arXiv cs.AI· Komal Thareja, Mats Rynge, Berent Aldikacti, Angelina Ju, Ian Lumsden, Anirban Mandal, Rajiv Mayani, Kin Wai Ng, Hamza Safri, Michael Stealey, Sai Swaminathan, Michela Taufer, Karan Vahi, Michael Zink, Ewa Deelman·· 3 小时前AI 评分39
PegasusAI Studio：面向科学工作流的可读网络基础设施与可问责 AI 智能体
PegasusAI Studio: Legible Cyberinfrastructure and Accountable AI Agents for Scientific Workflows
AI 导读
研究团队发布论文，介绍基于 Pegasus 与 HTCondor 构建的 AI 辅助科学网关 PegasusAI Studio，让 AI 编码智能体在共享网络基础设施（CI）上安全运行科学工作流。
正文 · 原文
Authors:Komal Thareja, Mats Rynge, Berent Aldikacti, Angelina Ju, Ian Lumsden, Anirban Mandal, Rajiv Mayani, Kin Wai Ng, Hamza Safri, Michael Stealey, Sai Swaminathan, Michela Taufer, Karan Vahi, Michael Zink, Ewa Deelman
Abstract:AI coding agents can now construct scientific workflows, but operating them on shared cyberinfrastructure (CI) raises problems that code generation does not solve. Agents need current knowledge of the workflow system and its execution sites and access to the runtime state of distributed executions, while facilities need to attribute every agent action to a user and bound it by that user's privileges. We present PegasusAI Studio, an AI-assisted science gateway built on the Pegasus and HTCondor stack. Studio makes CI legible to agents by projecting one facility-curated knowledge base into the native configuration of several coding agents, by introspecting workflow interfaces and execution sites, and by grounding diagnosis in workflow and scheduler telemetry. Agent actions operate within a trust envelope that combines federated authentication, per-user execution environments and scheduler credentials, provider-neutral model access, observability, and approval controls for state-changing actions. The same application runs as a JupyterHub-based gateway on a research cloud and as an Open OnDemand application on HPC systems. We evaluate Studio through case studies of legacy CASA weather workflow modernization, a gravitational-wave reanalysis, an EEG analysis built with an open-weight model, and an Event Horizon Telescope (EHT) M87 image reconstruction that recovered from two failed runs to a zero-failure execution. The evaluation establishes that these mechanisms operate as designed.
| Subjects: | Distributed, Parallel, and Cluster Computing (cs.DC); Artificial Intelligence (cs.AI); Software Engineering (cs.SE) |
| Cite as: | arXiv:2611.13309 [cs.DC] |
| (or arXiv:2611.13309v1 [cs.DC] for this version) | |
| https://doi.org/10.48550/arXiv.2611.13309 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Komal Thareja [view email]
[v1]
Mon, 9 Nov 2026 22:22:25 UTC (5,133 KB)
来源：arXiv cs.AI · arxiv.org