second-brain

Second Brain

An LLM-maintained personal knowledge base built on the LLM Wiki pattern. Drop raw sources into a folder, let the LLM compile them into a structured wiki, and browse it all in Obsidian.

How It Works

You feed raw material (articles, papers, notes, transcripts) into a raw/ folder. The LLM reads everything, writes structured wiki pages, creates cross-references, and maintains an index. You browse the results in Obsidian — following links, exploring the graph view, and asking questions.

The LLM is the librarian. You're the curator.

Prerequisites

• Obsidian — the markdown editor you'll browse your wiki in
• An AI coding agent — Claude Code, Codex, Cursor, Gemini CLI, or any agent that supports Agent Skills
• Node.js — required for installing the skills via npm

Install

npx skills add NicholasSpisak/second-brain

This installs four skills into your AI agent (Claude Code, Codex, Cursor, Gemini CLI, and 40+ others):

Skill	What it does
/second-brain	Set up a new vault (guided wizard)
/second-brain-ingest	Process raw sources into wiki pages
/second-brain-query	Ask questions against your wiki
/second-brain-lint	Health-check the wiki

Quick Start

1. Install the skills (see above)
2. Run the wizard: type /second-brain in your AI agent — it walks you through naming, location, domain, and tooling
3. Install Web Clipper: Obsidian Web Clipper — configure it to save to your vault's raw/ folder
4. Open in Obsidian — launch Obsidian, choose "Open folder as vault", select your vault folder
5. Clip your first article to raw/, then run /second-brain-ingest — the LLM will discuss key takeaways and build wiki pages
6. Browse your wiki in Obsidian — follow [[wikilinks]], explore the graph view, check wiki/index.md
7. Keep going — /second-brain-query to ask questions, /second-brain-lint to health-check

What You Get

your-vault/
├── raw/                    # Your inbox — drop sources here
│   └── assets/             # Images and attachments
├── wiki/                   # LLM-maintained wiki
│   ├── sources/            # One summary per ingested source
│   ├── entities/           # People, orgs, products, tools
│   ├── concepts/           # Ideas, frameworks, theories
│   ├── synthesis/          # Comparisons, analyses, themes
│   ├── index.md            # Master catalog of all pages
│   └── log.md              # Chronological operation record
├── output/                 # Reports and generated artifacts
└── CLAUDE.md               # Agent config (varies by agent)

Optional Tools

The wizard offers to install these. All optional but recommended:

• summarize — summarize links, files, and media from the CLI
• qmd — local search engine for markdown files (useful as wiki grows)
• agent-browser — browser automation for web research

FAQ

The wizard failed or I need to re-run setup. Run /second-brain again — the onboarding script is idempotent. It won't overwrite existing files, so your data is safe. If you need a fresh start, delete the vault folder and re-run.

I accidentally modified a file in raw/. That's OK. The wiki was built from the original content. If you need the original back, check your git history (if the vault is a git repo) or re-clip the source. The wiki pages are unaffected.

wiki/index.md is out of sync with actual pages. Run /second-brain-lint — it checks index consistency and offers to fix mismatches.

Wikilinks are broken after renaming a page. Run /second-brain-lint — it scans for broken [[wikilinks]] and reports which files need updating.

The wiki is getting large and queries are slow. Install qmd (npm i -g @tobilu/qmd). The query skill uses it automatically when available. It provides fast hybrid search across your wiki files.

Can I use this with multiple AI agents? Yes. The wizard generates config files for each agent you select. They all follow the same wiki schema, so multiple agents can work on the same vault.

How do I handle images in clipped articles? In Obsidian, set Settings → Files and links → Attachment folder path to raw/assets/. After clipping an article, use "Download attachments for current file" to save images locally.

How often should I lint? After every 10 ingests or monthly — whichever comes first. Also run it before any major query or synthesis work.

Based On

• Andrej Karpathy's LLM Wiki pattern
• Agent Skills open standard
• Blueprint & origin story — the founding document for this project

---