agent memory management

4 articles · 13 co-occurring · 0 contradictions · 0 briefs

Context rot relates to what information agents retain/lose across conversation segments; implies need for deliberate memory architecture

Related concepts

multi agent orchestration 2 memory persistence 2 context window optimization 2 context window management 2 state persistence across sessions 1 state machine patterns 1 prompt optimization 1 prompt engineering 1 multi turn conversation management 1 multi turn conversation degradation 1 context window scaling 1 context collapse 1 api abstraction layers 1

Evidence chain (4 articles, showing 4)

LangGraph Crash Course - Agent Workflows in Python - YouTube example_of

The explicit 'Adding Memory' section at 13:28 demonstrates that multi-turn agent persistence is a learnable, discrete component—directly exemplifying the memory management pattern.

Agentic Context Engineering: Evolving Contexts for Self-Improving Language Models | OpenReview supports

ACE framework addresses how agent memory should be structured and evolved—both offline (system prompts) and online (agent memory during execution).

Context Rot: Why AI Gets Worse the Longer You Chat (And How to Fix It) extends

Context rot relates to what information agents retain/lose across conversation segments; implies need for deliberate memory architecture

@charlespacker: stateful agents connected through stateless APIs 🔄 extends

Implies architectural decisions about where agent memory lives and how it survives API consumption patterns

query this concept

$ db.articles("agent-memory-management")

$ db.cooccurrence("agent-memory-management")

$ db.contradictions("agent-memory-management")