Submit a resource

Navigation

Back to directory

NVIDIA Agent Skills

An official, NVIDIA-verified catalog of Agent Skills that teach coding agents such as Claude Code and Codex how to use NVIDIA software, including Physical AI, robotics, simulation, CUDA-X and RAG workflows.

Visit website (opens in a new tab)docs.nvidia.com
NVIDIA Agent Skills

NVIDIA Agent Skills is a catalog of official, NVIDIA-verified Agent Skills that teach AI coding agents how to use NVIDIA software, spanning Physical AI and robotics workflows, simulation, CUDA-X libraries, retrieval-augmented generation, AI Blueprints and platform tools. Skills are maintained in their own product repositories and mirrored into this catalog daily through an automated sync pipeline.

Skills install through a command-line tool run via npx, which prompts for a skill and an install destination without requiring the repository to be cloned or its folders copied by hand; a specific skill can also be installed without prompts, and installation can target a particular coding agent such as Claude Code, Codex or Snowflake CoCo. Once installed, a skill activates automatically when the agent encounters a task it covers, such as solving an optimization problem with the cuOpt Python API.

It is written in Python and released under the Apache 2.0 license.

More in Agent skills

See all 581

  • Provenance gate

    github.com

    A deterministic authorization gateway that sits between an AI agent and its tools, tracing where each parameter of a side-effect call came from, and blocking the call or asking a human when the source does not check out.

  • Expected Harm Calculus (EHC-4): Asimov's Four Laws for Autonomous AI Agents

    github.com

    A markdown framework of system rules that restates Isaac Asimov's Four Laws of Robotics as a set of prioritized loss calculations for large-language-model agents to follow, published as a discovery file for agents and crawlers to read.

  • Vitruvius

    github.com

    An engineering research agent that runs a discover, read, synthesize, verify and review loop over technical questions, reading sources directly rather than inferring from titles, and publishing its own scored verifier benchmark, weaknesses included.

Cookies

We use necessary cookies to run the site. Measurement and marketing are optional; neither is used today. Privacy policy