用状态约束机制解决多智能体调度中的对齐难题，提升企业级流程可靠性。
SDOF: Taming the Alignment Tax in Multi-Agent Orchestration with State-Constrained Dispatch
- 将多智能体执行建模为受状态机约束的有限自动机，分层防御
- 在真实招聘系统中实现86.5%任务完成率，0漏报注入攻击
- 适合需高可靠性的企业级工作流应用，如人力资源自动化
多智能体编排框架如LangChain、LangGraph和CrewAI通过图结构管道分配任务，但未强制执行真实业务流程中的阶段约束。我们提出SDOF，将多智能体执行视为受约束的状态机。SDOF通过两层防御机制实现，由三个组件构成：(1) 基于生成式奖励建模（GRPO）训练的在线强化学习人类反馈专用意图路由器；(2) 带目标阶段有限自动机检查与先决/后置条件技能注册表验证的状态感知调度器。在基于贝森iTalent平台（覆盖6000+企业）的招聘系统上，185个专家标注场景触发1671次真实API调用。我们的GSPO对齐7B意图路由器在该状态机约束的对抗性路由基准上，联合准确率达80.9%，优于零样本GPT-4o的48.9%。端到端执行中，SDOF达成86.5%任务完成率（95%置信区间80.8至90.7），并成功阻断全部22项非法HR操作。在消息级阻断审计中，精度达100%，召回率为88%，专家一致性kappa=0.94。对960条源自SGD对话、涵盖8个服务领域的评估发现，按有限自动机映射存在201处阶段顺序冲突，其中41处出现在正常划分数据中。当前arXiv版本报告已验证范围；后续将发布多种子训练对比与更深入的工作流评估。
原文摘要 · Abstract (English)
Multi-agent orchestration frameworks such as LangChain, LangGraph, and CrewAI route tasks through graph-based pipelines but do not enforce the stage constraints that govern real business processes. We present SDOF, a framework that treats multi-agent execution as a constrained state machine. SDOF operates through two primary defensive layers, implemented by three components: (1) an Online-RLHF Specialized Intent Router trained via Generative Reward Modeling (GRPO) and (2) a StateAwareDispatcher with GoalStage finite-automaton checks and precondition/postcondition SkillRegistry validation for auditable execution control. On a recruitment system backed by the Beisen iTalent platform (6000+ enterprises), 185 expert-curated scenarios trigger 1671 live API calls. Our GSPO-aligned 7B Intent Router achieves higher joint accuracy than zero-shot GPT-4o on this FSM-constrained adversarial routing benchmark (80.9% versus 48.9%). In end-to-end execution, SDOF reaches 86.5% task completion (95% confidence interval 80.8 to 90.7) and blocks all 22 operations in the injection, illegal HR subset. Under a broader message-level blocking audit, SDOF attains precision 100% and recall 88%, expert agreement kappa=0.94. A separate evaluation on 960 SGD-derived dialogues spanning 8 service domains surfaces 201 stage-order conflicts under our FSM mapping, 41 of which arise in the normal split. This arXiv version reports the current validated scope; extended multi-seed training comparisons and deeper workflow evaluations will be released in a subsequent update.
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。