Hands-on, step-by-step tutorial for setting up an Obsidian vault that works as an agent-maintained knowledge base — combining Karpathy’s LLM Wiki pattern with a developer’s “second brain” (ADRs, debriefs, projects). One afternoon to set up, running forever.

I would like to announce my Substack: https://roanbrasil.substack.com/ or My X – @roanbrasil

Follow me there and subscribe to receive my articles.

Why this article exists

In the first article of this series, I showed why Obsidian won as the base for personal LLM harness — layer separation, data sovereignty, open format, community convergence. Now we leave the conceptual and enter the concrete: by the end of this article, you’ll have a working vault where you say /wiki-ingest and Claude reads the article, extracts concepts, creates pages, links them with existing notes, and updates the index. You ask /wiki-query "what do I know about X?" and it answers by synthesizing what's in your vault, with links to sources.

But there’s an important decision to make before starting. Most tutorials you find on Medium and Substack today implement only the LLM Wiki — a “second memory” for read articles, books, podcasts. That’s useful for…