automodel turns coding agents into model-discovery assistants
Researchers describe a portable skill that lets general-purpose coding agents search for scientific models, with case studies across five domains.
In plain words
Scientists often need mathematical models that explain data, and writing them by hand is slow. The automodel project packages a model-discovery method as a skill that coding agents like Claude Code can follow. The authors report these agents then found useful new models on their own.
What was announced
A new arXiv paper, automodel, addresses the shift from hand-designed models to automated discovery. Rather than building another dedicated search algorithm, the authors encode a discovery recipe as a skill for commercial coding agents such as Claude Code and OpenAI Codex, asking whether such a recipe alone could let these systems discover or improve models.
In case studies spanning astrophysics, cardiovascular dynamics, precision agriculture, traffic flow, and vehicle tire dynamics, the authors report that agents equipped with automodel identified novel, scientifically relevant models. They argue this makes a portable skill a simple, interactive, and broadly applicable alternative to domain-specific discovery pipelines. The skill is available on GitHub.
What it does not show
- Evidence is limited to the paper's abstract; no benchmark numbers, baselines, or independent evaluation of the claimed discoveries are provided.
Sources
Source-support review completed 20 Nov 2026. This story was drafted with AI assistance and checked against retained source passages. Automated checking is not independent verification.
Coverage: arXiv: abstract.
Updates and corrections
No corrections so far.
First published 20 Nov 2026
New to the subject? AI from the Ground Up starts from the beginning.