Skip to content

Publish CAD-to-SimReady 0.2.0 - #445

Merged
mosheabr merged 1 commit into
NVIDIA:mainfrom
doyubkim:codex/publish-cad-to-simready-0.2.0
Aug 12, 2026
Merged

Publish CAD-to-SimReady 0.2.0#445
mosheabr merged 1 commit into
NVIDIA:mainfrom
doyubkim:codex/publish-cad-to-simready-0.2.0

Conversation

@doyubkim

Copy link
Copy Markdown
Contributor

Onboarding type

  • New product onboarding (new components.d/<slug>.yml file)
  • Other (catalog change, README fix, infrastructure, etc.)

Public release affirmations for this update

  • The CAD-to-SimReady 0.2.0 skills/ content is cleared for public OSS release.
  • License: Apache 2.0.
  • No unreviewed new license or third-party component is introduced.
  • The existing skills/omniverse-cad-to-simready catalog path is retained.

The source repository is internal and this skill uses the existing Physical AI direct-publication exception in .github/scripts/manual-components.yml.

Reviewer checklist (OSS Skills PIC)

  • SKILL.md frontmatter sampled and confirmed spec-compliant.
  • No new license or third-party dependency requiring OSRB filing.
  • Signed content and public-release scope confirmed.

All PRs

  • All commits signed off with DCO (git commit -s).

Other context

Summary

Publish CAD-to-SimReady 0.2.0 by replacing the existing 0.1.0 catalog directory with the exact signed snapshot from NVIDIA-dev/physical-ai-skill-hub-dev@fed157be.

This update includes the 0.2.0 changelog, refreshed evaluation and benchmark evidence, pinned public upstream versions, current conversion and preflight workflows, and the reviewed version-stable SkillSpector baseline.

Source release PR: NVIDIA-dev/physical-ai-skill-hub-dev#275.

Validation

  • Catalog directory is byte-for-byte identical to the signed source snapshot.
  • Signature manifest covers all 129 non-signature files.
  • Zero signature hash mismatches, missing files, or unsigned extras.
  • No NVIDIA-dev references remain in the published skill directory.
  • git diff --check passes.
  • Source NVSkills CI and required source-repository checks passed before merge.

Draft follow-ups

Signed-off-by: Doyub Kim <doyopk@nvidia.com>
mosheabr added a commit that referenced this pull request Aug 12, 2026
The --no-ai case asserted on main()'s bare exit code. --check returns 1
for output drift as well as for skill warnings, so any PR that edits
skill content failed this test for reasons unrelated to enrichment.
PR #445 is the first to hit it: 70 changed files including SKILL.md at
62+/147-, which legitimately drifts the checked-in metadata.

Neutralise diff_text for the duration of the run so the exit code is
attributable to the warning gate alone. Drift is already covered by
case 4.

A first attempt simply excused drift with 'rc == 0 or drifted'. That was
worse than the original bug: removing the --no-ai gate entirely still
passed, because drift was present. Mutation-checked properly this time.

Verified against PR #445's content: reproduces the failure without this
change, passes with it, and still fails when the --no-ai gate is removed
-- on both #445's tree and a clean one.

Signed-off-by: Moshe Abramovitch <moshea@nvidia.com>
@mosheabr

Copy link
Copy Markdown
Collaborator

Merging via admin override. Recording why, since the normal gate did not run.

Why the gate is absent. This is a fork pull request (doyubkim/skills). Fork-based signing was reverted on 2026-08-06 in #422, so NVSkills CI structurally cannot post a status here — the failing require-nvskills-ci check is not a transient failure and re-running will not clear it. The team publishes from an internal repository and cannot move the branch into NVIDIA/skills.

Precedent. This is the same path used in #150 on 2026-05-29, which staged these five Physical AI skills into the catalog as a "one-time content bridge" because the sync could not clone NVIDIA-dev/physical-ai-skill-hub-dev. omniverse-cad-to-simready is one of the five, and it remains listed in .github/scripts/manual-components.yml and absent from components.d/, so direct publication is the sanctioned path for it. What has changed is the reason the normal path is unavailable, not the situation.

What I verified by hand, in place of the gate:

  • Signature covers the content: 129/129 resources match, subject omniverse-cad-to-simready, checked resource by resource against the fork branch
  • BENCHMARK.md: Overall verdict: PASS, evaluated 2026-08-12. Codex 48% → 82% (+33). Claude Code reports "Not available", so this is single-agent evidence
  • SPDX headers on every new .py file: Apache-2.0 with NVIDIA CORPORATION & AFFILIATES
  • Manual-staging claim confirmed against manual-components.yml
  • New config/skillspector-baseline.yaml: 7 suppressions, each narrow and individually justified — 3 for os.environ.copy() passed to local subprocesses, 4 for .env string matches in documentation. No fingerprint-based blanket suppressions. Spot-checked the largest claim in preflight.py, which derives redaction values from the environment specifically to scrub secrets from output. I spot-checked these; I did not audit all 2750 lines
  • 68 files, all within skills/omniverse-cad-to-simready/. Nothing outside

What this does not mean. Fork pull requests are not supported. This merged on a manual verification, not because the gate passed. The durable fixes are #428, which restores fork support but needs allow_fork_pull_requests enabled on the nvskills-ci side, or a deliberate carve-out for manually-staged skills, whose signatures travel with the content and are independently verifiable. This is the fourth team to hit the fork limitation, and #150 described this path as one-time — it is now the standing mode for five skills, which is worth resolving properly rather than case by case.

@doyubkim
doyubkim marked this pull request as ready for review August 12, 2026 20:53
@mosheabr
mosheabr merged commit aa11667 into NVIDIA:main Aug 12, 2026
7 of 10 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants