Forget the flashy visuals. Here's what a second brain is.
Those glowing knowledge graphs you see online are there because they look good on video. They aren't what makes a second brain work.
A second brain does two jobs. It gives your AI access to the files you choose. And it makes the right file easy to find at the right moment.
So it's a filing cabinet, and that's good news. A filing cabinet doesn't break, doesn't need updates, and doesn't lock you into one company.
The four files that do all the work
| File | What it does |
|---|---|
| me.md | Everything about you. Your business, your offers, your customers, how you talk. |
| INDEX.md | The table of contents. One line for every file in the brain. Your AI reads this first, so it knows where everything is. |
| recent.md | A running diary. A short summary of each conversation, with the date. |
| The rules note | Lives in your CLAUDE.md file (or AGENTS.md if you use Codex). It tells your AI: check the table of contents before answering, and file anything new. |
The rules note is what makes this work. Your AI doesn't get any smarter. It gets a habit of checking the cabinet before it answers.
Paste the prompt
Open the tool you already use: Claude Code, Claude Cowork, or Codex. Paste the prompt below.
It does not start by quizzing you. It first gathers what your AI already knows about you. That includes your instruction files, your memory, your project folder, and your recent sessions. It fills in your profile by itself and shows you the draft.
You only answer what's missing, which is five questions at most.
Tell it where your business lives
This is Step 4 inside the prompt. Your AI asks which folders hold your work.
Point it at your SOPs, your client folders, your projects, and anything else you want it to know about.
It shows you the list before it scans anything. You approve the list, and then it goes.
It builds the map
Most people copy everything into their second brain. That gives them two versions of every file, and both go stale.
Yours doesn't copy anything. It writes one summary page per project or client: what it is, where it lives on your computer, the key files, and the current status. Your files stay exactly where they are. Your AI now knows they exist and where to find them.
The three phrases you'll use
| Say this | What happens |
|---|---|
| “Save this” or “remember this” | Files what you said, right now, in the right folder. |
| “Clean up my brain” | Merges duplicates, files anything it missed this week, archives old entries, and re-scans your folders for new work. |
You don't need to learn any commands beyond these.
The Sunday night cleanup
The prompt sets up a weekly task that runs “clean up my brain” every Sunday night.
Two things to know. It only works in apps that can schedule tasks. And your computer has to be on with the app open.
If the app can't schedule, it will tell you. If you miss a week, say “clean up my brain” yourself and it catches up.
The rules that keep you safe
Your AI reads this folder every conversation, so treat what goes in it accordingly.
- No passwords and no API keys, ever.
- No banking or tax documents.
- The prompt already tells your AI to skip all of these and report what it skipped. Read that report.
- If you point it at client folders, remember that the client info ends up in summaries. Keep the brain folder private.
Quick answers
My history is in the ChatGPT app, not Claude Code or Codex. Paste this into ChatGPT, copy the answer, and paste it in when the brain prompt shows you its draft. It will use it.
How do I back it up?Put the brain folder inside iCloud Drive, Google Drive, or Dropbox. That's all it takes.
Can I read the files myself? Yes. They are plain text, so you can open any of them and fix anything.
Does it work with more than one AI? Yes. The prompt creates both CLAUDE.md and AGENTS.md, so Claude and Codex read the same brain.