jetson-package

Select Jetson-compatible containers and Python package indexes for GPU-native installs.

110|19|Updated May 25, 2026
One-click install
npx skills add https://github.com/NVIDIA-AI-IOT/jetson-device-skills --skill jetson-package
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: jetson-package
Source: https://github.com/NVIDIA-AI-IOT/jetson-device-skills/tree/main/skills/jetson-package
Command: npx skills add https://github.com/NVIDIA-AI-IOT/jetson-device-skills --skill jetson-package

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires python3, and includes scripts (resource) and references (resource) components.

What problem does it solve?

It helps agents choose the right Jetson-compatible container images, package indexes, and runtime options so GPU-native installs do not fail on mismatched ARM wheels or the wrong CUDA and SM target.

Core Features & Use Cases

  • Detects Jetson generation details and converts them into practical package guidance for Orin and Thor devices.
  • Recommends NVIDIA-curated container sources, including Jetson-oriented GHCR images and the current multi-arch CUDA and PyTorch paths on NGC.
  • Points users to the correct Jetson AI Lab devpi subtree for GPU wheels such as PyTorch and onnxruntime-gpu.
  • Helps with common workflows like selecting a vLLM image, replacing legacy l4t-cuda guidance, or fixing a CUDA and SM mismatch after a failed install.

Quick Start

Ask the agent to run the Jetson package skill, inspect your device, and recommend the right container image or Python package index for your workload.

Frequently Asked Questions about jetson-package

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I install Jetson-compatible Python packages for GPU acceleration?

To install Jetson-compatible Python packages for GPU acceleration, you need to use the correct Jetson AI Lab devpi subtree for GPU wheels like PyTorch and onnxruntime-gpu. This prevents failures from mismatched ARM wheels.

What is the correct container image for running vLLM on Jetson Orin devices?

The correct vLLM container for Jetson Orin devices is sourced from the NVIDIA-AI-IOT GHCR registry. The skill detects your device generation and L4T version to recommend the appropriate Jetson-ready image.

Why does my PyTorch wheel install fail with a CUDA SM mismatch on Jetson?

A CUDA SM mismatch on Jetson happens when installing wheels built for the wrong CUDA target. You must detect your L4T version and CUDA SM hints to select the correct NGC PyTorch paths or Jetson AI Lab devpi indexes.

Can I use the legacy l4t-cuda container guidance for Jetson Thor devices?

Legacy l4t-cuda container guidance should be replaced for Jetson Thor devices. The skill provides updated NVIDIA-curated multi-arch CUDA and PyTorch container paths from NGC to ensure compatibility with newer Jetson generations.

How do I find the right onnxruntime-gpu package index for my Jetson device?

Finding the right onnxruntime-gpu package index for Jetson requires checking your device's L4T version and CUDA SM hints. The skill points you to the specific Jetson AI Lab devpi subtree containing the matching GPU wheels.

Does the jetson-package skill work without python3 installed?

No, the jetson-package skill requires python3 as a dependency to function. It uses python3 to run artifact_hints.sh, which detects your Jetson generation, L4T version, and CUDA SM hints for accurate package recommendations.