# PegasusAI Studio: Legible Cyberinfrastructure and Accountable AI Agents for Scientific Workflows

DevFeed: [PegasusAI Studio: Legible Cyberinfrastructure and Accountable AI Agents for Scientific Workflows](<https://devfeed.tech/articles/pegasusai-studio-legible-cyberinfrastructure-and-accountable-ai-agents-for-scientific-workflows-127802.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.13309>)

Author: 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

Published: 2026-11-11T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>), [arXiv: Software Engineering](<https://devfeed.tech/sources/arxiv-software-engineering.md>)

Topics: [observability](<https://devfeed.tech/topics/observability.md>), [jupyterhub](<https://devfeed.tech/topics/jupyterhub.md>), [Legacy Modernization](<https://devfeed.tech/topics/legacy-modernization.md>)

Tags: [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-coding-agents](<https://devfeed.tech/tags/ai-coding-agents.md>), [authentication](<https://devfeed.tech/tags/authentication.md>), [ci](<https://devfeed.tech/tags/ci.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [cs-se](<https://devfeed.tech/tags/cs-se.md>), [distributed](<https://devfeed.tech/tags/distributed.md>), [eeg](<https://devfeed.tech/tags/eeg.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [event-horizon-telescope](<https://devfeed.tech/tags/event-horizon-telescope.md>), [execution](<https://devfeed.tech/tags/execution.md>), [gateway](<https://devfeed.tech/tags/gateway.md>), [hpc](<https://devfeed.tech/tags/hpc.md>), [jupyterhub](<https://devfeed.tech/tags/jupyterhub.md>), [knowledge-base](<https://devfeed.tech/tags/knowledge-base.md>), [m87](<https://devfeed.tech/tags/m87.md>), [model-access](<https://devfeed.tech/tags/model-access.md>), [modernization](<https://devfeed.tech/tags/modernization.md>), [observability](<https://devfeed.tech/tags/observability.md>), [runtime](<https://devfeed.tech/tags/runtime.md>), [workflows](<https://devfeed.tech/tags/workflows.md>)

## AI overview

PegasusAI Studio combines workflow-aware agent context with controls that attribute and constrain actions on shared computing infrastructure. Built on Pegasus and HTCondor, it projects a facility-curated knowledge base into several coding agents, introspects workflow interfaces and execution sites, and uses workflow and scheduler telemetry to support diagnosis. Federated authentication, per-user environments and scheduler credentials, provider-neutral model access, observability, and approvals for state-changing actions form its trust controls. The same application runs through JupyterHub on a research cloud or Open OnDemand on HPC. Case studies cover CASA workflow modernization, gravitational-wave reanalysis, EEG analysis using an open-weight model, and an EHT M87 reconstruction that recovered from two failed runs to zero failures. The evaluation reports that the mechanisms operate as designed.

## Source excerpt

arXiv:2611.13309v1 Announce Type: new 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.