brainoutsourcing

Your AI-powered Second Brain. Clone, /setup, answer 3 questions — done.

brainoutsourcing combines the command-driven session workflow introduced by KI-OS with the LLM knowledge structuring approach described by Andrej Karpathy, and extends both into a structured, persistent second brain system.

The goal is not to replace existing systems, but to make them more usable, more consistent, and easier to adopt — within minutes.

Prerequisites

• Claude Code (CLI, Desktop App, or IDE Extension)
• Optional: Obsidian as a viewer for your wiki

Setup (30 seconds)

git clone https://github.com/josudia/brainoutsourcing.git my-brain
cd my-brain
claude   # or: open Claude Code in this folder

Then in Claude Code:

/setup

Answer 3 questions (who you are, what you do, what you're working on) — Claude sets up everything.

What /setup does

• Creates your personal folder structure (context, wiki, plans, raw, personal)
• Generates context files from your answers
• Personalizes the CLAUDE.md (Claude's working instructions)
• Sets up an empty wiki with index
• Creates your Kanban board

Commands

Command	When to use
/checkin	Start every new session — load context, see status
/checkout	End session — save progress, write daily log
/add-todo	Quickly capture an idea or task
/add-wiki	Add a new source (URL, file) to the wiki
/review	Weekly review — clear inbox, plan the week
/create-plan	Create a detailed implementation plan
/implement-plan	Execute a plan step by step
/backlog	Kanban board: In Progress, Prioritized, Backlog
/wiki-health	Wiki health check (orphans, broken links)
/nightshift	Prepare an autonomous overnight run (optional, macOS)

How this differs from the original projects

brainoutsourcing is a synthesis of existing ideas, extended in three key ways:

• Persistent knowledge system: Extends the command-driven session approach from KI-OS into a long-lived wiki, context layer, and planning system that compounds over time.
• Operationalized LLM Wiki: Takes Karpathy's concept of LLM-compiled knowledge (raw sources → structured wiki pages, lint, cross-references) and integrates it into a practical daily workflow with /add-wiki, /wiki-health, and session-based ingestion.
• Near-zero setup: Reduces onboarding to a single /setup command with 3 questions, making the system accessible to non-technical users within seconds.

The focus is on usability and integration, not reinvention.

Architecture

brainoutsourcing/
├── CLAUDE.md              <- Claude's working instructions (personalized after /setup)
├── context/               <- Who you are, what you do, your strategy
├── wiki/                  <- LLM-maintained knowledge base (Karpathy pattern)
│   ├── concepts/          <- Technical and domain concepts
│   ├── tools/             <- Software and frameworks
│   └── people/            <- People and sources
├── plans/                 <- Kanban board, WIP, implementation plans
├── raw/                   <- Sources (articles, transcripts, papers) — immutable
├── personal/              <- Daily logs, health, finances — private
├── reference/             <- Reference materials
├── scripts/               <- Automation
├── outputs/               <- Generated artifacts
└── projects/              <- Own projects (each with its own git)

Credits

Full credits and license details: CREDITS.md

License

MIT — see LICENSE