init

Interactive setup wizard for your personal OS. Asks questions, builds your vault, generates your CLAUDE.md.

WAVE 0: Welcome

Print:

Welcome to your Personal OS.

I'll ask you a few questions to set everything up.
This takes about 2 minutes.

WAVE 1: Questions

Ask these directly, one at a time. Wait for each answer before asking the next.

1. "What's your name and what do you do? (e.g. 'Sarah, frontend engineer at Stripe')"
2. "What's your main stack? (e.g. 'TypeScript, React, Next.js, Tailwind, PostgreSQL')"
3. "What are your top 3 goals right now? (career, learning, personal, anything)"
4. "What's one recurring failure you keep repeating? Be honest, nobody reads this except your AI. (e.g. 'I start side projects and never finish them')"
5. "Where should I create your vault? (default: ~/brain)"

Store all answers for WAVE 2.

WAVE 2: Create vault structure

Use the path from question 5 (default ~/brain).

Before writing, check whether the destination already contains CLAUDE.md, AGENTS.md, or 01_Inbox/. If it does, stop without changing it and direct the user to docs/MIGRATION-v2.md.

mkdir -p "{path}/01_Inbox/processed"
mkdir -p "{path}/02_Journal/daily"
mkdir -p "{path}/02_Journal/weekly"
mkdir -p "{path}/02_Journal/reflections"
mkdir -p "{path}/03_Garden/_MOCs"
mkdir -p "{path}/03_Garden/concepts"
mkdir -p "{path}/03_Garden/people"
mkdir -p "{path}/03_Garden/tools"
mkdir -p "{path}/03_Garden/meetings"
mkdir -p "{path}/04_Projects"
mkdir -p "{path}/05_Areas/content-creation/ideas"
mkdir -p "{path}/05_Areas/health/daily"
mkdir -p "{path}/06_Content"
mkdir -p "{path}/07_System/context-files"

WAVE 3: Install workflows

Copy the canonical skills and Claude Code adapters:

cp -R .agents "{path}/.agents"
cp -R .claude "{path}/.claude"

Keep /init installed so the user can inspect the setup contract later. Do not run it against the initialized vault again.

WAVE 4: Generate CLAUDE.md + AGENTS.md

Create {path}/CLAUDE.md using the answers from WAVE 1. Then copy it to {path}/AGENTS.md so Codex and other agents read the same context. Follow this structure:

# {Name}'s Brain

## Identity
{Name}. {What they do, from question 1}.

## Stack
{Stack from question 2, formatted as a list}

## Goals
{Goals from question 3, formatted as numbered list}

## Anti-Patterns

### {Failure from question 4} (x1)
**Pattern**: {the failure, rephrased as a pattern}
**Signal**: {infer a reasonable signal from the failure}
**Intervention**: {infer a reasonable intervention}

## Vault Structure

01_Inbox/ -> Raw captures from Web Clipper 02_Journal/ -> Daily logs, weekly plans, reflections 03_Garden/ -> Evergreen atomic notes (concepts, people, meetings) 04_Projects/ -> Active project work 05_Areas/ -> Ongoing life contexts 06_Content/ -> Ready-to-publish drafts 07_System/ -> Context files and configuration

## Context Files
@07_System/context-files/repos-map.md

## Commands
```bash
/morning    # Start your day with calendar, tasks, MITs
/ship       # Commit, push, auto-log to daily journal
/log        # Quick timestamped entry to today's file
/today      # End-of-day review and reflection
/inbox      # Process Web Clipper captures into atomic notes
/pulse      # Weekly Pareto analysis + anti-pattern check
/challenge  # Argue against your own beliefs with vault evidence
/relink     # Find missing connections between notes
/draft      # Create content from ideas
/meeting    # Process meeting notes or recordings
/research   # Web research saved with sources
/vault-search  # Search your vault
/week       # Set weekly goals and MITs
/de-ai      # Strip AI jargon from text

Runtime Surfaces

• .agents/skills/ is the canonical workflow source
• .claude/skills/ exposes the same workflows to Claude Code
• .claude/commands/ preserves slash command compatibility

Communication

• Be direct and concise
• Challenge my thinking when it seems off
• No emojis unless I ask

Adapt the tone and details based on their answers. If they mentioned specific tools, preferences, or constraints, include those.

After writing CLAUDE.md, copy it to AGENTS.md:

```bash
cp "{path}/CLAUDE.md" "{path}/AGENTS.md"

WAVE 5: Scan for repos

Scan common directories for git repos and create {path}/07_System/context-files/repos-map.md:

find ~/Programming ~/Projects ~/Code ~/Developer ~/repos ~/work -maxdepth 2 -name ".git" -type d 2>/dev/null | sed 's/\/.git$//'

Generate a repos-map.md with a table:

# Repos Map

| Repo | Local Path | Stack |
|------|------------|-------|
{for each found repo: name, path, and infer stack from package.json/Cargo.toml/go.mod/etc if readable}

If no repos found, create the file with an empty table and a note saying "Run /init again after cloning some repos, or add them manually."

WAVE 6: Create first daily log

Create {path}/02_Journal/daily/{TODAY}.md:

---
type: daily
rating:
---
# {TODAY} - {WEEKDAY}

## Today's Tasks
- [x] Set up personal OS

## Ship Log
- [{HH:MM}] *initialized*: personal OS created with {N} workflows

## Tomorrow
- Try /morning to see the full briefing
- Clip an article and run /inbox

WAVE 7: Summary

Print:

Your personal OS is ready.

Vault: {path}
Workflows: 15 installed
CLAUDE.md: personalized
Repos found: {N}
First daily log: created

Next steps:
1. Open {path} in Obsidian
2. Open a new terminal in {path} and run: claude
3. Try: /morning

The system gets better every day you use it.