AI Agent 论文播报·2026-06-02

这是 AI Agent 论文播报 在 2026-06-02 的论文播报。本期内容由 AI 自动生成，欢迎留言交流。

本期重点

• SPADE-Bench: Evaluating Spontaneous Strategic Deception in Agents via Plan-Action Divergence
• Adaptive Auto-Harness: Sustained Self-Improvement for Agentic System Deployment on Open-Ended Task Streams
• SeClaw: Spec-Driven Security Task Synthesis for Evaluating Autonomous Agents

['今日主线：Agent栈从能力堆叠转向诊断、治理与运行时演化。']