解决分布式大模型团队中计划过时导致误执行的问题
Fresh Memory, Stale Plans: Dependency-Scoped Validation for Distributed LLM-Agent Memory
- 通过依赖范围验证机制，只校验影响当前动作的关键记录
- 在30次测试中，传统方法全部执行过时计划，新方法零错误完成任务
- 适合高并发协作的智能体系统，尤其适用于状态频繁更新场景
分布式大模型智能体团队可能读取最新共享事实，却仍基于过时计划执行。规划者依据需求 $r_3$ 生成动作，另一智能体提交 $r_4$，执行者接收 $r_4$ 却未替换由 $r_3$ 推导出的计划。我们称此为‘过时计划执行’：状态新鲜度无法保证计划有效性。为此提出 PlanFence，一种依赖范围的动作验证协议。计划明确引用其依赖的公开记录，执行者仅验证可能影响待执行外部动作的记录，验证失败则重规划或阻断。在30个含计划修订的受控实时工作流中，仅依赖新鲜度的执行者在每项任务中均执行过时计划，而 PlanFence 全部任务无无效动作。受控回放揭示两个条件边界：主动同步在低变更率下减少协调延迟；随着变更率上升，PlanFence 避免重复更新路径协调，且在共享键空间扩大时不会验证无关状态。这些是受控安全与系统开销结果，非通用任务精度提升。
原文摘要 · Abstract (English)
Distributed LLM-agent teams can read the latest shared facts and still act on an obsolete plan. A planner may derive an action from requirement $r_3$, another agent may commit $r_4$, and an executor may receive $r_4$ without replacing the plan derived from $r_3$. We call this \emph{stale-plan execution}: state freshness does not establish that the plan authorizing an action remains valid. We introduce PlanFence, a dependency-scoped action-validation protocol. Plans cite the exact public records they used, and an executor validates only the records that can affect the pending external action, replanning once or blocking when validation is incomplete. In 30 controlled live workflows with a post-plan revision, a freshness-only executor acts on the obsolete plan in every task, whereas PlanFence completes all tasks without an invalid action. Controlled replay reveals two conditional boundaries: proactive synchronization yields lower coordination stall at low churn, while PlanFence avoids repeated update-path coordination as churn grows and avoids validating unrelated state as the shared keyspace grows. These are controlled safety and systems-cost results, not general task-accuracy gains.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。