FlowSteer：通过强化递进画布编辑设计代理工作流
FlowSteer: Towards Agents Designing Agentic Workflows via Reinforced Progressive Canvas Editing
AI总结：
本文提出FlowSteer，一种通过强化学习递进画布编辑设计代理工作流的新范式，解决传统工作流构建中依赖人类、缺乏图级执行反馈和无法在长周期内修复错误的问题。
摘要：
近年来，代理工作流已被广泛应用于解决复杂的人类任务。然而，现有工作流构建仍面临关键挑战，包括依赖人类的工作流构建、缺乏图级执行反馈以及在长周期构建中无法修复错误。为解决这些问题，我们提出了FlowSteer，一种新的代理设计代理工作流的范式——一个代理本身端到端地设计一个下游执行器运行的工作流。为了支持这一范式，我们引入了工作流画布，一种新的可执行图-状态环境，能够为每个原子编辑返回语法检查的执行反馈。基于画布，我们进一步提出了强化递进画布编辑，其中一个轻量级策略代理根据真实的画布反馈，每回合发出一个原子编辑，并通过强化学习端到端训练。此外，FlowSteer提供了一个即插即用的框架，支持多种运算符库和可替换的LLM后端。在十二个数据集上的实验结果表明，FlowSteer在各种任务上均显著优于基线。我们的代码可在https://anonymous.4open.science/r/FlowSteer-9B2E获取。
英文摘要：
In recent years, agentic workflows have been widely applied to solve complex human tasks. However, existing workflow construction still faces key challenges, including human-dependent workflow construction, the lack of graph-level execution feedback, and the inability to repair errors in-loop during long-horizon construction. To address these challenges, we propose FlowSteer, a new paradigm of Agent Designing Agentic Workflows - a single agent itself end-to-end designs the workflow that a downstream executor runs. To support this paradigm, we introduce the Workflow Canvas, a novel executable graph-state environment that returns syntax-checked execution feedback for every atomic edit. Built on the canvas, we further propose Reinforced Progressive Canvas Editing, in which a lightweight policy agent issues one atomic edit per turn conditioned on real canvas feedback, and is trained end-to-end via reinforcement learning. Moreover, FlowSteer provides a plug-and-play framework that supports diverse operator libraries and interchangeable LLM backends. Experimental results on twelve datasets show that FlowSteer significantly outperforms baselines across various tasks. Our code is available at https://anonymous.4open.science/r/FlowSteer-9B2E.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
