SHE: Trajectory-driven Safety Harness Evolution for LLM Agents· LLM 安全动态演化框架 SHE
The safety of large language model (LLM) agents depends not only on model weights but also on the agent harness that manages context, memory, tools, permissions, and runtime control. Existing safety mechanisms often treat the harness as a fixed deployment artifact, limiting their ability to evolve with emerging risks. Moreover, coupled functions across harness components obscure safety responsibility attribution, making localized evolution difficult. We propose Safety Harness Evolution (SHE), a framework that learns evolving safe boundaries from rollout trajectories. SHE decomposes the harness into four artifacts with explicit safety responsibilities, including the System Prompt, Rule Bank, Safety Memory, and Tool Policy, defining clear functional boundaries for localized evolution. Based
提出动态演化框架 SHE，提高 LLM 代理的安全性。
- 核心方法
- SHE 框架通过分解安全带为四个具有明确安全责任的组件（系统提示、规则库、安全记忆和工具策略），并基于轨迹学习安全边界，实现局部化演化。
- 适合谁读
- 研究者、工程师
- 要解决的问题
- 现有的 LLM 代理安全机制无法随新兴风险演化，且各组件功能耦合影响安全责任归属。
- 关键实验
- 未提供
- 主要贡献
- 提供了一种动态演化的方法，使 LLM 代理的安全机制能够适应新的风险，并明确了各安全组件的功能边界。
- 意义与局限
- 意义在于提高了 LLM 代理的安全性和灵活性，但可能存在的局限是需要大量轨迹数据来训练模型。