Physicsnemo Discover

Official NVIDIA skill that routes you to the right PhysicsNeMo model, datapipe, or example for scientific ML tasks like surrogates, forecasting, and physics-informed learning. It searches the live repo every time instead of relying on hardcoded paths, so it won't point you to renamed classes or moved examples. Treats PhysicsNeMo as a composable framework (model family × datapipe × training strategy) and shows you multiple options along each axis rather than picking one for you. Strictly for discovery and navigation. It will not write training code, handle installation, or answer questions outside the SciML/AI4Science domain. If your task is reinforcement learning or generic computer vision, it abstains rather than guessing.

skills.sh installs: 1.9K
GitHub stars: 3.9K
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add nvidia/skills --skill physicsnemo-discover --agent claude-code

Installs into .claude/skills of the current project.

Files

PhysicsNeMo Discoverability

Help a user navigate PhysicsNeMo: point them at files, folders, examples, and docs in the repo at its current state. Never write training code; never cite a path from memory.

Core principle

PhysicsNeMo evolves — classes get renamed, examples move, experimental/ graduates. Any static list of class names and paths rots, so discover, don't remember: enumerate from the live repo every turn.

PhysicsNeMo is composable: each solution is a product (model family × datapipe × training strategy × config). An example is one reference instantiation of that product, not a prescription. Surface the axes and the menu along each axis, then cite examples as concrete starting points to fork and recombine.

What a correct answer satisfies

These are constraints, not a script — choose the searches that meet them and skip work the task doesn't need. Search patterns per axis live in references/RECIPES.md.

SKILL.md on GitHub