AI Coding-Assistant Skills
The Holoscan team publishes a collection of agent-readable skills that help you install, run, and explore the Holoscan SDK from inside an AI coding assistant such as Claude Code, Cursor, or GitHub Copilot CLI. The skills are organized in the public NVIDIA/skills repository, alongside skills from other NVIDIA teams.
A skill is a self-contained playbook — a description of what it does, when to use it, and the exact steps to follow — that an AI assistant invokes by name. When
you ask the assistant to “install Holoscan in a container,” the assistant matches
your request to the holoscan-install-container skill and follows its steps:
pulling the right NGC image for your platform, verifying GPU passthrough, and
running the bundled example tests on your behalf. Skills give the assistant a
reliable, reviewable procedure to follow instead of improvising.
Skills are installed with the skills, which
runs through npx and prompts you to choose a skill and an install destination —
no manual git clone or folder copying required:
npx skills add nvidia/skills
Follow the interactive prompts to select the skills you want. The CLI installs each skill into the location appropriate for your AI assistant (project scope, user scope, or global).
For advanced install options — including non-interactive flags, custom destinations, and integration with multiple assistants — see the skills CLI advanced install guide.
Available skills cover a range of Holoscan SDK tasks. Today they focus on installing the SDK — walking you through any supported install path (container, Debian package, Python wheel, Conda, or from source) and helping you pick the one that fits your platform.
The canonical list of skills — including each skill’s full description,
prerequisites, and versioning — lives in
NVIDIA/skills. Run the install command
above to browse them interactively in the CLI.
We’re actively iterating on these skills and want to hear from users. Share feedback, ideas, and skill-specific issues (bad install steps, outdated commands, unclear prompts) through the Holoscan community channels — developer forum, Discord, and more. Tell us which workflows you’d like a skill for, what worked, and what didn’t.
Please do not open issues or pull requests against NVIDIA/skills for Holoscan skills. That repository is shared across many NVIDIA teams; reports filed there are unlikely to reach the Holoscan team. Route all Holoscan-specific feedback through the community channels above.
The skill collection is curated by the Holoscan team while the format and content
stabilize. At this stage, the Holoscan team is not accepting external pull requests that add, remove, or modify skills — either in the SDK’s local
skills/
mirror or in the canonical NVIDIA/skills
repository. This is a narrower constraint than the SDK’s general
Contributing Guide,
which continues to welcome community contributions to the rest of the SDK.
Once the skill format is stable, we expect to open contributions. Until then, please route ideas and improvements through the forum link above.