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.
Claude Code
$ npx -y skills add NicholasSpisak/second-brain --agent claude-code
Codex
$ npx skills add NicholasSpisak/second-brain -a codex
Cursor
$ npx skills add NicholasSpisak/second-brain -a cursor
Gemini CLI
$ npx skills add NicholasSpisak/second-brain -a gemini-cli
GitHub Copilot
$ npx skills add NicholasSpisak/second-brain -a github-copilot
OpenCode
$ npx skills add NicholasSpisak/second-brain -a opencode
Windsurf
$ npx skills add NicholasSpisak/second-brain -a windsurf
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.
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.
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 |
/second-brain in your AI agent — it walks you through naming, location, domain, and toolingraw/ folderraw/, then run /second-brain-ingest — the LLM will discuss key takeaways and build wiki pages[[wikilinks]], explore the graph view, check wiki/index.md/second-brain-query to ask questions, /second-brain-lint to health-checkyour-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)
The wizard offers to install these. All optional but recommended:
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.
Stars on GitHub
View @nicholasspisak’s profileA comprehensive ecosystem of 76+ specialized AI agent personas designed to work seamlessly with Claude Code's Task tool, providing expert-level assistance across the complete spectrum of business and technical challenges.
@nicholasspisakFetches up-to-date documentation and code examples for libraries, frameworks, SDKs, CLI tools and cloud services, so an AI agent can answer API questions and generate code against current references.
@upstashGives Claude Code skills and agents for writing React documentation on react.dev, covering MDX components, voice and style rules, blog post structure, and interactive Sandpack code examples.
@reactjsAn agent skill that turns complex terminal output into styled HTML pages you actually want to read. Ask your agent to explain a system architecture, review a diff, or compare requirements against a plan.
@nicobailonA WYSIWYG markdown editor so you can edit markdown files like a Google Doc or Notion page, with a desktop app and web UI for navigation, search, and wiki-style linking. It also ships skills for working in codebase wiki, personal CRM, and knowledge base projects.
@inkeepAn 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. Install it with the command on its page. It includes second-brain-ingest, second-brain-lint, second-brain-query. Its skills do not fire on their own yet. Free and open source.
Is this plugin yours?
Claim it with GitHubSubmit a pluginPromote it