ModelRefs / Short-Term Memory — Agent Pattern
Short-Term Memory — Agent Pattern
Maintain a rolling message buffer with summarization to fit the agent's working context. Cheapest memory tier and the default for single-session agents.
Overview
Append every turn to a buffer; when token budget is exceeded, summarize the oldest turns and keep the summary plus recent turns. Cheapest memory tier and the default for single-session agents.
When to use it: Your agent needs coherent multi-turn conversation within a single session.
Pattern details
| Pattern class | memory |
|---|---|
| Difficulty | beginner |
| Autonomy | guided |
| Also known as | conversation buffer, rolling memory |
| Last reviewed | 2026-06-07 |
Known failure modes
- Summary loss — Important detail dropped during summarization. Mitigation: Pin critical facts; use structured summaries with required fields.
When not to use it
- Storing PII in plaintext buffers.
Continue your research
Use these connected ModelRefs sections to compare alternatives, inspect implementation paths, and review the evidence and governance boundaries relevant to Short-Term Memory — Agent Pattern.
Frequently asked questions
When should I use the Short-Term Memory agent pattern?
Your agent needs coherent multi-turn conversation within a single session.
What are common failure modes of Short-Term Memory?
Summary loss
Is Short-Term Memory production-ready?
Yes when paired with the safety controls and observability hooks documented on the pattern page.