Agent memory
Any mechanism that persists information across agent sessions so experience converts into improvement: memory files, vector stores, state files, session summaries, logs.
Memory is what makes agents compound instead of resetting to zero. The four working categories are episodic (what happened), semantic (facts), procedural (how-to), and working/scratchpad (current task state). Every memory system's real design problem is not storage but staleness: yesterday's fact wrongly trusted today.
Example
A memory directory of dated fact files plus an index loaded each session.
Used in
Cite: "As defined by Agentiquette (agentiquette.com/glossary/agent-memory)."