本文へ移動
cccskills
無料GitHub で公開

jetson-print-device-info

Use when you need to print Jetson device info (module model, L4T version, kernel, OS version, current power mode) from a running Jetson target. This is an example skill.

インストール方法を見る

含まれるファイル(5)

  • SKILL.md5.1 KB
  • BENCHMARK.md3.1 KB
  • evals/evals.json5.7 KB
  • skill-card.md3.6 KB
  • skill.oms.sig4.5 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

jetson-print-device-info

Prints a concise summary of the Jetson device this skill runs on.

This skill is intended as a reference example for the jetson-device-skills repo and the NVIDIA-wide skills CI. It runs on the Jetson target (not the host PC) and performs read-only inspection — useful as a baseline capture before running performance tests.

Purpose

Capture a baseline snapshot of a Jetson target's software stack and power mode before running performance, regression, or compatibility tests.

When to use

  • Starting a performance run and you want a captured baseline of the device's software stack and power mode.
  • Verifying that a freshly flashed Jetson matches an expected L4T / JetPack version.

Prerequisites

  • Running on a Jetson target (not the host PC).
  • Standard CLIs available: tr, cat, uname, uptime, lsb_release (or /etc/os-release).
  • nvpmodel available for power mode (optional — skill falls back gracefully).

Inputs

None. The skill reads only from the local Jetson filesystem and standard CLIs.

Instructions

Run each step in order and print the captured values into the report shown under Output format.

  1. Capture the module model and validate it is a Jetson target — exit early otherwise:
    # Device-tree strings are null-terminated, so strip NULs before printing.
    model=$(tr -d '\0' < /proc/device-tree/model 2>/dev/null)
    case "$model" in
      *Jetson*) ;;
      *) echo "Not running on a Jetson target (model: '${model:-unknown}')"; exit 1 ;;
    esac
    echo "$model"
    
  2. Extract the L4T release header line — skip the rest of the file (long list of library SHAs):
    head -1 /etc/nv_tegra_release 2>/dev/null || echo "L4T release info not found"
    # Equivalent: grep -m1 '^# R' /etc/nv_tegra_release
    
  3. Run nvpmodel -q and join its two output lines (NV Power Mode: <name> and the mode number) onto one line. paste -sd only uses the first char of its delimiter, so use awk to insert the literal / separator:
    nvpmodel -q 2>/dev/null | awk 'NR==1{a=$0; next} {print a" / "$0}' \
      || echo "nvpmodel not available"
    
  4. Print the kernel version and uptime:
    uname -r
    uptime -p
    
  5. Print the OS version (prefer lsb_release, fall back to /etc/os-release):
    lsb_release -ds 2>/dev/null || (. /etc/os-release && echo "$PRETTY_NAME") || echo "OS version not found"
    

Output format

Print a short report with these sections, one line each where possible:

Model:           <device-tree model string>
L4T release:     <release header line>
Power mode:      <nvpmodel name> / <mode number>
Kernel:          <uname -r>
Uptime:          <uptime -p>
OS version:      <lsb_release / /etc/os-release output>

Examples

Example output on an Orin AGX dev kit (L4T R36 / Ubuntu 22.04):

Model:           NVIDIA Jetson AGX Orin Developer Kit
L4T release:     # R36 (release), REVISION: 3.0
Power mode:      NV Power Mode: MAXN / 0
Kernel:          5.15.136-tegra
Uptime:          up 2 hours, 14 minutes
OS version:      Ubuntu 22.04.4 LTS

Example output on an AGX Thor (L4T R39 / Ubuntu 24.04):

Model:           NVIDIA Jetson AGX Thor Developer Kit
L4T release:     # R39 (release), REVISION: 0.0
Power mode:      NV Power Mode: 120W / 1
Kernel:          6.8.0-tegra
Uptime:          up 12 minutes
OS version:      Ubuntu 24.04 LTS

Error handling

Each command falls back to a clearly labeled "... not found" / "... not available" string if the underlying file or binary is missing — the skill never errors out mid-report. If /proc/device-tree/model is missing or does not contain a Jetson string, exit early with a clear "not running on a Jetson target" message.

Limitations

  • Read-only inspection only — does not detect GPU/CPU clocks, thermal state, or per-rail power.
  • nvpmodel output format varies between L4T versions; the skill prints it verbatim rather than parsing.

Troubleshooting

  • Error: /proc/device-tree/model: No such file or directory Cause: Running on a host PC, not a Jetson target. Solution: Run the skill on the Jetson device directly (e.g. via SSH).

  • Error: nvpmodel: command not found Cause: L4T BSP not installed, or running in a minimal container without nvpmodel. Solution: Expected on non-Jetson or stripped environments — the skill prints "nvpmodel not available" and continues.

Notes

  • Read-only. Do not change power mode, install packages, or modify any files.
  • If the skill is invoked on a host PC by mistake, /proc/device-tree/model will not contain a Jetson model string — detect that and exit with a clear message rather than printing misleading info.

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

Official NVIDIA-authored guidance for NVIDIA cuDF GPU DataFrames, pandas acceleration, dask-cuDF, ETL, joins, groupby, CSV/Parquet I/O, nullable semantics, and multi-GPU DataFrame workloads.

日本語の概要は準備中です。原文の説明を表示しています。

NVIDIA/skills3,5702026年10月10日 更新

Customize NVIDIA Nemotron Voice Agent's Generic Pipecat example for healthcare appointment, five-field patient intake, or custom tool-calling workflows without a separate backend.

日本語の概要は準備中です。原文の説明を表示しています。

NVIDIA/skills3,5702026年10月10日 更新

Calibrate a new dataset from live RTSP camera streams via the AutoMagicCalib REST API. Use when the user provides RTSP URLs or asks to calibrate live cameras; VIOS records clips, AMC ingests them, then runs calibration.

日本語の概要は準備中です。原文の説明を表示しています。

NVIDIA/skills3,5702026年10月10日 更新

Run end-to-end calibration on the shipped sample dataset (sdg_08_2_sample_data_010926.zip) against a running AMC microservice. Use when user says 'test sample dataset', 'run sample calibration', 'verify AMC install', or 'launch and test'.

日本語の概要は準備中です。原文の説明を表示しています。

NVIDIA/skills3,5702026年10月10日 更新

Calibrates pre-recorded `cam_*.mp4` datasets through the AutoMagicCalib REST API. Use for user-supplied local MP4s; route live RTSP streams to `amc-run-rtsp-calibration`.

日本語の概要は準備中です。原文の説明を表示しています。

NVIDIA/skills3,5702026年10月10日 更新

Launch AutoMagicCalib microservice and web UI from NGC release images via Docker Compose. Use when user says 'deploy auto calibration', 'launch auto calibration', 'launch AMC', 'start MS+UI', or 'set up auto-magic-calib'. Requires NGC API key.

日本語の概要は準備中です。原文の説明を表示しています。

NVIDIA/skills3,5702026年10月10日 更新

NVIDIA のスキルをすべて見る

このスキルの問題を報告する