arXiv：cs.CL· Xueyang Feng, Chaozhuo Li, Xiaohe Bo, Jiakai Tang, Chenxi Li, Qiwei Ye, Xu Chen·· 11 小时前
POrchestra：主动式智能体编排框架
Proactive Agentic Orchestration
AI 导读
POrchestra 是一个实时主动式智能体编排框架，允许子智能体在执行过程中主动向编排器请求调整 harness，从而适应动态环境。该机制还衍生出基于事后反馈的 On-Policy Orchestration Distillation（OPOD），可将子智能体的主动修改请求归因到先前的编排决策以优化编排策略。
正文
Abstract:Agentic orchestration delegates tasks to sub-agents with independent contexts, providing a key path to extending the effective execution horizon of large language model agents. However, existing orchestration methods face two limitations: in execution, current methods fix sub-agents' goals and harnesses at creation time, limiting their adaptability to dynamic real-world environments; in optimization, sparse and delayed feedback constrains effective learning of the orchestration policy. To address these limitations, we introduce POrchestra, an on-the-fly proactive agentic orchestration framework. It allows sub-agents to proactively communicate with the orchestrator at any logical point to request harness adjustments while maintaining continuity of local execution. This mechanism not only enables sub-agents to adapt seamlessly to dynamic environments but also naturally inspires On-Policy Orchestration Distillation (OPOD) based on hindsight feedback. OPOD attributes sub-agents' proactive revision requests back to preceding orchestration decisions to derive targeted improvement signals, and leverages these signals as process supervision alongside the final reward to optimize the orchestration policy. We evaluate POrchestra on three long-horizon agent benchmarks spanning deep research, dynamic digital assistance, and software engineering. POrchestra improves task success by an average of 9.1% relative to the strongest baseline; combined with OPOD, it improves small-orchestrator success by an average of 10.0% relative to existing training baselines. POrchestra also achieves a more favorable cost-performance trade-off. Our code is publicly available at this https URL.
| Comments: | 41 pages including references and appendices, 11 figures. Code: this https URL |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.13485 [cs.CL] |
| (or arXiv:2610.13485v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.13485 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Xueyang Feng [view email]
[v1]
Fri, 9 Oct 2026 12:18:49 UTC (740 KB)
来源：arXiv：cs.CL · arxiv.org