AgentDB Memory Patterns

What problem does it solve? AI agents lose context between sessions and cannot learn from past interactions, forcing developers to rebuild memory, retrieval, and learning infrastructure from scratch for every stateful agent or chat system. ## Core Features & Use Cases - Session and Long-Term Memory: Store conversation history, user facts, and preferences in a persistent vector database with HNSW indexing for sub-millisecond retrieval. - Pattern Learning with ReasoningBank: Record successful interaction patterns, retrieve them with reasoning-based context synthesis, and train learning plugins such as Decision Transformer, Q-Learning, and Actor-Critic. - CLI and MCP Integration: Initialize databases, run queries, import/export vectors, benchmark performance, and expose memory tools to Claude Code through an MCP server. - Use Case: Build a customer support assistant that remembers each user's preferences across sessions, retrieves relevant past resolutions via vector search, and improves its responses by training on successful interactions. ## Quick Start Set up an AgentDB memory store for my agent and show me how to save conversation history and retrieve relevant past context with vector search.