Official agent skill

Jetson Init Target

by NVIDIA in NVIDIA/skills

Author a new Jetson target-platform profile (referencedevkit + optional customcarrier) and update the active pointer.

OfficialApache-2.0Auto-check passedAI & LLM Engineering

Install Jetson Init Target

skills CLI
$ npx skills add NVIDIA/skills --skill jetson-init-target -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install NVIDIA/skills jetson-init-target --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/NVIDIA/skills.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/jetson-init-target .claude/skills/jetson-init-target && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
jetson-init-target
GitHub stars
3.5k
Token cost
~4.8k tokens
SKILL.md length
2,350 words
Files
7 (incl. references)
Skills in repo
380
Repo updated
First seen
Licence
Apache-2.0

At a glance

Author a new Jetson target-platform profile (referencedevkit + optional customcarrier) and update the active pointer.

  • Works in 3 steps: If target-platform/ does not exist,… → Read target-platform/active_target.yml. → List target-platform/*.yaml (excluding…
  • Create a target
  • SKILL.md covers Overview, When to invoke, Procedure and Prerequisites, plus 3 more sections
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

Jetson Init Target is an agent skill from NVIDIA/skills, published by the product's own GitHub organization. Author a new Jetson target-platform profile (referencedevkit + optional customcarrier) and update the active pointer. Use to create a target; not for switching existing profiles.

Its SKILL.md is about 4.8k tokens, which your agent loads only when the skill is triggered. The skill folder holds 8 other files, including reference files (for example `BENCHMARK.md`, `evals/evals.json` and `references/gotchas.md`).

It sits in AI & LLM Engineering, covering GPU and accelerator computing. It works with NVIDIA AI Platform. The repository describes itself as: Agent Skills for NVIDIA products — install into Claude Code, Codex, and other coding agents to run Physical AI, robotics, simulation, CUDA, and RAG workflows end to end. The licence is Apache-2.0.

When your agent uses it

  • Create a target
  • Not for switching existing profiles

Example prompts

  • “/jetson-init-target”

Workflow steps

3 steps, taken from the first numbered list in SKILL.md.

  1. If target-platform/ does not exist, create it and copy
  2. Read target-platform/active_target.yml.
  3. List target-platform/*.yaml (excluding active_target.yml).

What it can do on your machine

Read from SKILL.md and the folder at commit 67a13c0. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Jetson Init Target loads about 4.8k tokens when it runs, and up to ~6k if it reads all its reference files. Until then it costs about 50 tokens; SKILL.md has 2,350 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~50
When it runs · the whole SKILL.md, loaded when a task matches
~4.8k
With references · SKILL.md plus every file in references/, read only if the agent opens them
~6k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from NVIDIA/skills at commit 67a13c0, republished under its Apache-2.0 licence (© NVIDIA). 2,350 words, ~4,761 tokens.

Download SKILL.mdSave it as .claude/skills/jetson-init-target/SKILL.md (or your agent's skills folder). This skill also uses 6 other files; get the full folder from GitHub.
name
jetson-init-target
description
Author a new Jetson target-platform profile (reference_devkit + optional custom_carrier) and update the active pointer. Use to create a target; not for switching existing profiles.
version
0.0.1
license
Apache-2.0
metadata.data-classification
public
metadata.author
Jetson Team
metadata.tags
target-platform, configuration, meta
metadata.domain
meta

Initialize Target Platform

Overview

This skill authors a new target-platform profile YAML — the write-side of the contract in ../../context/target-platform-contract.md. It uses ../../references/bsp-platforms-catalogue.md as the canonical product / flash-config catalogue and ../../references/platform_template.yaml as the field schema.

Flow is linear: pick a reference product + flash config, then optionally add custom-carrier details. The two contract cases — reference-only vs reference + custom carrier — are distinguished only by whether the user provides custom-carrier details; there is no up-front mode prompt.

Scope is reference_devkit: and custom_carrier: only. Sibling skills own the other blocks: bsp_image: → jetson-init-image; source: → jetson-init-source; documents: → jetson-link-docs.

Outputs: a new target-platform/<name>.yaml and (by default) an updated active_target.yml pointing at it. To switch among already-authored profiles, use sibling jetson-set-target.

When to invoke

  • The user asks to create / author / add a new target platform profile.
  • A downstream skill refused with "no active target" and target-platform/ contains no profile YAML files yet (otherwise prefer jetson-set-target).
  • The user wants a target whose profile does not yet exist on disk.

Procedure

Quick-start prefill mapping

Follow the shared quick_start_prefill contract. This skill consumes:

  • quick_start_prefill.target.active_platform as the selected catalogue row when it resolves to exactly one product.
  • quick_start_prefill.target.custom_carrier as the custom-carrier intent.

If active_platform resolves to one catalogue row, use it directly. Do not re-ask reference-product or flash-config questions; use the row's default flash config unless the prefill names a catalogue variant. Ask only when the prefill is missing, invalid, ambiguous, or names an unknown variant.

How the skill consumes the template

Load ../../references/platform_template.yaml once at skill startup and treat it as the schema for prompted blocks. Each placeholder value carries one of three sentinel markers:

MarkerBehavior
<REQUIRED: description>Prompt the user. Value must be provided; NA is accepted with a warning that downstream skills may refuse.
<OPTIONAL: description>Prompt the user. Value may be skipped (Enter or NA); skipped fields are omitted from the written profile (no empty key, no NA placeholder).
<DERIVED: description>Filled programmatically from bsp-platforms-catalogue.md and the user's product / flash-config choice in the "Pick the reference product" and "Pick the flash config" steps. Not prompted.

Use the marker description as the prompt text verbatim. Match markers with the regex ^<(REQUIRED|OPTIONAL|DERIVED):\s*(.*)>$ after YAML parsing strips the surrounding quotes.

Block-level inclusion (custom_carrier) is decided by the skill flow, not by markers — adding/removing fields inside a block is a template-only change.

Survey current state
  1. If target-platform/ does not exist, create it and copy ../../references/active_target_template.yaml to target-platform/active_target.yml (note .yml extension).
  2. Read target-platform/active_target.yml.
  3. List target-platform/*.yaml (excluding active_target.yml).
SituationAction
No profile YAMLs existProceed — this is the bootstrap case. The new profile will be set as active_target automatically in the "Write" step (no activation prompt in the "Confirm" step).
Profiles exist, active: NAShow the list of existing profiles and confirm the user wants to author a new one (rather than activate one of the existing profiles via jetson-set-target).
Profiles exist, active: names a present fileShow the active profile and the full list. Ask whether the user wants to (a) author a new profile alongside, or (b) switch to one of the existing profiles via jetson-set-target. If (b), stop and tell the user to run jetson-set-target.
active: names a missing fileTell the user the pointer is broken; offer to author a new profile (and replace the pointer) or to fix the pointer manually.
Read the platforms catalogue

Read ../../references/bsp-platforms-catalogue.md. Parse the "Product / Chip / SKU / Flash Config Mapping" table — that is the source of truth for product names, CVM/CVB SKUs, default flash confs, and variant suffixes. Do not scan Linux_for_Tegra/*.conf or parse conf filenames; the catalogue already collapses that.

Pick the reference product (baseline)

Print the full catalogue list before asking. Group rows by chip family and number every entry with stable indexes for this run. Build the list only from bsp-platforms-catalogue.md; do not copy platform rows from examples or other docs.

Build explicit shortcut choices from the printed list by taking the first row in each chip-family group. Each shortcut label must include the full-list index and product name, for example Index 1 — Jetson AGX Orin 64GB. Do not hard-code product names; derive shortcuts from the parsed catalogue rows. These shortcuts are the only explicit choices for this question. Manual index entry must use the tool's built-in Type something input; do not add a third custom choice for manual entry.

Prompt text:

Pick a platform shortcut, or choose Type something and type the numeric index from the full platform list printed above. If your hardware is a custom carrier with one of these Jetson modules, pick the closest matching reference devkit as the baseline. Type cancel to cancel this run.

Do not show chip-family choices such as T234 / Orin family or T264 / Thor family. Do not show Type index, Manual entry, Other platform, or any custom manual-entry choice. The only manual-entry path is the tool-provided Type something row. Do not add Other as a choice or paraphrase Type something as Other in prompt text. Do not accept product-name substring matches in this step; typed input must be a numeric index from the printed full list or cancel. skip is not allowed because this skill cannot author a target profile without a reference platform.

Resolve the shortcut or typed index to exactly one catalogue row. If the typed index is out of range or invalid, show the valid index range and ask again. If the user types skip, explain that jetson-init-target requires a reference platform and ask again. If the user types cancel, stop without writing a profile.

Pick the flash config

From the catalogue row, the Default flash conf is option 1; the Variants column lists suffixes — option 2..N. Reconstruct each variant's full conf filename by appending the suffix to the default conf base. If a row says (same as <other>) in the Variants column, reuse the referenced row's variant list verbatim. Build this list only from the selected catalogue row.

Prompt:

Pick a number, or press Enter to accept the default (1).

If the chosen row has no variants (— in the catalogue), proceed to the "Optional: add custom carrier details" step with the default conf.

Edge cases:

  • Raw-conf variants (catalogue marks (raw)). For Orin Nano, -a0-maxn (raw) → p3768-0000-p3767-0000-a0-maxn.conf. Use the raw filename verbatim.
  • (same as <other>) placeholders. Reuse the referenced row's variants verbatim.
Optional: add custom carrier details

After the baseline + flash config are chosen, resolve whether the profile should record a custom carrier on top. If quick_start_prefill.target.custom_carrier is a valid concrete intent, consume it directly:

  • add_custom_carrier → record custom_carrier: and collect the missing carrier metadata below.
  • no_custom_carrier → omit custom_carrier: and proceed to the "Confirm" step.
  • keep_existing_custom_carrier is valid only when editing an existing active profile that already has custom_carrier:. Because this skill authors a new profile, treat it as ambiguous and ask the custom-carrier intent question below.
  • skip or a missing prefill value → ask the custom-carrier intent question below.

When intent is not already concrete, ask the user whether they also want to record a custom carrier on top using the AskUserQuestions UI. This is a bounded choice; show add and skip only:

Optionally add custom-carrier details for a customer-designed carrier using this Jetson module? (add / skip, default skip)

If the resolved intent is skip, proceed to the "Confirm" step with the custom_carrier: block omitted.

If the resolved intent is add, collect every still-missing custom_carrier: value through the AskUserQuestions UI. Do not collect custom-carrier metadata with plain-text prompts, inline chat questions, freeform summary confirmation, inferred defaults, or example values presented as selectable options. Split the collection into multiple AskUserQuestions forms if the UI question limit would otherwise be exceeded.

For the first AskUserQuestions form, ask custom_carrier.name, custom_carrier.id, custom_carrier.sku, and custom_carrier.revision in template document order. For these four fields, the only selectable path shown to the user must be the tool's built-in Type something row. Do not add example values, "Other", guessed board IDs, guessed SKUs, common revision strings, or any explicit domain options. These are customer-provided metadata: preserve them exactly as entered and do not require or validate any naming convention, prefix, case, character set, digit count, or SKU format. Validate only that required fields are non-empty unless the user explicitly enters NA. NA is accepted per the contract for required fields; warn that downstream skills may refuse if a missing field is required for their edit.

After custom_carrier.name is known, collect custom_carrier.flash_config through a second AskUserQuestions form if it is still missing. Show the suggested default as the only explicit choice and allow the built-in Type something row for a custom .conf filename. Require a non-empty filename ending in .conf; do not derive it from board ID, SKU, or any guessed convention.

Field-specific behavior (defaults / suggestions not expressible via markers alone):

  • custom_carrier.flash_config — suggest a default of kebab-cased custom_carrier.name + ".conf" (e.g. "Acme Vision X1" → acme-vision-x1.conf).
  • custom_carrier.name — prompt: "Type the customer-facing custom carrier or product name exactly as you want it recorded. No format is required." Use Type something only.
  • custom_carrier.id — prompt: "Type the custom carrier board ID or project ID exactly as you want it recorded. No format is required." Use Type something only.
  • custom_carrier.sku — prompt: "Type the custom carrier SKU or variant identifier exactly as you want it recorded. No format is required." Use Type something only; quote in the written YAML so numeric-looking values and leading zeros survive.
  • custom_carrier.revision — prompt: "Type the custom carrier revision or build identifier exactly as you want it recorded, or press Enter to omit it. No format is required." Use Type something only; omit when skipped.

If the user does not know a value, suggest reading /proc/device-tree/chosen/ids on the booted target. If genuinely unavailable, accept NA for required fields or omit optional fields.

The template intentionally omits custom_carrier.module and custom_carrier.chip_family — module identity comes from reference_devkit.module (the same physical Jetson module plugs into both carriers). Don't add them back.

Show full SKILL.md (732 more words)Show less
Confirm

Show the proposed YAML. Shape depends on whether the user added custom-carrier details in the "Optional: add custom carrier details" step. Example: see references/ui-samples.md for the reference-only and reference + custom-carrier YAML shapes, plus the note on which sibling skill appends which block.

Suggested filename:

  • Reference-only: kebab-cased product name from the selected catalogue row, with .yaml appended. E.g. Jetson AGX Thor T5000 → jetson-agx-thor-t5000.yaml; Jetson AGX Orin 32GB → jetson-agx-orin-32gb.yaml. Do not derive the filename from flash_config; multiple products can share one default conf.
  • With custom carrier: kebab-cased custom_carrier.name. E.g. "Acme Vision X1" → acme-vision-x1.yaml.

Use the suggested filename without asking when it is unique and was generated from the selected platform or custom-carrier name. Print the filename and continue. Ask only on collision, empty generated name, ambiguous input, or explicit custom-filename request. Then handle activation:

  • Bootstrap case (no profile YAMLs existed at the "Survey current state" step): do not prompt. The new profile is the first one, so it is auto-activated in the "Write" step. Tell the user this is happening so the auto-activation is not silent.
  • Profiles already exist: ask whether to activate the new profile (default: yes). Decline only if the user is authoring multiple profiles to choose from later — they can activate any of them with jetson-set-target.
Write (and optionally activate)
  1. Write target-platform/<filename>.yaml. If the file already exists, ask before overwriting; never silently clobber.

  2. Edit target-platform/active_target.yml to set active: <filename>.yaml when either of the following holds:

    • This was the bootstrap case from the "Survey current state" step (no profile YAMLs existed before this run) — activation is automatic, regardless of any prior user input.
    • Profiles already existed and the user accepted activation in the "Confirm" step.

    Preserve the header comment block; only modify the active: line. If profiles already existed and the user declined activation, leave active_target.yml untouched.

Confirm

Print a summary: profile path, active pointer (or "unchanged" if activation was declined), Tegra family (T234 / T264), whether custom_carrier: was recorded, and a reminder that downstream skills will resolve to whichever profile is active. If the user authored multiple profiles without activating, point them at jetson-set-target to pick one.

Next steps (Setup → Customize chain):

  1. /jetson-init-image — extract the BSP and append bsp_image:.
  2. /jetson-init-source — clone the shared repos (and optionally override <workspace>/Source if the user wants source.root_path written into the profile).
  3. /jetson-link-docs (optional) — register paths to pre-downloaded BSP / schematic / pinmux documents in the profile's documents: block.

Then, before any customize-* skill (Customize):

  1. /jetson-generate-kb — KB that customize-* skills consult for target-specific file locations.
  2. /jetson-derive-carrier (custom carriers only) — materialize the custom carrier's flash conf + supporting DTSIs from the reference devkit. Required before per-knob customize-* on a custom carrier.

If a downstream skill triggered this run, tell the user to re-issue their original request; do not silently re-trigger it.

Prerequisites

  • bsp-platforms-catalogue.md and platform_template.yaml reachable via the relative paths in the References section.
  • Write access to target-platform/ (and target-platform/active_target.yml if activation is requested).
  • For "what module/carrier is this?" lookups: /proc/device-tree/chosen/ids on the booted device.

Limitations

  • Owns only reference_devkit:, custom_carrier:, and the active_target.yml pointer; refuses to write bsp_image:, source:, or documents:.
  • Will not invent SKUs, module IDs, custom carrier metadata, or flash configs — records NA or aborts when the user does not know a required value.

Troubleshooting

  • User asks to switch instead of author — refuse and route to jetson-set-target; do not re-author an existing profile.
  • Catalogue row not found for the product the user named — add the row to bsp-platforms-catalogue.md first; do not patch the product list inside this skill.
  • Required field stuck at NA — downstream customize-* skills may refuse; capture the missing value (read /proc/device-tree/chosen/ids) and rerun.
  • active_target.yml lost its comment block after activation — a non-round-tripping YAML writer was used; switch to ruamel.yaml and restore from the template.

See references/gotchas.md for the full pitfalls list (catalogue/template separation, NA handling, conf collision rules, etc.).

References

© NVIDIA, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 6 other files (references) in skills/jetson-init-target of NVIDIA/skills.

  • SKILL.md
  • BENCHMARK.md
  • evals/evals.json
  • references/gotchas.md
  • references/ui-samples.md
  • skill-card.md
  • skill.oms.sig

Open the folder on GitHubat commit 67a13c0

Compare with similar skills

Jetson Init Target next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Jetson Init Target compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Jetson Init Target this skillNVIDIA/skills3.5k—~4.8kAutomated safety check: PassApache-2.0
Fla Triton To Gluonfla-org/flash-linear-attention5.8k—~4.2kAutomated safety check: PassMIT
DGX Spark Memory and Thermal Opswshobson/agents40k1 repos~2kAutomated safety check: PassMIT
DGX Spark Training Gotchaswshobson/agents40k1 repos~2kAutomated safety check: PassMIT
Megatron-LM on SLURMNVIDIA/Megatron-LM18k—~1.8kAutomated safety check: PassApache-2.0
OpenVLA-OFT Fine-TuningOrchestra-Research/AI-Research-SKILLs13k1 repos~3.7kAutomated safety check: PassMIT

Similar skills

  • Fla Triton To Gluon

    fla-org/flash-linear-attention

    Workflow for porting an existing Triton kernel in fla/ops/ to Gluon (triton.experimental.gluon) to gain explicit control over tensor layouts, shared memory, async data movement (cp.async / TMA), MMA…

    5.8k GitHub stars~4.2k tokensUpdated today
    AI & LLM EngineeringAuto-check passed
  • Plans memory headroom, works through out-of-memory failures and watches temperature and power during long ML training jobs on NVIDIA DGX Spark.

    40k GitHub starsUsed in 1 repo~2k tokens
    AI & LLM EngineeringAuto-check passed
  • Preflight checks and diagnosis for ten known failure modes of ML training on NVIDIA DGX Spark's GB10, spanning launch errors, memory, thermals, bandwidth and precision.

    40k GitHub starsUsed in 1 repo~2k tokens
    AI & LLM EngineeringAuto-check passed
  • Megatron-LM on SLURM

    NVIDIA/Megatron-LM

    Official

    Shows how to launch distributed Megatron-LM training on a SLURM cluster: sbatch skeleton, torch.distributed.run setup, CUDA_DEVICE_MAX_CONNECTIONS rules and failure diagnosis.

    18k GitHub stars~1.8k tokensUpdated today
    AI & LLM EngineeringAuto-check passed
  • OpenVLA-OFT Fine-Tuning

    Orchestra-Research/AI-Research-SKILLs

    Fine-tunes and evaluates OpenVLA-OFT and OFT+ robot policies with LoRA and continuous action heads on LIBERO simulation and ALOHA real-robot setups.

    13k GitHub starsUsed in 1 repo~3.7k tokens
    AI & LLM EngineeringAuto-check passed
  • Cosmos Policy Evaluation

    Orchestra-Research/AI-Research-SKILLs

    Sets up and runs NVIDIA Cosmos Policy evaluations on the LIBERO and RoboCasa simulators, including headless GPU rendering and inference latency profiling.

    13k GitHub stars~3.7k tokensUpdated 3 mo ago
    AI & LLM EngineeringAuto-check passed

More from NVIDIA/skills

All 380 skills in this repo
  • Official

    A skill your agent uses when the user wants to deploy, run, debug, tear down, or call the REST API of the RTVI-CV 2D detection / tracking microservice.

    3.5k GitHub starsUsed in 1 repo~4.5k tokens
    Auto-check passed
  • Official

    Generates, validates, compares and explains HOLOLINK_def.svh macro files for the HSB IP, using bundled Python scripts and asking before it writes anything.

    3.5k GitHub stars~2.9k tokensUpdated today
    Auto-check passed
  • Official

    Runs and validates an end-to-end Mission Control demo in a locally installed Isaac Sim, with a Nova Carter robot driven through a Python server.

    3.5k GitHub stars~4.8k tokensUpdated today
    Auto-check passed
  • Orchestrates defect image generation for PCBA, metal surface and glass inspection with NVIDIA Cosmos AnomalyGen on OSMO, from cold-start Day 0 to real-photo Day 1 labeling.

    3.5k GitHub stars~5k tokensUpdated today
    Auto-check: notes
  • Orchestrates video data augmentation and auto-labeling workflows on OSMO, from flow selection and preflight checks to submission, monitoring and output download.

    3.5k GitHub stars~4.7k tokensUpdated today
    Auto-check: notes
  • Official

    Runs NVIDIA TAO Data Services KPI analysis on object detection results, comparing predictions to ground truth and writing per-class precision, recall and AP to a CSV.

    3.5k GitHub stars~2.7k tokensUpdated today
    Auto-check: notes

Questions about Jetson Init Target

What does Jetson Init Target do?

Author a new Jetson target-platform profile (referencedevkit + optional customcarrier) and update the active pointer. Jetson Init Target is an agent skill from NVIDIA/skills, published by the product's own GitHub organization. Author a new Jetson target-platform profile (referencedevkit + optional customcarrier) and update the active pointer.

When should I use Jetson Init Target?

Jetson Init Target fits situations like: create a target; not for switching existing profiles.

How do I install Jetson Init Target in Claude Code?

Run `npx skills add NVIDIA/skills --skill jetson-init-target -a claude-code`. Or copy the skill folder (skills/jetson-init-target in NVIDIA/skills) into .claude/skills/jetson-init-target in your project. Claude Code loads it when a task matches its description.

How do I install Jetson Init Target in Codex?

Run `npx skills add NVIDIA/skills --skill jetson-init-target -a codex`. Or copy the skill folder (skills/jetson-init-target in NVIDIA/skills) into .agents/skills/jetson-init-target in your project. Codex loads it when a task matches its description.

Can I use Jetson Init Target in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add NVIDIA/skills --skill jetson-init-target -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/jetson-init-target, .gemini/skills/jetson-init-target, .github/skills/jetson-init-target and .opencode/skills/jetson-init-target in your project.

What does Jetson Init Target need to run?

SKILL.md names no scripts, command-line tools or credentials: Jetson Init Target is instructions for the agent only.

Does Jetson Init Target access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Jetson Init Target safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Jetson Init Target use?

Jetson Init Target is published under the Apache-2.0 licence (declared in SKILL.md). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Jetson Init Target use?

About 4.8k tokens (SKILL.md is roughly 19k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 1.2k tokens, read only when the agent opens those files.

What are the alternatives to Jetson Init Target?

Skills that share tags, products or a category with Jetson Init Target: Fla Triton To Gluon (fla-org/flash-linear-attention, 5.8k stars), DGX Spark Memory and Thermal Ops (wshobson/agents, 40k stars), DGX Spark Training Gotchas (wshobson/agents, 40k stars) and Megatron-LM on SLURM (NVIDIA/Megatron-LM, 18k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Jetson Init Target?

NVIDIA (a GitHub organization, an official publisher) maintains it in NVIDIA/skills, which has 3,539 GitHub stars. The repository holds 380 skills in this directory. The repository was last updated on October 7, 2026.

Source: NVIDIA/skills on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.