Brainoutside

BrainOutside

**Your brain, kept outside your head — where your agents can read it.**

A self-hosted memory server for your AI agents. One git repo full of markdown is your brain; BrainOutside serves it over REST and MCP, with visibility tiers enforced server-side and a human gate on every write.

Single-user and single-brain by design. That is the product, not a limitation waiting to be fixed.

**Two ways to run a brain, one repo between them.** Start local: clone [brainoutside-template](https://github.com/hassancs91/brainoutside-template), open it in VS Code, and Claude Code is the whole interface — free, private, zero infrastructure. Or self-host this server as the brain's online head, so every agent you run anywhere can read your mind over MCP and REST. They compose: the local repo IS the repo the server clones. Start local today, add the server when you want it, migrate nothing. Site and docs: [brainoutside.com](https://brainoutside.com).

Status: pre-release. The engine and the first-run wizard are built and verified end to end; what remains before the public beta is packaging, docs, and launch assets.

📖 Read the guide

The full walkthrough lives on my site: what a second brain is actually for, why plain markdown beats a vector database for personal knowledge, and how to go from an empty repo to a self-hosted brain your agents can read. **[How to Build an AI Second Brain](https://learnwithhasan.com/guide/ai-second-brain/?utm_source=github&utm_medium=readme&utm_campaign=brainoutside&utm_content=body)**

Free to read, no login.

Where this came from

This project grew out of [Andrej Karpathy's llm-wiki](https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f) and the wave of markdown knowledge bases around it. The core insight there is right, and it is the foundation this builds on: plain markdown plus a coding agent beats RAG for personal knowledge — compile knowledge in at ingestion time and the artifact compounds, instead of being re