Memory
Memory
Information written to persist across sessions, outside any one context window
✓ Concepts & BehaviorWhat is Memory
Memory, in an agent context, is information deliberately written somewhere outside the session so it survives into future sessions that never saw the original conversation, a project instructions file like CLAUDE.md, a running log of learnings, a notes file the agent updates as it works. Without something like this, every new session genuinely starts blank.
This is not automatic, and that's worth being explicit about: a model does not remember a previous session just because it happened, the only continuity that exists is whatever got written down and gets read back in as part of a later session's context. Memory is the deliberate practice of making that continuity real instead of assumed.
Memory is broader and more durable than a handoff, a handoff is usually scoped to one piece of work and read once by whoever picks it up next, while memory tends to be standing information relevant across many future sessions, house rules, recurring gotchas, project conventions, that keeps getting read in every time, not just once.
Build your own agent skills, not just prompts.
We design, build, and ship the Claude Code skill libraries production teams actually run on.
Talk to AY Automate ↗