How this brain works

Everything you are reading is a plain text file in a Git repository. There is no database and no app that owns it.

The rule

Tools change. Notes should not have to. So the notes are the boring part on purpose — Markdown files with a small block of labels on top.

type: meeting
date: 2026-08-26
client: [[some-client]]
about: [[some-project]]

Those labels are what let a machine answer "what did we decide about this project?" instead of just finding the word.

Two repositories, not one

Filtering happens when files are copied, not when they are rendered. A private note is never written into the published folder at all — so a bug means the site fails to update, never that something leaks.

Publishing is one word

- share: private
+ share: public

That is the whole action. A robot does the rest.

Why not just use a notes app

If your knowledge lives inside an application, you are renting it.

I want to still be able to read these in ten years, with whatever tool exists then.

See also What I do.