结果监控器：针对静默工具故障的恢复可行性
Outcome Monitors: Recovery Affordances for Silent Tool Failures
AI总结：
该研究针对智能体决策中的静默工具故障问题，提出结果监控器方法，可检测结果契约违反并提供恢复工具，显著提升ToolMaze和tau-bench任务的完成率。
AI中文摘要：
当工具调用超时，智能体（agent）会识别故障并绕过它。而缓存的错误页面或负价格可能以预期格式到达，并被当作事实使用。我们提出结果监控器（Outcome Monitors），用于检测从任务无关轨迹中挖掘或从公开模式中衍生的结果契约的违反情况。当发生违反时，监控器会保留结果，并发布一份非约束性回执，指明被违反的属性及公开的恢复工具。在注入故障的冻结、预先指定的评估中，结果监控器使四个模型在两个提供商系列的ToolMaze任务完成率从10.9%提升至28.1%，并在第三个提供商系列中得到复现；在tau-bench零售任务中，两个层级的完成率分别提升14.0和12.0个百分点。在单独的ToolMaze对照实验中，移除恢复工具列表会消除测得的增益，恢复该列表则可恢复效果；诊断细节和时间未产生可检测差异，增益集中在故障阻碍完成的场景。在从已发布事件分类法转录的一组测试中，挖掘词汇外的检测率降至46%，但交付仍继续且完成率不变。恢复工具是这些对照中活跃的回执内容，将检测扩展至契约词汇之外仍是未解决的问题。
英文摘要：
When a tool call times out, the agent sees the failure and can route around it. A cached error page or negative price can instead arrive in the expected format and be consumed as fact. We introduce Outcome Monitors, which detect violations of outcome contracts mined from task-disjoint traces or derived from public schemas. On a violation, the monitor preserves the result and issues a nonbinding receipt naming the violated property and public recovery tools. In frozen, prespecified evaluations with injected failures, Outcome Monitors raise ToolMaze completion from 10.9% to 28.1% across four models in two provider families and replicate in a third. In tau-bench retail, completion improves by 14.0 and 12.0 points on two tiers. In separate ToolMaze controls, removing the recovery-tool list eliminates the measured gain and restoring it recovers the effect; diagnostic detail and timing produce no detectable differences. Gains concentrate where the fault blocks completion. On a suite transcribed from a published incident taxonomy, detection outside the mined vocabulary falls to 46%, though delivery continues and completion is unchanged. Recovery tools are the active receipt content in these controls; extending detection beyond the contract vocabulary remains open.
