Add Physical AI Skill Hub component#125
Conversation
Signed-off-by: Doyub Kim <doyopk@nvidia.com>
|
looks good. let's gooo |
mosheabr
left a comment
There was a problem hiding this comment.
Approved. Physical AI Skill Hub source repo is fully prepared — 4 of 5 declared skills are 5/5 artifact-compliant (omniverse-cad-to-simready, omniverse-realtime-viewer, omniverse-usd-performance-tuning, physical-ai-infrastructure-setup-and-resilient-scaling). The 5th (physical-ai-neural-reconstruction) is missing evals.json and will be dropped by sync enforcement until evals lands. Doyub flagged this transparently in the PR body — appreciate the honesty.
Note: source repo is currently internal (NVIDIA-dev/physical-ai-skill-hub-dev), but is explicitly allowlisted in nvskills-ci config (allowed_owners + per-repo entry). Worth a follow-up to add links.discussions: false in a small subsequent PR — repo has Discussions disabled and the help table would otherwise render a broken link.
One-time content bridge while the catalog sync workflow's GITHUB_TOKEN gains cross-org read scope for internal NVIDIA-dev repos. Without that access, the sync logs "Clone failed for NVIDIA-dev/physical-ai-skill-hub-dev" on every run and these skills cannot land via the normal mirroring path, despite being fully 5/5 compliant in source. Skills staged (all sig + skill-card + evals + BENCHMARK + SKILL.md from the source repo's main branch): - omniverse-cad-to-simready - omniverse-realtime-viewer - omniverse-usd-performance-tuning - physical-ai-infrastructure-setup-and-resilient-scaling - physical-ai-neural-reconstruction Two source skills are intentionally excluded from the catalog and NOT staged here: - physical-ai-defect-image-generation (excluded by policy) - physical-ai-video-data-augmentation (excluded by policy) Behavior of the sync workflow with this manual content in place: the sync will continue to log "Clone failed" for this component and skip the source-side refresh, but the existing catalog content under skills/<skill-name>/ is preserved (the sync only writes for successful clones — it does not remove orphaned content). Once cross-org token access lands, the next sync will overwrite this manually-staged content with the live source-of-truth state. components.d/physical-ai-skill-hub.yml registered the same 5 skills in PR #125; no yml changes here. Signed-off-by: Moshe Abramovitch <moshea@nvidia.com>
Onboarding type
components.d/<slug>.ymlfile)For new product onboarding — author affirmations
By submitting this PR, I confirm on behalf of my team:
.agents/skills/orskills/path used for new entries (or existing path retained for legacy entries percomponents.d/<slug>.yml)Reviewer checklist (OSS Skills PIC)
components.d/<slug>.ymlentry valid (required fields, uniquecatalog_dir, path exists in source repo, filename slug matches name)SKILL.mdfrontmatter spec-compliant (at least one sampled)All PRs
git commit -s).If you forgot, run
git rebase --signoff origin/main && git push --force-with-leaseto retroactively sign all commits in your branch.Other context
Adds the first-round Physical AI Skill Hub sync entries:
omniverse-cad-to-simreadyomniverse-realtime-vieweromniverse-usd-performance-tuningphysical-ai-infrastructure-setup-and-resilient-scalingphysical-ai-neural-reconstructionSync gate note:
physical-ai-neural-reconstructionhas anevals.jsonbefore the sync gate runs; the current source tree showsSKILL.md,skill-card.md,skill.oms.sig, andBENCHMARK.md, but noevals.jsonwas found.