🧠 the-brain

An open memory platform for AI — in the making. 3-layer cognitive architecture with an extension-first philosophy. Runs locally, costs nothing, and lets you swap any component.

🧪 Experimental — pre-alpha. Far from stable. This repository is for experimental purposes only. Do not use in production. the-brain explores what persistent, private, 3-layer AI memory looks like. Breaking changes, missing features, and rough edges are the norm. If you're curious about the concept, contribute on GitHub or build an extension.

Why the-brain?

AI forgets everything between sessions. the-brain remembers — across any AI tool you use. It harvests your conversations, filters noise from signal, and consolidates what matters. No cloud. No telemetry. Just a background daemon learning your patterns.

Quick Links

Architecture at a Glance

Your AI tools feed into three plugin slots:

Any AI tool → ⚡ INSTANT slot — what happens now (default: graph-memory)
            → ⚖️ SELECTION slot — what's worth keeping (default: spm-curator)
            → 🌌 DEEP slot — permanent consolidation (default: mlx-lora + auto-wiki)

Each slot is replaceable. Add extension names to config.json ("extensions": ["my-ext"]) and daemon loads them on start. Extensions are disabled by default for security — you must explicitly enable each one.

What the-brain Learns

• Corrections — "No, actually use useCallback here" → remembers for next time
• Preferences — "I prefer arrow functions" → injected into every context
• Patterns — Repeated choices across sessions → auto-suggests related concepts
• Identity — Your stable preferences and style, maintained across retrains

Documentation Sections

Tech Stack

License

MIT © 2026 Oskar Schachta