is gonna live.
00:07:21The folder could be anywhere on your computer.
00:07:23It could be on your desktop.
00:07:24So you just have the folder.
00:07:26It's known as obsidian vault,
00:07:28but you don't have to call it the vault for this demo.
00:07:30It is literally called the vault.
00:07:31And inside of that folder,
00:07:33that's where all our markdown files are going to be.
00:07:35So the vault could have sub folders like, you know,
00:07:39daily notes, research and projects.
00:07:41And inside daily notes are the markdown files, right?
00:07:45All of the daily notes for that particular day
00:07:48as just text documents.
00:07:50And like you saw, if I click on those text documents,
00:07:52I have the obsidian document.
00:07:56When I say obsidian document, I'm talking about this, right?
00:07:57We can look at this inside of something like our terminal,
00:08:00or we can actually open up obsidian and look at it.
00:08:02Same thing.
00:08:02So after you download it,
00:08:03just decide where you want the vault to live.
00:08:05I suggest putting it wherever you put
00:08:07most of your cloud code projects.
00:08:09And now to have this relationship with obsidian cloud code,
00:08:13you will just start a new project inside of that vault.
00:08:16Once you create the vault,
00:08:17all you have to do is open your terminal,
00:08:18navigate to a folder inside the vault,
00:08:21or just be in the vault itself and start up cloud code.
00:08:23You can see here, I'm inside mine, users chase the vault.
00:08:27And over here on the left-hand side,
00:08:29you can see all of the files and folders.
00:08:32And this is my personal thing.
00:08:33And most of it's related to content creation.
00:08:35So you can see stuff for research, for projects, for people,
00:08:38inbox, daily notes, et cetera, et cetera.
00:08:40Now, when I look at these files inside of cursor,
00:08:43it looks just like a normal markdown file
00:08:44because that's all it is.
00:08:45Now there's certain conventions that obsidian follows,
00:08:48but guess who knows these conventions?
00:08:51Cloud code knows these conventions.
00:08:53So when you're working inside the vault,
00:08:55all you have to do is tell cloud code
00:08:57inside of its cloud.md file and say, hey,
00:09:01all markdown files need to follow obsidian conventions.
00:09:05Furthermore, there are a million and one repos out there
00:09:08that have to do with creating obsidian skills for cloud code.
00:09:11So if you wanna take it a step further
00:09:13and give cloud code specific obsidian skills,
00:09:16just Google obsidian skills repos,
00:09:19or literally just tell cloud code, hey,
00:09:21go ahead and do a web search on the best practices
00:09:25for cloud code and obsidian skills and create those, right?
00:09:28That easy.
00:09:29So now let's talk about how we can use this markdown file
00:09:32set up in this sort of memory infrastructure
00:09:35to actually improve cloud code,
00:09:36especially from the position of a personal assistant.
00:09:39And the way we're gonna do that is through the cloud.md file.
00:09:43And that's what you're looking at right here.
00:09:45So remember the cloud.md file traditionally is a file
00:09:49that you use in your project
00:09:50that essentially lists all your conventions.
00:09:52You also have the ability to do something like slash init,
00:09:55right?
00:09:56And cloud will automatically go through your entire code base,
00:09:59figure out what you're doing and why
00:10:01and put it in this file, this cloud.md.
00:10:03It almost acts like a system message in the sense
00:10:06that cloud code is always gonna be looking at this, right?
00:10:09To make sure it's doing what you say.
00:10:12Now, recently, relatively recently,
00:10:15there has been a big push against things like cloud.md files
00:10:20and slash init in general.
00:10:22Now this is in response to this study
00:10:24that came out last month called evaluating agents.md,
00:10:28are repository level context files helpful for coding agents?
00:10:32I'll put a link to this down below.
00:10:33The long and short of it is,
00:10:35is that this study pretty much said
00:10:37that using stuff like cloud.md,
00:10:40these like system message add-ons
00:10:42that are supposed to enforce particular conventions
00:10:45are actually a net negative and they hurt the AI system.
00:10:49The reason for that is,
00:10:50is oftentimes these conventions don't make sense
00:10:54with how we actually create projects, right?
00:10:58Oftentimes we're doing like the UI in one section
00:11:00and the backend in another section,
00:11:01the authentication in another place,
00:11:02yet it's constantly bringing in all the conventions.
00:11:06All set to say is a lot of people are saying,
00:11:08don't use cloud.md.
00:11:10However, in the context of a personal assistant
00:11:13where our conventions don't have to do with code,
00:11:16they're conventions about how we think
00:11:19and conventions about how I want you
00:11:20to write the markdown files in relation to Obsidian,
00:11:25cloud.md is perfect for that, right?
00:11:27Cloud.md is perfect for that.
00:11:28And one of the ways Obsidian allows us to improve this
00:11:31is because all of these files are already organized.
00:11:35It is really easy to turn this cloud.md file
00:11:39into a living, breathing document.
00:11:41What you see here is a few weeks
00:11:44of essentially using Obsidian and cloud code.
00:11:46I am by no means an Obsidian expert.
00:11:48I've not been using this for months and months and months.
00:11:50However, three months from now, six months from now,
00:11:53nine months from now, or just like a month from now,
00:11:55as I continue to add onto this,
00:11:57one of the great Obsidian and cloud code plays you can do
00:11:59is say, hey, take a look at all our notes,
00:12:02compare it to our cloud.md file.
00:12:05Now make them kind of match and improve the conventions.
00:12:08The idea being that your personal assistant cloud code
00:12:11continues to become closer and closer
00:12:14to this Jarvis type character you want
00:12:18by virtue of having more information about you.
00:12:21And in essence, what we get is we get this brain
00:12:24within a brain, right?
00:12:25This entire vault system is essentially your second brain,
00:12:28right, but this cloud.md is like, you know,
00:12:32whatever you call it, like the full cortex.
00:12:34Like this is the one that's actually making decisions
00:12:36and it's distilled our thinking template into one file.
00:12:41And it can always take a look at the details as needed.
00:12:43And this is the sort of way you should approach
00:12:45the cloud code and Obsidian relationship.
00:12:48Now, how you ultimately leverage all of this
00:12:52into something that is useful for you
00:12:54is obviously up to you.
00:12:56How I've been using this lately
00:12:58is pretty much what you saw from my last video.
00:13:00And if you haven't seen it,
00:13:01it was essentially using cloud code
00:13:04as a research agent on steroids
00:13:05in combination with a YouTube search skill and notebook LM.
00:13:09Well, what do you think I'm doing
00:13:10with all this research information
00:13:11I'm gathering from that workflow?
00:13:13I'm dumping it here into my second brain
00:13:15to continue to turn cloud code
00:13:18into my personal research content assistant I want it to be.
00:13:21And in fact, tomorrow's video, at least that's the plan,
00:13:25is to show you exactly how to do that.
00:13:27And really what you should be thinking about
00:13:29in terms of the use cases is again,
00:13:31like if you actually did have your version of Jarvis,
00:13:33what would it need to know and what would it need to do?
00:13:36And figuring out how to execute that
00:13:39in this Obsidian cloud code ecosystem
00:13:41is where you're gonna make your money.
00:13:42So this is where I'm gonna leave you guys today.
00:13:44I hope this video and this conversation
00:13:47was if nothing else able to sort of shed some light
00:13:50on this relationship between all three,
00:13:52because I think once you understand
00:13:54how they improve one another,
00:13:55you get a good grasp of what Obsidian does
00:13:57and does not buy you.
00:13:59Because I think the waters are still kind of murky
00:14:01for a lot of people.
00:14:02Lastly, as always, hey,
00:14:04if you are trying to learn more about cloud code
00:14:07and you're someone who doesn't come
00:14:07from a technical background or even someone who does,
00:14:10it's essentially trying to supercharge their process
00:14:12of learning how to actually use it in a practical manner,
00:14:14I just released my cloud code masterclass.
00:14:17Inside of the Chase AI+ community,
00:14:19there is a link to that in the comments.
00:14:21So make sure you check that out
00:14:22if you're someone who's serious about this stuff
00:14:24and is looking to eventually make it some sort of career.
00:14:26So make sure to leave some comments
00:14:28and let me know what you thought.
00:14:29Also make sure to go into the description
00:14:31and check out the free Chase AI community.
00:14:33Again, tons of free resources there.
00:14:35If you're just getting started on your AI journey,
00:14:37as always, I'll see you around.

Key Takeaway

Integrating Claude Code with an Obsidian vault creates a symbiotic relationship where the AI automates note organization and linking, providing the user with a visual second brain while giving the AI structured personal context for improved assistant performance.

Highlights

Obsidian serves as a free orchestration layer for markdown files that provides persistent memory and relationship mapping for Claude Code.

Claude Code automates the creation of Obsidian-formatted markdown files, including manual linking tasks like double-bracket notation.

The 2,736 community plugins available in Obsidian allow users to extend the base functionality of their AI memory vault.

Maintaining a claude.md file inside the Obsidian vault acts as a 'full cortex' system message that enforces specific thinking templates and conventions.

A study titled 'Evaluating Agents.md' suggests repository-level context files can be a net negative for coding, but they remain effective for personal assistant use cases.

Navigating the terminal to a folder inside an Obsidian vault allows Claude Code to access and organize historical notes as a 'second brain'.

Timeline

Obsidian as a Local Orchestration Layer

  • Obsidian functions as an organization layer on top of a local folder known as a vault.
  • Users retain full ownership of all data created within the system because it relies on standard markdown files without vendor lock-in.
  • The visual graph view reveals how different documents and projects relate to one another through a web of connections.

The primary problem solved by this integration is the lack of persistent memory in standard AI sessions. Unlike cloud-based tools, Obsidian keeps data local and accessible. The interface allows for visual insight into note connections, supported by a massive library of over 2,700 community plugins for customization.

The Symbiotic Relationship Between User and AI

  • Claude Code removes the friction of manual note-taking by drafting properly formatted and linked markdown files from raw prompts or brain dumps.
  • Organized note hierarchies improve the performance of Claude Code by providing clear relationships between wide-ranging personal contexts.
  • A structured vault prevents the AI from becoming a 'black box' where information is lost in a disorganized file system.

The value exchange is cyclical: the user provides raw information, Claude Code organizes it into an Obsidian-compatible structure, and that structure then allows the AI to offer better insights. This is particularly effective for personal assistant roles involving daily notes and multi-year projects. The result is a more capable assistant that acts like a human personal assistant with a long-term memory.

The Spectrum of Data Organization

  • A disorganized folder of markdown files forces the AI and the user to operate at sub-optimal capacity.
  • Complex Graph RAG systems with embeddings are often too resource-intensive for individual users to build and maintain.
  • Obsidian represents a 'happy medium' filing cabinet that provides organized value without the cost of extra tokens or heavy frameworks.

Data management is viewed as a spectrum ranging from 'papers on the ground' (disorganized files) to the 'Library of Congress' (complex RAG systems). Obsidian provides a middle-ground solution that is free in terms of both financial cost and computational overhead. It functions as a marginal improvement that provides consistent value across all AI interactions.

Setting Up the Obsidian Vault for Claude Code

  • The vault is a standard folder that can be placed anywhere on a computer, such as the desktop or a project directory.
  • Starting a Claude Code session inside the vault directory allows the AI to recognize and interact with all sub-folders like daily notes and research.
  • Claude Code is natively aware of Obsidian conventions, requiring only a simple instruction in a configuration file to follow them.

Installation involves downloading the software from obsidian.md and defining a vault location. Best practices include placing the vault where most Claude projects are stored. Users can further enhance the AI's capabilities by instructing it to perform a web search for specific 'Obsidian skills' and creating custom skill files within the environment.

Optimizing Claude Code via the claude.md File

  • The claude.md file acts as a persistent system message that enforces conventions for how the AI should think and write.
  • Updating the claude.md file based on existing notes creates a living document that evolves as the user's second brain grows.
  • Research workflows are enhanced by dumping data from YouTube or Notebook LM into the vault to refine the AI's role as a personal content assistant.

While recent studies suggest repository-level context files can hinder coding agents, they are highly effective for maintaining personal thinking templates. The claude.md file serves as the 'full cortex' of the system, distilling the user's logic into a single reference point. Over time, comparing new notes to this file helps the AI move closer to a highly personalized assistant character.

Community Posts

No posts yet. Be the first to write about this video!

Write about this video