Agent Memory Systems
State persistence architecture dividing context into short-term working context, long-term vector memory, and episodic logs.
Mechanism & Definition
The state management subsystem of an AI agent. Comprises working short-term memory (in-context prompt buffer), long-term semantic memory (vector database of past interactions), and episodic/procedural memory (structured logs of successful tool sequences).