agentleFS
herreraspaintingba-hubherreraspaintingba-hub/second-brain-starter-kit/skills/second-brain-init/SKILL.md

second-brain-init

Skill0 starsChanged 41 days agoMITView on GitHub

First-run setup (and v1 upgrade) for the Second Brain Starter Kit. Builds the Obsidian vault, checks the eight companion skills are installed, puts the brain's tools in the vault (search index, frontmatter contract, project Doors, health check), writes the personalized CLAUDE.md with the three-layer startup, turns on the monthly health check, and verifies everything. Trigger on "set up my second brain," "install the starter kit," "initialize my brain," "set up my vault," "upgrade my brain to v2," "configura mi segundo cerebro," "instala el starter kit," "inicializa mi cerebro," "configura mi bóveda," "actualiza mi cerebro a la v2," or when there is no CLAUDE.md with `second_brain_initialized: true` and the user asks how to start. Runs once per brain; re-running asks before changing anything. Bilingual EN+ES.

The file

Show its rules and text

We split this skill into the rules it gives an agent

19 rules. About 0 are common, 8 are distinctive, and 11 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈8

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 11

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Do not run it: when the brain is set up and the user is just working: they want another skill.

    line 36
  2. Distinctive:

    Use what you find as defaults for the questions.

    line 63
  3. Very specific:

    If it is missing on a Mac, macOS offers to install it the first time python3 runs.

    line 66
  4. Very specific:

    Ask them one at a time (with AskUserQuestion or the host's multiple choice).

    line 73
  5. Very specific:

    Q3. Vault name / Nombre de la bóveda: Default "My Second Brain" / "Mi Segundo Cerebro".

    line 80
  6. Very specific:

    If any is missing, copy it from the kit's skills/ folder.

    line 151
  7. Distinctive:

    If Python is not available, still copy the folder (it will work the day Python is installed) and say so in the summary.

    line 172
  8. Very specific:

    Write {{WORKING_DIR}}/CLAUDE.md from this template.

    line 177
  9. Very specific:

    Write {{VAULT_PATH}}/CURRENT-CONTEXT.md (type: note, with date and time): active projects (each linking its Door once it has one), this …

    line 223
  10. Very specific:

    If {{HEALTH_SCHEDULED}} is yes:

    line 229
  11. Very specific:

    When CLAUDE.md says kit_version: 1.0.0, ask once: "Upgrade your brain to v2?

    line 273
  12. Very specific:

    Add the missing folders: 03 Ideas & Notes/References, 07 Assets, 08 Projects, 09 MOCs, Archives, .brain/tools.

    line 276
  13. Very specific:

    Run brain_health.py --write and show the first light.

    line 281
  14. Distinctive:

    Do not mass-edit old notes to the new contract.

    line 283
  15. Distinctive:

    The user already has a vault: use it. Add missing folders, never move or delete notes, and say so.

    line 288
  16. Distinctive:

    Re-run by accident: stop and ask: keep it, change some answers, or start over.

    line 292
  17. Very specific:

    Windows: use OneDrive or Documents.

    line 294
  18. Distinctive:

    The host cannot run commands: create the folders and notes by hand, skip the tools, and tell the user what they are missing.

    line 295
  19. Distinctive:

    Start Here notes. Upgrade path from v1.

    line 315

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Second Brain: Setup

One-time setup. Builds your vault, sets up the tools, writes your CLAUDE.md and checks that everything works. About 10 minutes.

Configuración única. Construye tu bóveda, instala las herramientas, escribe tu CLAUDE.md y verifica que todo funcione. Unos 10 minutos.


When to run it

  • First install of the kit, or the user asks to set up or initialize the brain.
  • No CLAUDE.md, or one without second_brain_initialized: true.
  • Upgrade: a CLAUDE.md with kit_version: 1.0.0. Then follow "Upgrading from v1" at the end instead of a fresh setup.

Do not run it when the brain is set up and the user is just working: they want another skill. If a setup was left half done, offer to resume it.

Triggers / Activadores

English: "set up my second brain" · "install the starter kit" · "initialize my brain" · "set up my vault" · "I just installed the kit" · "upgrade my brain to v2"

Español: "configura mi segundo cerebro" · "instala el starter kit" · "inicializa mi cerebro" · "configura mi bóveda" · "acabo de instalar el kit" · "actualiza mi cerebro a la v2"

Load first

obsidian-power-user, for the formatting of every note this setup writes.


Step 1. Look around before asking

ls "$WORKING_DIR/CLAUDE.md" 2>/dev/null           # already set up?
ls "$WORKING_DIR" | grep -iE "vault|brain|obsidian" # a vault nearby?
uname -s                                            # macOS, Linux, Windows
python3 -c "import sqlite3; c=sqlite3.connect(':memory:'); c.execute('create virtual table t using fts5(x)'); print('python ok, fts5 ok')"

Use what you find as defaults for the questions. Python is optional. With Python 3 and FTS5 the brain gets real search, the frontmatter check and the monthly health check. Without it, every skill still works with plain text search and manual checks. If it is missing on a Mac, macOS offers to install it the first time python3 runs; on Windows, python.org has an installer. Never block the setup on it.


Step 2. Seven questions

Ask them one at a time (with AskUserQuestion or the host's multiple choice).

Q1. Your name / Tu nombre. "What should I call you?" / "¿Cómo te llamo?" → {{USER_NAME}}

Q2. Your role or business / Tu rol o negocio. Optional. → {{BUSINESS_NAME}}

Q3. Vault name / Nombre de la bóveda. Default "My Second Brain" / "Mi Segundo Cerebro". → {{VAULT_NAME}}

Q4. Where it lives / Dónde vive. Most people pick a synced folder.

  1. iCloud Drive: ~/Library/Mobile Documents/com~apple~CloudDocs/{{VAULT_NAME}} (Mac)
  2. Dropbox: ~/Dropbox/{{VAULT_NAME}}
  3. OneDrive: ~/OneDrive/{{VAULT_NAME}} (Windows)
  4. Documents, local only: ~/Documents/{{VAULT_NAME}}
  5. A path they type → {{VAULT_PATH}}

Q5. Main language / Idioma principal. English / Español / Both. Every skill works in both; this sets the default for headings and summaries. → {{PRIMARY_LANGUAGE}}

Q6. Notion. "Do you also use Notion? project-advisor can mirror its verdicts there." Yes / No / Later. → {{NOTION_ENABLED}}

Q7. Monthly health check / Revisión mensual. "Once a month (the 1st, 07:00) the brain can check itself: broken links, orphan notes, backups left inside, and keep a history. Turn it on?" Yes (recommended) / No, I'll ask for it. → {{HEALTH_SCHEDULED}}. Skip this question if Python is not available.


Step 3. Build the vault

Create {{VAULT_PATH}} with:

{{VAULT_NAME}}/
├── 00 Inbox/
├── 01 Personal Knowledge/   People/ Places/ Routines/ Lessons Learned/
├── 02 Strategy/             Vision/ Goals/ Decision Log/ North Star/
├── 03 Ideas & Notes/        References/
├── 04 Learning/             Books/ Courses/ AI & Tech/ Business/
├── 05 AI System/            Skills/ Integrations/ Architecture/
├── 06 Session Logs/
├── 07 Assets/
├── 08 Projects/
├── 09 MOCs/
├── Archives/
├── Templates/
├── Excalidraw/
└── .brain/                  tools/  vocabulary.json  (hidden from Obsidian)

Sections are found by number prefix, so "01 Personal Knowledge" can become "01 Company Knowledge" later. Then write two starter notes so the graph has a center:

  • Home.md at the root: one paragraph on what this brain is, and a link to each MOC.
  • 09 MOCs/MOC, Start Here.md (type: moc): links to Home, to the session log of this setup, and a short "how to use me" list (the four phrases from Step 8).

Step 4. Check the skills are installed

install.sh copies all nine skills. Confirm they are in the host's skills folder (~/.claude/skills/, ~/.codex/skills/, ~/.gemini/skills/, or where the user's AI loads skills from):

SkillJob
second-brain-initThis setup
obsidian-power-userObsidian formatting and features
save-to-obsidianSave knowledge and the session log
project-advisorScore an idea before committing
brain-searchSearch the brain before asking
project-doorOne current-state note per project
session-checkpointSave point for long sessions
close-sessionClose a session without losing anything
brain-healthMonthly checkup with history

If any is missing, copy it from the kit's skills/ folder.

Step 5. Put the tools in the vault (when Python works)

The tools ship inside this skill, in second-brain-init/tools/. Copy them:

SRC="<skills folder>/second-brain-init/tools"
mkdir -p "{{VAULT_PATH}}/.brain/tools"
cp "$SRC"/*.py "$SRC"/*.sh "{{VAULT_PATH}}/.brain/tools/"
cp "$SRC/vocabulary.json" "{{VAULT_PATH}}/.brain/vocabulary.json"

In .brain/vocabulary.json, set time_required_since to today's date (from the clock), so every note from today on carries the real time and older notes are not flagged for it. Then build the first index:

python3 "{{VAULT_PATH}}/.brain/tools/brain_index.py"

If Python is not available, still copy the folder (it will work the day Python is installed) and say so in the summary.

Step 6. Write CLAUDE.md

Write {{WORKING_DIR}}/CLAUDE.md from this template. Keep it short: it is read at the start of every conversation, and every line costs memory.

# {{VAULT_NAME}}: brain configuration

second_brain_initialized: true
initialized_date: {{TODAYS_DATE}}
kit_version: 2.0.0

## Owner
- Name: {{USER_NAME}} · Role: {{BUSINESS_NAME}} · Language: {{PRIMARY_LANGUAGE}}

## Vault
- Name: {{VAULT_NAME}} · Path: {{VAULT_PATH}}
- Tools: {{VAULT_PATH}}/.brain/tools/ · Notion mirror: {{NOTION_ENABLED}}
- Monthly health check: {{HEALTH_SCHEDULED}}

## Startup, in three layers
1. Always: this file and `{{VAULT_PATH}}/CURRENT-CONTEXT.md`.
2. By topic, before answering:
   - A project is mentioned: read its Door front page first (`project-door`).
   - A past fact, decision or "have we done this": search the brain (`brain-search`).
   - Anything written into the vault: `obsidian-power-user` rules.
3. On demand only: the rest of a Door, checkpoints, old session logs, MOCs.

## Rules
- Search the brain before asking the user for a fact.
- Save only when asked, or offer once at the end of a session with something real.
- Real date and time on every note; frontmatter checked with check_frontmatter.py.
- Never delete in the vault: archive. No .bak copies inside it.
- Every link must resolve; no file names with # ^ [ ] | :

## Phrases
- "save to my brain" / "guárdalo en mi cerebro" → save-to-obsidian
- "should I do this?" / "¿vale la pena?" → project-advisor
- "where are we with X?" / "¿en qué quedamos con X?" → project-door
- "close the session" / "blindemos" → close-session
- "brain health" / "salud del cerebro" → brain-health

## North Star
(Fill this in after a few weeks: what you are building toward. project-advisor uses it.)

Step 7. Seed CURRENT-CONTEXT.md

Write {{VAULT_PATH}}/CURRENT-CONTEXT.md (type: note, with date and time): active projects (each linking its Door once it has one), this week's focus, open questions, recent decisions. Short on purpose: it is read at every start. Keep it under one screen.

Step 8. Turn on the health check

If {{HEALTH_SCHEDULED}} is yes:

bash "{{VAULT_PATH}}/.brain/tools/schedule_health.sh" install
python3 "{{VAULT_PATH}}/.brain/tools/brain_health.py" --write     # first row of the history

Tell the user it runs on the 1st of each month and that "turn off the monthly check" switches it off. On a Mac with the vault in iCloud, mention the Full Disk Access tip the script prints.

Step 9. Verify

V="{{VAULT_PATH}}"
for d in "00 Inbox" "06 Session Logs" "09 MOCs" "Archives" ".brain/tools"; do
  test -d "$V/$d" && echo "OK   $d" || echo "MISSING $d"; done
for s in obsidian-power-user save-to-obsidian project-advisor brain-search project-door session-checkpoint close-session brain-health; do
  test -f "<skills folder>/$s/SKILL.md" && echo "OK   $s" || echo "MISSING $s"; done
grep -q "second_brain_initialized: true" "{{WORKING_DIR}}/CLAUDE.md" && echo "OK   CLAUDE.md" || echo "MISSING CLAUDE.md"
python3 "$V/.brain/tools/check_frontmatter.py" --all --summary   # the starter notes must pass
python3 "$V/.brain/tools/brain_search.py" "start here"           # finds MOC, Start Here

Fix anything missing before calling the setup done.

Step 10. One next step, and the setup log

Give the user exactly one thing to try:

EN: Your brain is ready. Talk about something you are working on, and at the end say "save to my brain".

ES: Tu cerebro está listo. Platica de algo en lo que estés trabajando y al final di "guárdalo en mi cerebro".

Then run save-to-obsidian to log this setup as 06 Session Logs/{{TODAYS_DATE}} Session, Second Brain Setup.md: the answers, what was installed, where it lives. It is the birth certificate of the brain.


Upgrading from v1

When CLAUDE.md says kit_version: 1.0.0, ask once: "Upgrade your brain to v2? Nothing is moved or deleted; I add what is new." If yes:

  1. Add the missing folders: 03 Ideas & Notes/References, 07 Assets, 08 Projects, 09 MOCs, Archives, .brain/tools.
  2. Steps 5, 7 (only if CURRENT-CONTEXT.md is missing) and 8.
  3. Rewrite CLAUDE.md with the v2 template, keeping every line the user added (people, vocabulary, North Star).
  4. Run brain_health.py --write and show the first light. Old vaults usually start yellow or red; that is the baseline, not a failure. Offer the three biggest fixes.
  5. Do not mass-edit old notes to the new contract. They get fixed when they are next touched; the health history shows the progress.

Edge cases

  • The user already has a vault: use it. Add missing folders, never move or delete notes, and say so.
  • No Obsidian installed: the notes are plain markdown; Obsidian adds the graph and clickable links. Suggest obsidian.md when they are ready.
  • Re-run by accident: stop and ask: keep it, change some answers, or start over. Never overwrite silently.
  • Windows: use OneDrive or Documents; the health schedule prints a schtasks line.
  • The host cannot run commands: create the folders and notes by hand, skip the tools, and tell the user what they are missing.

How to talk during setup

Plain words, concrete steps ("Creating the vault in iCloud...", "Building the search index..."). At the end give three things only: a short confirmation in the user's language, the verification lines, and the one next step.

What this skill does not do

  • It does not write the user's knowledge; that is save-to-obsidian.
  • It does not import notes from Notion, Apple Notes or Evernote; obsidian-power-user has the import guide.

Changelog

  • 2.0.0 (2026-10-02): Installs nine skills. Tools in .brain/ (index, search, frontmatter contract, Doors, health). Folders 07, 08, 09, Archives, References. CLAUDE.md with three-layer startup and rules. Monthly health check (Q7). Home and Start Here notes. Upgrade path from v1. Python optional throughout.
  • 1.0.0 (2026-05-29): First public version.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.