Architecture for Intelligent Networked Growth
Dual-Brain Knowledge Metabolism Engine
Order Brain × Growth Brain
🧠 双脑代谢 · 🌿 意识神经 · 💬 会话入库 · 📦 一键部署
让知识库自己活着 —— 进入主动代谢时代
Two subsystems, one shared infrastructure — zero external services
Stable storage & structured governance
Active metabolism & non-linear evolution
SQLite (Graph + FTS5) · Node Single Process · Zero Dependencies
No PostgreSQL · No Redis · No Docker · No npm install
Knowledge doesn't just sit — it lives, grows, and renews
From compilation to metabolism — the paradigm shift
| Dimension | Karpathy LLM Wiki | aing |
|---|---|---|
| Paradigm | Knowledge Compilation | Knowledge Metabolism |
| Growth | Linear: ingest→compile→query | Non-linear: self-sustaining evolution |
| LLM Role | Single "programmer" | Dual-brain: compile + metabolize |
| Automation | Manual trigger | Time-triggered, automated |
| Upper Limit | ~100-200 sources / 50K tokens | Theoretically unbounded |
| Git | Manual init required | ✅ First-class native |
| Knowledge Liveness | Static after compile | ✅ Active expiry · pruning · regeneration |
Zero dependencies. Just Node.js. One command to start.