onboard-brain
Skill0 starsChanged 2 months agoMITView on GitHub
Interview the user to fill in an empty or thin second brain. Use when the user invokes /onboard-brain, when the brain is mostly TODOs, or when they say they want to "set this up" or "teach you about me".
In this repository11 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- daily-briefSkill
- daily-email-summarySkill
- draft-letterSkill
- meeting-notesSkill
- person-briefSkill
- plan-daySkill
- process-inboxSkill
- recallSkill
- rememberSkill
- weekly-reviewSkill
The file
We split this skill into the rules it gives an agent
12 rules. About 0 are common, 7 are distinctive, and 5 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈7
Distinctive · estimate
This author's own choices. Worth a look.
5
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 11
If CLAUDE.md still shows a SETUP-REQUIRED notice, the brain isn't set up yet.
- Distinctive: line 13
Ask, in one message:
- Distinctive: line 22
If the chosen assistant name belongs to someone in their life (a spouse, a child), point out the mix-up it can cause, and offer to add a …
- Distinctive: line 28
Write as you go: After every 2–3 answers, save to the right file immediately.
- Very specific: line 30
Never invent: If an answer is vague, ask once, then write it as given with (unconfirmed) — do not smooth it into something cleaner than …
- Distinctive: line 33
Show progress: Every few rounds: "That's identity done.
- Very specific: line 39
Read brain/00-index.md first.
- Distinctive: line 40
Ask which area they want to start with, and offer the recommended order:
- Distinctive: line 89
When are you sharpest? When do you crash?
- Distinctive: line 115
Ask directly: "What phrases do you hate seeing in your own name?"
- Very specific: line 119
Update every status marker in brain/00-index.md (⚪ → 🟡 → 🟢).
- Very specific: line 120
Write today's memory/log/YYYY-MM-DD.md noting what was captured.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Fill the brain by conversation. This is the single most important skill here — everything else is only as good as what this collects.
Step 0 — First-time setup (only if needed)
If CLAUDE.md still shows a SETUP-REQUIRED notice, the brain isn't set up yet. Do this first:
- Ask, in one message:
- What's your full name?
- What's your main email address? (Only the email-summary skill uses it, to send you your summary.)
- What would you like to call your assistant? (Suggest "Mona"; any name works.)
- From the brain's folder, run:
bash setup.sh --owner "<name>" --email "<email>" --assistant "<assistant name>" - Relay any warning setup prints, especially about git remotes. Re-read
CLAUDE.md: it now carries their name and the notice is gone. Then continue below.
If the chosen assistant name belongs to someone in their life (a spouse, a child), point out the
mix-up it can cause, and offer to add a rule to CLAUDE.md §0 so the two are never confused.
Rules
- One question at a time, or a small cluster at most. This is a conversation, not a form.
- Write as you go. After every 2–3 answers, save to the right file immediately. Never hold answers in context waiting for the end — if the session dies, the work is lost.
- Never invent. If an answer is vague, ask once, then write it as given with
(unconfirmed)— do not smooth it into something cleaner than what was said. - Let them skip. "Skip" moves on and leaves the
TODO:. Don't push. - Show progress. Every few rounds: "That's identity done. Three sections left."
- The user can stop at any point. Say at the start that this can be done in pieces and resumed later.
Where to start
Read brain/00-index.md first. If some sections are already filled, skip them and say so.
Ask which area they want to start with, and offer the recommended order:
- Identity — 10 min — biggest immediate payoff
- People — 15 min — family first, then work
- Work — 15 min
- Routines & the year's calendar — 10 min
- Life admin — health, finance, home — 15 min
- Voice — 10 min, needs writing samples
Section 1 — Identity → brain/identity/
- Full name, what people call you, where you're based, timezone, languages.
- What do you do for a living, in a sentence you'd say at a party?
- What other roles do you carry? (father, treasurer of the society, the one who handles parents' paperwork…)
- What's the one thing you'd want an assistant to know on day one so they don't embarrass you?
- Hard rules about your time — anything that must never be scheduled over?
- What's your current single biggest focus? What would make this year a good one?
- What are you deliberately not doing right now?
Section 2 — People → brain/people/
Family first: "Walk me through your immediate family — name, relationship, where they live, anything I'd need to know to help you take care of them."
Then: "Who are the 5–10 people you deal with most for work?"
For each person, create a file from brain/people/_template.md. Filename:
firstname-lastname.md, lowercase. Ask for birthdays as you go. Don't interrogate — get
name, relationship, one line of context, and the birthday; depth accumulates later.
Mirror every birthday into brain/routines/calendar-of-the-year.md and add a row to
brain/people/directory.md.
Section 3 — Work → brain/work/
- What entities exist? Company, freelance, side projects — which are actually active?
- How does money come in? Who pays you, on what cadence?
- What are you working on right now? Create a project file per active project from
brain/work/projects/_template.md. - What comes back on a clock — standing meetings, reports, invoices, filings?
→
recurring-obligations.md - What tools do you live in? →
tools-and-accounts.md(locations of logins, never logins)
Section 4 — Routines → brain/routines/
- Walk me through a normal weekday, hour by hour, roughly.
- What happens every day no matter what?
- When are you sharpest? When do you crash?
- What does a week look like — anything fixed on particular days?
- Now the year: birthdays, anniversaries, insurance and policy renewals, passport/licence expiry, tax deadlines, festivals that change your plans, annual checkups.
The year calendar is the highest-value thing in this whole section. Push a little here.
Section 5 — Life admin
- Health: blood group, allergies, ongoing conditions, regular medication, your doctors, insurance policies, when your last full checkup was.
- Finance: accounts by nickname only, fixed monthly outgoings, subscriptions, income sources, who your accountant is. Stop them if they start reading out a number — say plainly that account numbers don't go in files, and record where the details live instead.
- Home: rent/EMI dates, utilities, vehicles and their insurance/service dates, document expiry dates, the plumber/electrician/mechanic you trust.
Section 6 — Voice → brain/identity/voice.md
This one needs artefacts, not answers.
- "Paste me 3–5 messages you actually wrote — one to family, one to a client, one to a vendor or official."
- From the samples, infer: greeting, sign-off, sentence length, formality per audience, emoji use, spelling variant. Write what you observe, and say what you inferred so they can correct it.
- Ask directly: "What phrases do you hate seeing in your own name?"
Finishing
- Update every status marker in
brain/00-index.md(⚪ → 🟡 → 🟢). - Write today's
memory/log/YYYY-MM-DD.mdnoting what was captured. - Move anything unfinished into
memory/open-loops.mdas "brain gaps". - Tell them the top 3 gaps remaining and what each one would unlock.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

