Sentry：在测试时学习从LLM智能体故障中恢复

中文摘要 AI 辅助

LLM智能体经常因无效的工具调用、重复的动作或缺乏依据的推理而在任务中途失败，从这些失败中学习是通往可靠性的途径。我们发现，失败知识到达智能体的方式与其内容同样重要。失败教训是有条件的：保留在智能体的上下文中时，当失败不存在时它们会误触发，而从不断演进的剧本中移除它们会提升性能。相比之下，运行时干预仅在失败发生时起作用，但不会从修复中学习。我们认为失败知识是条件性知识，应当有条件地暴露，并将这一原则实例化在Sentry中，这是一个与智能体并行运行的失败管理层。当Sentry检测到失败时，它从外部剧本中检索匹配的教训以指导恢复，在无法访问任务奖励的情况下验证智能体是否已恢复，并且仅在恢复成功时存储新教训；完整剧本从不进入智能体的上下文。在多个智能体基准测试中，Sentry在每个基准上都优于最强的运行时干预基线，平均提升37%，在两个同时评估的基准上比最强的上下文演进基线提升39%；将Sentry与上下文演进结合可获得进一步收益。学到的教训可迁移到保留任务，受控实验表明，即使相关教训按需可用，向智能体暴露完整剧本也会降低性能。

英文摘要

LLM agents often fail mid-task due to invalid tool calls, repeated actions, or poorly grounded reasoning, and learning from these failures is a path to reliability. We find that how failure knowledge reaches the agent matters as much as what it contains. Failure lessons are conditional: kept in the agent's context, they misfire when their failure is absent, and removing them from an evolving playbook improves performance. Runtime interventions, in contrast, act only when a failure occurs but do not learn from their repairs. We argue that failure knowledge is conditional knowledge and should be conditionally exposed, and instantiate this principle in Sentry, a failure-management layer that runs alongside the agent. When Sentry detects a failure, it retrieves matching lessons from an external playbook to guide recovery, verifies without access to task rewards whether the agent recovered, and stores a new lesson only if it did; the full playbook never enters the agent's context. Across multiple agentic benchmarks, Sentry outperforms the strongest runtime-intervention baseline on every benchmark, by 37\% on average, and the strongest context-evolution baseline by 39\% on the two benchmarks where both are evaluated; combining Sentry with context evolution yields further gains. Learned lessons transfer to held-out tasks, and controlled experiments show that exposing the full playbook to the agent lowers performance even when relevant lessons remain available on demand.

发表机构

• Stanford University（斯坦福大学）

机构由 AI 辅助整理，请以论文原文为准。