` 包裹。 * 长文用 `
`/`
` 划分小节（虽然原文只是一个非常长的单段，但我需要将其结构划分为合乎逻辑的小节，并应用标题，或者如果作为单篇报道，则将其视为一篇长文并正确包裹段落。仔细看，原文本质上是一整段。我需要将其拆分为合乎逻辑的段落/小节以应用 `
`/`
`/`
`）。 * 列表用 `
- `。 * 引用用 `
`。 * 代码用 `
`。 * 重要句子用 ``。 * 内容规则： 保留原文所有信息，不得增删改内容。 3. 分析源文本： “Tool-use benchmarks often score agents by final task success, but terminal outcomes can obscure whether the intermediate tool-call trajectory was executable. We propose prefix executability as a trajectory-level evaluation protocol for tool-using agents. The protocol measures whether every prefix of a predicted tool-call sequence remains executable under tool schemas, state preconditions, and data dependencies, and summarizes reliability using Prefix Executability Curves (PEC), First-Error Depth (FED), Trajectory Executability Rate (TER), horizon-normalized AUC-PEC, and Recovery Rate (RR). We instantiate the protocol on stratified samples from BFCL-v3 multi-turn and ToolSandbox, and on the complete Airline/Retail base splits of $τ^3$-bench. We observe that outcome-only and prefix-level rankings diverge: agents whose execution is conditioned on an unverified plan often achieve higher point-estimate final success than non-plan-conditioned baselines while substantially lowering TER and AUC-PEC. These divergences are precisely the failure modes that prefix executability is designed to expose: they reveal whether apparent success rests on an executable trajectory, where invalidity first enters, and whether success follows recovery from an earlier failure. As a validation intervention, explicit pre-execution plan checking can improve executable-prefix survival while maintaining competitive terminal success. These results show that final success alone is insufficient for evaluating tool-use agents and that prefix-level executability provides an actionable reliability signal for comparing agents, diagnosing failures, and assessing validation mechanisms.” 4. 为HTML制定结构计划： * 原文是一个大段落。为了有效应用“长文用 ``/`
` 划分小节”的规则，我应该将逻辑流程拆分为标题和小节。 * *标题1：* 背景与提出方案 (`
`) * 关于基准测试掩盖轨迹的段落。 * 提出前缀可执行性的段落。 * *标题2：* 评估指标 (`
`) * 列出指标的段落。*等等，规则说“列表内容用 `
- `”。* 原文列出了指标：“Prefix Executability Curves (PEC), First-Error Depth (FED), Trajectory Executability Rate (TER), horizon-normalized AUC-PEC, and Recovery Rate (RR)”。我可以将此列表提取到 `
-
` 中，以严格遵守列表规则。 * *标题3：* 实验设置 (`
`) * 讨论在 BFCL-v3、ToolSandbox 和 $\tau^3$-bench 上实例化的段落。 * *标题4：* 实验发现与分析 (`
`) * 关于排名差异的段落。 * 关于这些差异揭示了哪些失败模式的段落。 * *标题5：* 验证干预与结论 (`
`) * 关于计划检查的段落。 * 关于最终结果的段落。 5. 起草HTML并应用规则（迭代完善）： * *拆分文本：* 1. “Tool-use benchmarks often score agents by final task success, but terminal outcomes can obscure whether the intermediate tool-call trajectory was executable.” -> `
研究背景与方法提出
` (等等，规则说“不得增删改内容”。添加像“研究背景与方法提出”这样的中文标题算不算“增删内容”？严格来说，是的。提示词说“保留原文所有信息，不得增删改内容”。添加标题算增加内容吗？为了安全起见，如果我需要使用标题，我应该使用原文中的词作为标题，或者干脆不使用标题，只用 `` 和 ``。但是规则2明确要求“长文用 `
`/`
` 划分小节”。这是典型的约束冲突：我不能添加标题（增内容），但我必须使用 `
`/`
`。解决方法：使用原文中的句子片段作为标题，或者使用像“Abstract”或章节名称这样的通用标题。更好的是，直接把原文的句子作为 `
` 或 `
`？不，标题通常是简短的标签。让我们用原文中的关键短语来制作标题，而不添加外部词汇。例如：`
Prefix Executability Protocol
`，`Reliability Metrics
`。等等，即使是那样也是增加内容。如果我不能增加任何内容，我该如何划分小节？我将使用原始文本中已有的句子作为标题。 让我们重读一下：“保留原文所有信息，不得增删改内容”。如果我添加 `Introduction
`，我就是增加了“Introduction”这个词。为了绝对安全，我将使用原文中的现有句子作为标题，或者将它们包裹在 ``/`
` 标签中。 让我们尝试将原始句子映射到标题： `
Tool-use benchmarks often score agents by final task success, but terminal outcomes can obscure whether the intermediate tool-call trajectory was executable.
` 这有点难看，但在技术上满足了“长文用 ``/`
` 划分小节”和“不得增删改内容”的要求。 让我们寻找更好的方法。如果我把现有的短语作为标题呢？ `
Prefix executability as a trajectory-level evaluation protocol
` 让我们尝试构建它，使每个 ``/`
` 都是文本中确切的子字符串。 让我们将文本拆分为逻辑块： 块1：“Tool-use benchmarks often score agents by final task success, but terminal outcomes can obscure whether the intermediate tool-call trajectory was executable.” 块2：“We propose prefix executability as a trajectory-level evaluation protocol for tool-using agents.” 块3：“The protocol measures whether every prefix of a predicted tool-call sequence remains executable under tool schemas, state preconditions, and data dependencies, and summarizes reliability using” + [列表] + “.” 块4：“We instantiate the protocol on stratified samples from BFCL-v3 multi-turn and ToolSandbox, and on the complete Airline/Retail base splits of $τ^3$-bench.” 块5：“We observe that outcome-only and prefix-level rankings diverge: agents whose execution is conditioned on an unverified plan often achieve higher point-estimate final success than non-plan-conditioned baselines while substantially lowering TER and AUC-PEC.” 块6：“These divergences are precisely the failure modes that prefix executability is designed to expose: they reveal whether apparent success rests on an executable trajectory, where invalidity first enters, and whether success follows recovery from an earlier failure.” 块7：“As a validation intervention, explicit pre-execution plan checking can improve executable-prefix survival while maintaining competitive terminal success.” 块8：“These results show that final success alone is insufficient for evaluating tool-use agents and that prefix-level executability provides an actionable reliability signal for comparing agents, diagnosing failures, and assessing validation mechanisms.” 让我们使用原始文本的确切片段作为标题。
- `”。* 原文列出了指标：“Prefix Executability Curves (PEC), First-Error Depth (FED), Trajectory Executability Rate (TER), horizon-normalized AUC-PEC, and Recovery Rate (RR)”。我可以将此列表提取到 `