depth	2

Second Brain Seed

DOWNLOAD AS ZIP — the green Code button above → Download ZIP. Don't clone unless you know what a remote is: a clone keeps pointing at this public repo, and a git push months from now would send your private thinking here. Unzip it into the folder you want your brain to live in (name the folder what you'll call the brain); the agent sets up version history fresh.

Here for a session with Lorin? Book at loringtonian.github.io/links. More about the Seed, sessions and the community: trimtablabs.xyz.

One prompt that grows into your own AI second brain. No 130-file template to strip down — just a seed an agent reads to build a personal knowledge base shaped to you.

Download it, open it in your AI agent from inside that folder, and say "build my second brain." The agent reads SEED.md, interviews you (purpose, your work, your gaps, your voice, what to pull in, how private to keep things), and grows the structure around what you actually put in.

Turn your stack up first. Best results come from the smartest model, the biggest context window, the highest thinking budget, and plan mode on — setup is a thinking task. (More in → more out.)

Start

• Claude Code — the terminal, or the Code tab of the Claude Desktop app (not Cowork: it has no command line or git, so setup can't run there): start Claude Code inside the brain folder (memory and config load from the folder you start in) and say "build my second brain." (SEED.md auto-loads via CLAUDE.md.) During setup the agent gives you a one-word shortcut — a shell alias named whatever you call the brain (2b, exo, cortex, sec, brain…) — so every later session opens in the right folder without the cd-then-launch ritual. If it doesn't, ask for one.
• Codex / Cursor / other coding agents: point it at AGENTS.md, or paste SEED.md as your first message.
• Any other model (Hermes, OpenRouter, etc.): paste SEED.md as the system / first prompt.

That's it. Answer a few questions, capture one real thought, and you have a working brain that grows from there.

Already know what you want, or short on patience? Hand the agent a few lines on who you are and what the brain is for. It files them as your spec, skips every question they answer, and still writes the core doc — the short path is built in.

The seed ships one example skill pair — /nano-banana-flash and /nano-banana-pro (image generation; add a free GEMINI_API_KEY to .env) — as the pattern for growing your own, and three note templates in templates/ (your own thinking · others' content · AI synthesis) so every note starts from the same shape.

Where it can grow

Get through setup and live with your brain for a week, and it will tell you how to see a fully built-out one — routers, templates, dozens of skills, hooks. Overkill on day one.