What does this skill do, and when should you use it?
This skill authors a new Jetson target-platform YAML profile rather than switching among existing profiles. It selects a reference devkit, module, and flash configuration from the platform catalogue, then optionally records custom-carrier metadata. It writes target-platform/<name>.yaml and can update active_target.yml. Its scope is limited to reference_devkit, custom_carrier, and the active pointer; it does not author BSP image, source, or document blocks.
Reads the platform catalogue, profile template, and current active pointer; presents and resolves a reference Jetson product and flash configuration; optionally collects custom-carrier name, ID, SKU, revision, and a .conf filename; writes a new target-platform/<name>.yaml; updates target-platform/active_target.yml on bootstrap or after activation confirmation; refuses to write bsp_image, source, or documents blocks.
- A Jetson project needs its first target-platform profile for a reference development kit.
- A product uses a customer-designed carrier with an existing Jetson module and needs that carrier recorded on a reference baseline.
- A downstream BSP workflow reports no active target and no target profile exists yet.
- A team wants to author several target profiles before selecting one with jetson-set-target.
How do you install this skill?
- Before execution, verify that the referenced catalogue, template, and contract files exist and match this skill revision.
- Add backup or recoverable-write procedures and specify recovery behavior if active_target.yml activation fails.
- Align the skill license metadata with the repository's dual-license statement and add formal examples, negative-trigger tests, and explicit maintenance ownership/change history.
- The workflow depends on AskUserQuestions-style agent UI capabilities; agents without that UI may not reproduce the procedure faithfully.
- Local filesystem
Install the individual skill with the documented CLI pattern: npx skills add nvidia/skills --skill jetson-init-target --yes. The README does not specify a separate installation directory; the CLI prompts for the destination.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/NVIDIA/skills.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/skills/jetson-init-target" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- create a new Jetson target platform profile
After the skill is loaded, ask the agent to “create a new Jetson target platform profile” and provide the reference platform plus any custom-carrier details. This skill authors new profiles; it does not switch existing profiles, which is handled by jetson-set-target. For a custom carrier, provide its name, ID, SKU, revision, and a flash-config filename ending in .conf.
What are this skill's strengths and limitations?
- Clearly separates profile creation from profile switching.
- Derives platform, SKU, and flash-configuration data from the catalogue instead of inventing values.
- Supports both reference-only and reference-plus-custom-carrier profiles.
- Can automatically activate the first profile created.
- Requires the referenced catalogue, templates, and active-pointer files to be reachable at their relative paths.
- Requires write access to target-platform/ and, when activated, active_target.yml.
- Does not prepare the BSP image, source repositories, documents, or knowledge base.
- The supplied SKILL.md provides no test suite or validated-platform matrix; missing required custom-carrier metadata may cause downstream skills to refuse the profile.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| Jetson Target Platform Initializer this page ✓ NVIDIA · Official | 57 · Use with care | ★ 3.8k | today | Apache-2.0 |
| Jetson Target Switcher ✓ NVIDIA · Official | 60 · Recommended | ★ 3.8k | today | Apache-2.0 |
| Jetson UPHY Lane Allocation ✓ NVIDIA · Official | 50 · Use with care | ★ 3.8k | today | Apache-2.0 |
| Jetson BSP Info Check ✓ NVIDIA · Official | 59 · Recommended | ★ 3.8k | today | Apache-2.0 |
| Jetson USB Port Customization ✓ NVIDIA · Official | 55 · Use with care | ★ 3.8k | today | Apache-2.0 |
Its role differs from jetson-set-target: this skill creates a new target-platform profile, while jetson-set-target switches the active pointer among existing profiles.
How did FollowSkills review this skill?
The skill narrowly scopes writes to target-platform configuration and the active pointer, requires confirmation for new files, overwrites, and activation, and forbids invented metadata; however, it lacks backup, rollback, or failure-recovery procedures, and bootstrap auto-activation remains an external effect, so full marks are not justified.
The workflow, edge cases, input validation, and routing of errors are described in substantial detail; however, the referenced catalogue, templates, and YAML-writing behavior were not actually verified in the supplied material, and no reproducible test suite or failure logs are provided, so the static ceiling applies.
Invocation conditions, non-fit cases, inputs, outputs, and boundaries with sibling skills such as jetson-set-target are clear; Chinese-language interaction support is not stated, and evidence for adaptation across agent UI capabilities is limited, so points are deducted.
The skill provides a version, license field, layered documentation, limitations, troubleshooting, references, and maintenance clues; however, formal Instructions and Examples sections are missing, the benchmark reports an author-format finding, version-change governance and maintenance ownership are not explicit, and the Apache-2.0 frontmatter differs from the repository's dual-license presentation.
The target task, output format, confirmation flow, and downstream steps are concrete, and the internal benchmark reports that core tasks can be completed; however, it covers only two positive tasks with no negative cases, and directly usable outputs are not verifiable from this static review, so the static ceiling limits the score.
The supplied skill files, evaluation report, evaluation tasks, and repository governance files are auditable evidence; however, there is no committed test suite, CI coverage, or independent corroboration, and the benchmark report cannot substitute for execution reproduction, so the score remains below the static maximum.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →