Agent_Memory_Techniques: 30 Runnable Notebooks Covering All Major Agent Memory Approaches

Agent_Memory_Techniques is a GitHub repository that organizes 30 memory techniques into 30 directly runnable notebooks, from conversation caching to production-grade solutions. It covers vector databases, knowledge graphs, episodic memory, semantic memory, and popular frameworks like Mem0, Letta, Zep, and Graphiti. It includes evaluation benchmarks for long-conversation memory, enabling data-driven selection. A decision tree in the README guides users to the most relevant notebook based on their scenario, reducing the need to study all 30. The resource is aimed at developers designing memory layers for AI agents.