今日速览
| 序号 | 标题 | 来源 | 日期 | 主题 | 推荐等级 |
|---|---|---|---|---|---|
| 1 | Persona-Execution Separation: An Architecture Pattern for Evolving LLM Agents under Execution Audit | arXiv | 2026-08-27 | AI-Agent | 高 |
| 2 | Verify Smarter, Evolve Further: Efficient Harness Evolution through Behavior-Aware Verification | arXiv | 2026-08-27 | RAG | 高 |
重点论文与技术动态
1. Persona-Execution Separation: An Architecture Pattern for Evolving LLM Agents under Execution Audit
- 来源：arXiv
- 日期：2026-08-27
- 作者/机构：Yisen Xi
- 主题标签：
AI-Agent,arXiv - 推荐等级：高
- 分类：cs.SE, cs.AI
一句话结论
Persona-Execution Separation (PES) 架构模式有效解耦了LLM代理的人格演化与执行可追溯性。
一段话。在受控组织中，LLM代理需自由演化人格（如指令、语气），同时确保执行过程可审计。PES通过将人格置于可漂移的单一信任域、执行置于需审计的受限制域，并以受控合约桥连接，实现自由漂移、执行可追溯和解耦，避免单信任域机制的高成本问题。
核心内容
- PES将人格设为可漂移的单一信任域，执行设为需审计的受限制域，通过审批矩阵、DLP和审计强制跨域操作。
- 该模式基于三大目标：人格自由演化、执行可追溯性及解耦，适用于多用户部署、执行审计和预期人格变化的场景。
- 试点中未发现执行侧重新验证或人格指纹影响硬性字段，验证了架构的稳定性。
方法与数据
- 试点在受监管数字员工平台实施，记录五项决策及被拒备选方案。
- 摘要未明确具体方法细节。
价值判断
- 值得关注：解决单信任域无法同时满足人格自由演化与执行可追溯性的核心矛盾。
- 可复用点：架构模式可直接应用于需动态调整人格的受控LLM代理系统。
- 局限/待核查：摘要未明确具体局限，但试点显示未采用PES时隔离可能被后续代码变更破坏。
摘要
Large language model (LLM) agents in governed organizations must let the persona (instructions, tone, self-presentation) evolve freely, while keeping execution (stateful, audited work) traceable. A single trust domain does not satisfy both cheaply. We present Persona-Execution Separation (PES): persona and execution reside in different trust domains, connected by a governed contract bridge. The persona is singly-homed and may drift; execution is faceless and audited. Status summaries may return; data bodies remain in the restrictive domain except a graded data-loss-prevention (DLP) exception; identity stays continuous. An approval matrix, DLP, and audit enforce the crossing. PES follows from three goals---free drift, execution traceability, and decoupling. Under LLM representational indistinguishability, any single-domain mechanism that meets all three must re-introduce typed change objects, an external gate, and a stable audit anchor: PES rebuilt at higher coupling cost. A development/pilot case in a regulated digital-employee platform records five decisions over one month, each with a rejected alternative. A mechanism check on the shipped implementation found no execution-side re-validation under persona perturbation (five model configurations) and no persona fingerprint on hard-asserted fields. A probe of a recovered pre-separation build found the governed execution path decoupled from the persona by omission, not by construction; a later wiring change could reverse that isolation, which PES makes an audited architectural rule. The pattern applies when multi-user deployment, execution audit, and expected persona churn hold jointly.2. Verify Smarter, Evolve Further: Efficient Harness Evolution through Behavior-Aware Verification
- 来源：arXiv
- 日期：2026-08-27
- 作者/机构：Jinghan Xu, Yikai Zhang, Aili Chen, Weiyuan Li, Jiaqing Liang, Deqing Yang
- 主题标签：
RAG,arXiv - 推荐等级：高
- 分类：cs.AI
一句话结论
行为感知验证结合显式归因能实现更可靠且样本高效的agent harness进化。
核心内容
- 现有 propose-and-verify 方法在固定任务集上评分所有候选，导致对无关行为的浪费rollouts，且聚合评分掩盖特定行为退化。
- HarnessLens 联合探索任务空间与用户可配置组件，从执行轨迹推导候选修改，并通过可归因证据门选择性验证行为相关任务。
- 在三个agent harnesses和四个基准测试中，HarnessLens 提升平均性能7.6-13.6%且评估预算消耗显著降低，验证了其在约束交互预算下的高效性。
方法与数据
- 方法：行为感知验证与可归因证据门。
- 数据：三个agent harnesses和四个基准测试。
价值判断
- 值得关注：行为感知验证大幅减少评估预算消耗，提升harness进化效率，尤其在资源受限场景。
- 可复用点：HarnessLens框架开源（GitHub链接），可直接复用于其他agent harness优化任务。
- 局限/待核查：摘要未明确。