A standalone memory management system for AI context and conversation state. Octobrain solves the "forgetting" problem by providing a persistent, semantic memory layer that lives outside the LLM's limited context window. It combines vector storage (LanceDB) with a knowledge graph, allowing AI agents to memorize decisions, user preferences, and technical insights. It implements a temporal decay system (Ebbinghaus forgetting curve) to ensure the most relevant and reinforced information surfaces first. - Semantic Memory: Natural language retrieval of past decisions and insights. - Knowledge Base: Index and search web content for persistent reference. - Memory Graph: Maps relationships between memories (depends_on, supersedes). - Temporal Decay: Intelligent importance weighting based on access and time.