Agent-readable docs index: /llms.txt. Full docs in one file: /llms-full.txt. Download /docs.zip to grep all markdown files locally.

A brain of one

brainkit is built for teams, but nothing about it requires more than one person. If the question is "can I use this for personal notes — journals, family stuff, health records?" the answer is yes: run your own instance, where you are the company.

Why your own instance, and not the company vault

It's tempting to keep personal notes in your People/ space at work — after all, nobody else's compiled vault can ever contain them. But be clear about the direction of the guarantee:
The compiler protects your notes from other people's vaults. It does not protect them from whoever administers the master repo. The master vault is one git repo owned by the company: anyone with access to it can read every space, and when you leave, your notes stay behind in the company's history.
Notes about your employer's work belong in your employer's brain. Notes about your life belong in a brain you own — same tool, different repo.

Set it up

The setup is Getting started with an org of one — about two minutes instead of five:
brain init ~/brain/master --company "Just Me"
Then make yourself the only entry in ~/brain/master/_meta/org.yaml:
people:
  me: {name: Your Name, roles: [admin], email: you@example.com}
Compile your slice and work in it exactly as an employee would:
brain compile --master ~/brain/master --out ~/brain/compiled
brain index --vault ~/brain/compiled/me
claude mcp add brain -- brain mcp --vault ~/brain/compiled/me
With a single person, the permission machinery has nobody to hide anything from — it sits idle until the day you add someone. And adding someone later is just an edit to org.yaml plus a recompile, no migration: a "family brain" that starts as yours alone can grow a second reader whenever you want, with the same structural guarantees a company gets — see A family brain for the full household setup.

What still earns its keep solo

Most of the toolbox has nothing to do with multi-tenancy:
  • Intake — email, chat, voice, webhooks, and uploads into your Inbox, from anywhere — including tools that push automatically, like meeting-notes apps.
  • Search — hybrid keyword + semantic + graph retrieval, served to any MCP client.
  • Orientation — a generated Map.md in the vault root tells the agent what is there before it searches: spaces and note counts, entity types, the most-connected notes, what is pending.
  • Dashboard — brain dashboard for the live inbox, actions, and note map.
  • Plain files — Obsidian-compatible markdown in a git repo you control, with full history. To explore it visually, see See your brain as a graph.
  • The loop — brain cycle and brain doctor on cron keep it maintained without you thinking about it.
The parts that stay quiet are the ones you'd expect: the promotion queue has no one to share with, and the permissions matrix has one row.

Two brains

The common end state is two instances of the same tool: the company brain at work, a personal brain at home. Your assistant can use both — register each vault under its own MCP server name:
claude mcp add brain-work -- brain mcp --vault ~/work-brain
claude mcp add brain-home -- brain mcp --vault ~/brain/compiled/me
Two repos, two owners, one habit. That separation isn't a workaround — it's the point: the boundary between your employer's knowledge and your own is exactly as structural as every other boundary in brainkit.
Powered byHolocron