CompressAgent 新基准显示：35% 预算下压缩方法成功率从 19.9% 到 47%，选压缩法别只盯省 token。
CompressAgent 基准覆盖 9 个代理控制上下文（ACC）、3 个任务族、3 个 Qwen API 模型和 6 档保留上下文预算，共运行 15,525 次。在保留 75% 上下文时，通用改写和章节式压缩分别达到 92.7% 和 92.4% 成功率，接近完整上下文的 93.8%。当预算降到 35%，章节式、义务感知式、通用改写成功率分别为 47.0%、39.0% 和 19.9%。压缩失败主要表现为工具执行和动作解析错误，且不同 ACC 间可靠性差异显著。
Control Under Compression: Reliability Frontiers for Tool-Using Agents
Tool-using language-model agents are governed not only by task prompts but also by persistent system-side instructions that specify tools, arguments, policies, execution protocols, and recovery. Compressing these agent control contexts (ACCs) can reduce input cost and context use, yet existing prompt-compression evaluations do not reveal whether the resulting control remains operationally reliable. We introduce CompressAgent, an environment-verified benchmark for ACC compression across nine independently constructed ACCs, three task families, three fixed Qwen API model identifiers, six retained-context budgets, and 15,525 runs. We uncover a nonlinear, method-dependent reliability frontier. At 75% retained context, generic rewriting and section-based compression achieve 92.7% and 92.4% success, close to the 93.8% full-context baseline. Between 50% and 35%, methods diverge sharply; at 35%, section-based, obligation-aware, and generic rewriting achieve 47.0%, 39.0%, and 19.9%. At retained-context budgets from 25% to 10%, executable protocols become fragile. Reliability also varies substantially across ACCs, making universal compressor rankings inappropriate and motivating per-context qualification. Failure analysis shows that compression primarily surfaces as tool-execution and action-parsing errors. These findings recast ACC compression from token reduction into a runtime-reliability problem that must be evaluated through executable outcomes.