Hebbian Mind

This is a knowledge graph that learns from use. Save content and it activates concept nodes (security, architecture, deployment, etc.), then strengthens edges between co-activated concepts using Hebbian learning. Query by nodes to surface related memories. Edges decay when unused, so the graph adapts to what you actually reference. It exposes eight MCP tools including save_to_mind, query_mind, get_related_nodes, and analyze_content for previewing activations. Ships with 100+ enterprise concepts and optional FAISS integration for semantic search. Runs on stdio with optional RAM disk mode for faster reads via dual-write pattern. Memories decay over time unless accessed, with configurable thresholds and immortal mode for critical content.