tonydzi/second-brain-starter-kit
15 stars · Last commit 2026-11-07
A second brain for Claude Code and Codex that keeps receipts, following Karpathy's LLM Wiki pattern: the agent keeps your markdown vault. Skills as a plugin marketplace, opt-in memory over MCP, vault templates, a CRM engine. MIT.
README preview
# Second Brain Starter Kit: persistent memory for Claude Code and Codex *Read this in other languages: [Русский](README.ru.md)* A second brain and persistent memory for **Claude Code and Codex**: 235 skills shipped as a plugin marketplace, a memory server you can attach with one manifest, a working CRM engine and vault templates. It is the system [Palo Alto AI Research Lab](https://github.com/tonydzi) runs every day, stripped of personal data and cut down to a portable core: **the method, not the data**. It follows [Andrej Karpathy's LLM Wiki pattern](https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f) (April 2026): the agent reads and writes an Obsidian vault of linked markdown notes, steered by a schema file (`CLAUDE.md` / `AGENTS.md`), with one deliberate difference: notes marked `maintained_by: human` stay yours, and agents only propose changes to them. Independent project, not affiliated with or endorsed by him. **Pick your door:** - **New to this?** Start with [the light onboarding guide](ONBOARDING-LITE.md). Download a folder, open it in Claude Code or Codex, and try one useful task. No terminal, GitHub account, shared vault, sync service, or background updater is needed; your work stays in a separate local folder. - **Already have an Obsidian vault?** Take the habits that keep it from rotting without the rest: the [concept template](templates/concept-note.md) with a `maintained_by: human` flag so agents never rewrite your notes, three [read-only subagents](docs/SUBAGENTS.md) that review and propose, and the [contract for unattended runs](docs/SCHEDULED-RUNS.md).