发表机构
Microsoft Research; National University of Singapore; The Ohio State University; Carnegie Mellon University（微软研究院; 新加坡国立大学; 俄亥俄州立大学; 卡内基梅隆大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
CUAWright提出一个以bash命令为唯一动作接口、文件系统为可演化空间的最小终端框架，通过可编程接口显著提升数字智能体在OSWorld、Online-Mind2Web等基准上的性能与成本效率。
AI 中文摘要
当前计算机使用智能体的主流方法是将模型与特定领域的工具框架相结合：一个配备人工设计工具的浏览器或桌面环境，这些工具在任务执行前就已固定。随着模型编码能力的提升，这种图形用户界面（GUI）原生且静态的框架阻碍了智能体直接以编程方式操作系统状态，也限制了工具的灵活构建。为此，我们提出了CUAWright，一个约3000行代码的最小终端框架，以bash命令作为其唯一的动作接口，并以文件系统作为其可演化的空间，用于动态创建工具和管理上下文。我们在广泛的数字任务上进行了全面实验，结果表明，通过为智能体提供一个最小化、可编程的接口，它在广泛的任务上相比GUI或混合命令行（CLI）接口取得了显著更强的结果。在OSWorld 2.0上，与已发布的GPT-5.5基线相比，CUAWright在部分奖励上实现了33.2%的相对提升，同时估计成本降低了37.5%。在Online-Mind2Web和长时程Odysseys基准上，CUAWright在成功率上分别大幅超越GUI原生框架4.7%和44.0%。此外，我们发现这些优势也扩展到需要精确视觉理解和CLI交互的CAD应用中：在CADGenBench和BenchCAD上，我们的统一框架相比其他基于CLI的框架（使用GPT-5.5）实现了8.1%-41.6%的相对改进。综合这些结果，表明数字环境远比其GUI界面所暗示的更具可编程性，而一个以终端为中心的最小框架是获得更好性能和效率的关键。
英文摘要
The prevailing approach to computer-use agents couples a model with a domain-specific harness: a browser or desktop environment equipped with human engineered tools that are fixed before task execution. As models' coding capabilities improve, the GUI native and static harness prevents them from direct programmatic operation on system state, as well as flexible construction of tools. To this end, we introduce CUAWright, a minimal terminal harness of roughly 3K lines of code that uses bash commands as its sole action interface, and a file system as its evolvable space for dynamically creating tools and managing the context. We conduct comprehensive experiments across a wide range of digital tasks, and demonstrate that by giving the agent a minimal, programmable interface, it achieves substantially stronger results compared to their GUI or hybrid CLI interface across a wide range of tasks. On OSWorld 2.0, CUAWright delivers a 33.2% relative improvement in partial reward while reducing estimated cost by 37.5% compared with the published GPT-5.5 baseline. On Online-Mind2Web and the long horizon Odysseys benchmark, CUAWright substantially outperforms GUI native harness by 4.7% and 44.0% in success rate, respectively. Furthermore, we found the gains extend to CAD applications that require accurate visual understanding and CLI interaction: on CADGenBench and BenchCAD, our unified harness yields 8.1%-41.6% relative improvements over other CLI based harnesses with GPT-5.5. Together, these results suggest digital environments are far more programmable than their GUI interfaces imply, and a minimal terminal-focused harness is the key for better performance and efficiency.
Comments20 pages, 10 figures
