# Agentic Memory
Agentic memory is durable information an agent can preserve, retrieve, and update beyond a single model context window. It may include structured records, working notes, summaries, decisions, provenance, and task state.
XDO's document, account, and system variables resemble a scoped persistent-memory design, but the historical system stored deterministic values rather than model-generated memories. The useful modern translation is the separation of local, shared, and global state with explicit promotion and access rules.
Related: [[wiki/Layered Context Architecture|Layered Context Architecture]], [[wiki/Durable Execution|Durable Execution]], [[wiki/Provenance|Provenance]].
Source: [Anthropic — Effective context engineering for AI agents](https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents)