Agent-based code generation for the Gammapy framework

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Kostunin, Dmitriy, Sotnikov, Vladimir, Golovachev, Sergo, Mehta, Abhay, Holch, Tim Lukas, Jones, Elisa
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918151602569216
author Kostunin, Dmitriy
Sotnikov, Vladimir
Golovachev, Sergo
Mehta, Abhay
Holch, Tim Lukas
Jones, Elisa
author_facet Kostunin, Dmitriy
Sotnikov, Vladimir
Golovachev, Sergo
Mehta, Abhay
Holch, Tim Lukas
Jones, Elisa
contents Software code generation using Large Language Models (LLMs) is one of the most successful applications of modern artificial intelligence. Foundational models are very effective for popular frameworks that benefit from documentation, examples, and strong community support. In contrast, specialized scientific libraries often lack these resources and may expose unstable APIs under active development, making it difficult for models trained on limited or outdated data. We address these issues for the Gammapy library by developing an agent capable of writing, executing, and validating code in a controlled environment. We present a minimal web demo and an accompanying benchmarking suite. This contribution summarizes the design, reports our current status, and outlines next steps.
format Preprint
id arxiv_https___arxiv_org_abs_2509_26110
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Agent-based code generation for the Gammapy framework
Kostunin, Dmitriy
Sotnikov, Vladimir
Golovachev, Sergo
Mehta, Abhay
Holch, Tim Lukas
Jones, Elisa
Software Engineering
Instrumentation and Methods for Astrophysics
Software code generation using Large Language Models (LLMs) is one of the most successful applications of modern artificial intelligence. Foundational models are very effective for popular frameworks that benefit from documentation, examples, and strong community support. In contrast, specialized scientific libraries often lack these resources and may expose unstable APIs under active development, making it difficult for models trained on limited or outdated data. We address these issues for the Gammapy library by developing an agent capable of writing, executing, and validating code in a controlled environment. We present a minimal web demo and an accompanying benchmarking suite. This contribution summarizes the design, reports our current status, and outlines next steps.
title Agent-based code generation for the Gammapy framework
topic Software Engineering
Instrumentation and Methods for Astrophysics
url https://arxiv.org/abs/2509.26110