Context window hygiene for long-running LLM agents. Decision rules for when and how to compress, mask, switch, or delegate context — backed by research (JetBrains/NeurIPS 2025, OpenHands, Letta/MemGPT, LLMLingua). Use when an agent runs for extended sessions, accumulates large tool outputs, approaches context limits, or suffers from compaction/overflow. Also use when designing agent architectures that need to manage context over time.

长时间运行大语言模型代理的上下文窗口维护。基于研究（JetBrains/NeurIPS 2025、OpenHands、Letta/MemGPT、LLMLingua），提供在何时以及如何压缩、掩码、切换或委派上下文的决策规则。适用于代理运行时间较长、工具输出累积过多、接近上下文限制或出现压缩/溢出问题的场景，也适用于需要长期管理上下文的代理架构设计。