Unverified50% confidenceFactExact time
An Agent's short-term memory is typically the large model's context window, which has limited capacity and will be truncated or compressed as the number of conversation turns increases.
1
Sources
50%
Confidence
Long-term
Relevance
9/18/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIn multi-step tasks, an Agent loses context, which is fundamentally a state management problem, because information relies on the context window's memory, and key information is easily buried or truncated as the number of conversation turns increases.75% similarUnverifiedThe main reasons Agent frameworks are slower than direct conversation are that system prompts inject large numbers of tool definitions, multi-round ReAct reasoning loops occur, and context accumulation causes the number of input tokens to surge.74% similarUnverified当前大模型并没有真正的持久化记忆机制，随着对话轮次增加，早期需求描述在注意力机制中的权重会逐渐衰减72% similarUnverifiedAgent交接时传递上下文过多会增加Token消耗和延迟，传递过少则导致接手Agent缺乏关键背景72% similarUnverifiedNative large models are constrained by context window size and tend to forget early information when handling ultra-long tasks.71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/931869API
curl https://kongchang.com/api/v1/knowledge/claims/931869MCP
get_claim(id=931869)