Obsidian's CEO Made His Second Brain Public: It's Free, but Don't Rush to Copy It

Obsidian CEO Steph Ango, also known as kepano, has put his personal note vault template on GitHub. Free.

Many people's second brains stop on the first page: create a new vault, face the blank page, and start wondering how to divide up the folders. kepano's template at least gives a reference point. He states clearly that this is just how he uses it, not dogma—take the parts that are useful to you.

He didn't just publish one vault. The free repositories connected to the Obsidian team add up to a starter pack covering everything from templates to plugins, themes, and canvas:

• kepano-obsidian kepano's own vault template—copy it and you don't have to start from a blank page.
• jsoncanvas Obsidian's open canvas format; canvas data is stored as ordinary files and isn't locked into a specific app.
• obsidian-releases the release list of community plugins and themes; when looking for plugins, start here.
• obsidian-sample-plugin the official plugin template—if you need a plugin that's missing, you can modify one yourself.
• 40-questions 40 questions to turn scattered notes into an annual review.
• obsidian-minimal a clean theme that at least makes the vault look less like a junk room.
• flexoki a color scheme kepano made for writing and code.

These repositories solve the starting problem. The real problem comes three months later.

A widely shared article divides second brains into two types: filing cabinet and compiler. The filing-cabinet model is collect, archive, and search when needed. The article cites Andrej Karpathy: RAG re-derives knowledge on every query, whereas a compiled wiki derives it once and keeps updating it. Based on this, the author built a raw/, wiki/, output/ structure, used CLAUDE.md to record a user profile, and had the model do daily archiving, linking, and flagging contradictions.

Sounds smooth. But someone in the comments threw cold water on it: copying the CEO's vault also copies his habits. The folders and tags in the template grew out of someone else's brain. Which ones you keep and which ones you throw away matters more than the download itself.

Another objection is more practical. A compiled wiki costs tokens up front to summarize, link, and update. Knowledge you use frequently is worth it; a document you happened to read once is not. Compiling everything may turn into a sunk cost. Others mentioned that once documents reach the hundreds, 'putting new files in the right place' becomes harder, and models will hallucinate. Automatic archiving isn't free.

So a more reliable way to use it is: treat kepano's vault as a reference, run it for a week first, and see which structures you actually use. Put high-frequency, important content into the compilation process, and keep low-frequency material searchable. Don't treat the template as the answer; it's just someone else's workspace.

If you want to take this a step further, tools like akm are doing something else: packaging knowledge, skills, and workflows into libraries that agents can call, and they also support indexing a Karpathy-style wiki. It doesn't solve content quality; it only solves invocation and reuse.

Skipping the blank page isn't hard. The hard part is keeping the system useful when no one is managing it. kepano's repositories provide a starting point; the rest you have to delete yourself.