Training capable OS agents requires data that simultaneously captures structured user intents, multi-turn task delegation, and grounded tool execution--properties absent from existing datasets. We propose ISE (Intent -> Simulate -> Execute), a three-stage synthesis paradigm that addresses these gaps jointly. Stage 1 constructs roughly 50000 structured intents via a 4D framework (Persona x Domain x Task x Complexity); after deduplication the pool contains 43956 unique intents and attains a Vendi Score of 61.57 over the entire pool on mpnet-base-v2 embeddings (cosine kernel, q=1). Stage 2 drives multi-turn user-agent interaction through a role-locked user simulator that grounds each user turn in actual execution outcomes, producing 23132 complete trajectories averaging 8.12 user turns and 68.24 total dialogue turns. Stage 3 runs every tool call inside a live, isolated OS workspace, generating authentic failure-recovery dynamics instead of simulated responses. Fine-tuning on ISETrace improves ClawEval pass@1 from 19.3 to 37.7 using Qwen3-8B on agent tool-use tasks with a standard protocol. This result outperforms zero-shot GPT-4o and the larger Qwen3-32B base model which is four times bigger. An ablation on Stage 2 proves multi-turn simulation brings a large portion of the performance gain. We release all source code and dataset at https://github.com/Valiere01/ISE-Trace.

翻译：训练高性能的操作系统代理需要同时包含结构化用户意图、多轮任务委派和基于实际执行结果的工具调用数据——这些特性在现有数据集中缺失。我们提出ISE（意图 -> 模拟 -> 执行），这是一个三阶段合成范式，旨在联合解决这些不足。第一阶段通过一个四维框架（用户画像 × 领域 × 任务 × 复杂度）构建约5万条结构化意图；去重后池中包含43,956条唯一意图，并在mpnet-base-v2嵌入（余弦核，q=1）上对整个池获得61.57的Vendi分数。第二阶段通过一个角色锁定的用户模拟器驱动多轮用户-代理交互，该模拟器将每一轮用户交互基于实际执行结果，产生23,132条完整轨迹，平均包含8.12轮用户交互和68.24轮总对话。第三阶段在实时的隔离操作系统工作区中运行每个工具调用，生成真实的失败-恢复动态而非模拟响应。在ISETrace上微调后，使用Qwen3-8B在标准协议下的代理工具使用任务上，将ClawEval pass@1从19.3提升至37.7。该结果优于零样本GPT-4o和体积大四倍的Qwen3-32B基础模型。对第二阶段的消融实验证明，多轮模拟带来了大部分性能提升。我们在https://github.com/Valiere01/ISE-Trace发布所有源代码和数据集。