Agent Harness Awareness：将验证遥测转化为工具型智能体的操作控制
原标题：Agent Harness Awareness: Turning Verified Telemetry into Operational Control for Tool-Using Agents
AI 摘要
论文提出 Agent Harness Awareness 架构，让主机在执行关键动作前向模型提供经过验证的、带新鲜度标签的执行状态快照，以解决模型只看到静态工具描述而无法感知工具健康状态的“操作盲区”。在 216 次决策的消融实验中，GPT-5.6 Sol、Claude Sonnet 5 和 Gemini 3.7 Flash 在获得验证状态后，平均可观测策略遗憾降低 96.9%，单次派发完成率从 69.4% 提升至 95.8%。另在 27 次决策的本地 TCP/HTTP 实时复现中，验证状态将平均遗憾从 8,956 ms 降至 0 ms，9/9 个模型-故障组合完成动作，而对照组仅 5/9。
正文节选
Agent Harness Awareness: Turning Verified Telemetry into Operational Control for Tool-Using Agents Abstract Tool descriptions tell an agent what an action can do, but they do not necessarily reveal whether that action is healthy or suitable at the moment of use. Execution harnesses often possess this information through monitors, traces, and recent failures while the model choosing the next action does not. We call this gap operational blindness. We introduce Agent Harness Awareness, a trust-sep