simready-foundation-conform-fet-001-minimal

작성자: nvidia

Use for repairing SimReady Minimal assets: units, upAxis, defaultPrim, hierarchy, mesh quality, extents, and origin placement.

npx skills add https://github.com/nvidia/simready-foundation --skill simready-foundation-conform-fet-001-minimal

SimReady Conform FET-001 Minimal

Purpose

Use this workflow skill after FET000_CORE passes and the next profile validation gate is FET001_BASE_NEUTRAL. It brings an existing USD asset toward the Minimal/Base Neutral feature contract while preserving visual scale, hierarchy intent, and source traceability.

This is an authoring and repair skill, not the final validator. Work on a staged copy under the requested output directory, rerun the same validation gate after each repair stage, and stop at the first remaining failing FET001 validation gate.

Prerequisites

Read the source-of-truth files named below before editing. Work on staged outputs where the skill requires them, and keep validation evidence with the result.

Source of Truth

Before changing an asset, load the exact FET001 manifest selected by the profile:

  • nv_core/sr_specs/docs/features/FET_001_base_neutral-0.1.0-minimal.json
  • nv_core/sr_specs/docs/features/FET_001_base_neutral-1.0.0-minimal.json
  • nv_core/sr_specs/docs/features/FET_001-minimal.md

Some profiles select FET001_BASE_NEUTRAL@0.1.0, while newer profiles may select 1.0.0. Treat the JSON manifest requirement list as authoritative for the selected profile/version.

For per-requirement repair details, read references/fet001-requirements.md when a FET001 validation report or inspection identifies matching failures.

Inputs

Collect these before editing:

InputRequirement
usd_assetRequired .usd, .usda, .usdc, or unpacked USD-family asset to repair.
output_rootRequired or inferred folder for staged assets and reports.
simready_profileProfile being validated, such as Prop-Robotics-Neutral.
profile_versionProfile version, if supplied by the user or validation command.
validation_reportPreferred JSON or markdown report from the failing profile/feature validation gate.
source_assetOriginal CAD, DCC, URDF, MJCF, or conversion input path, used for provenance.
physical_size_policyWhether to preserve physical size. Default to preserve size.

Instructions

Use this checklist when changing the repository:

  1. Confirm the input asset exists and that FET000_CORE is already passing on the same staged asset. If Core still fails, hand back to simready-foundation-conform-fet-000-core first.
  2. Parse the validation report and filter to FET001_BASE_NEUTRAL failures. Do not repair later profile features in this skill.
  3. Load the selected FET001 manifest version and the requirement repair map.
  4. Create a staged output folder under output_root; do not mutate the source unless the user explicitly asks for in-place repair.
  5. Run or inspect minimum USD metadata: stage openability, default prim, root prims, upAxis, metersPerUnit, composed bounds, mesh/geometry counts, authored references, and used layers.
  6. Apply deterministic fixes in this order:
    • Atomic asset/file fixes for AA.001 and AA.002.
    • Hierarchy fixes for HI.001, HI.003, and HI.004.
    • Unit and orientation fixes for UN.001, UN.002, UN.006, and UN.007.
    • Minimal visual geometry fixes for VG.001 or VG.MESH.001.
    • Mesh quality fixes for VG.002, VG.014, VG.027, VG.028, and VG.029.
    • Origin placement fixes for VG.025 only when the asset placement convention is clear.
  7. Rerun the same profile validation gate, or the narrowest available FET001 validation gate.
  8. Summarize the stage as passed, failed, skipped, or blocked. Stop when FET001 passes or the next FET001 failure requires upstream conversion, source geometry, or user intent.

Unit Repair Policy

Treat UN.007 with extra care. Setting metersPerUnit = 1.0 without adjusting authored coordinates can change the apparent physical size of an asset.

Default policy:

  • Preserve physical size.
  • Record the original metersPerUnit, composed bounds, default prim, and root transforms before editing.
  • If the asset uses millimeters, such as metersPerUnit = 0.001, normalize to meter units by scaling authored linear data by 0.001 or by applying an equivalent root-scale normalization in the same staged USD file.
  • Recompute authored extents after any geometry-space rescale.
  • Validate that the composed bounds in meters are unchanged within a small tolerance.

Prefer these approaches in order:

StrategyUse WhenNotes
Re-export from converter with meter unitsConversion options are available and reliable.Best long-term fix because geometry, metadata, and derived data are generated consistently.
Rescale staged USD dataAll relevant layers are local and editable.Multiply points, translate ops, extents, and other linear authored values by old_meters_per_unit / 1.0, then set stage metersPerUnit = 1.0.
Apply root-scale normalization in the same staged USDThe asset root has no existing authored xform ops and the selected FET001 version does not reject a root scale.Set stage metersPerUnit = 1.0 and add a root scale op equal to old_meters_per_unit / 1.0. This strategy is useful for millimeter- or centimeter-authored converter output when a root scale is acceptable.
Create a meter-unit wrapperThe package/validator contract allows a local USD payload reference without violating Core or atomic-asset rules.Verify FET000_CORE and AA.001 immediately. Current validators may reject extra USD files under simready_usd (NP.005) or ../payloads/... references as outside the asset root (AA.001).
Metadata-only changeThe coordinates are already in meters, or the user accepts a physical size change.Report this explicitly because it can make the asset too large or too small.

For any asset with metersPerUnit = 0.001, a size-preserving single-file repair can set metersPerUnit = 1.0 and author root op xformOp:scale:meter_normalization = (0.001, 0.001, 0.001), assuming the root transform is otherwise safe to modify.

Examples

Example request:

Repair FET001_BASE_NEUTRAL minimal stage failures on a staged USD asset.

Expected result summary:

staged_asset: repaired copy or output directory
validation: selected feature/profile gate and report path
remaining_failures: next failing requirement IDs, if any

Repair Policy

Make automatic repairs only when the result is mechanical and locally verifiable:

  • Set missing upAxis to Z when the asset is already visually Z-up or when the selected profile requires Z-up.
  • Set defaultPrim to the single root prim when there is exactly one valid root.
  • Wrap multiple roots under a new Xform only when all asset content can be moved or referenced without breaking paths.
  • Convert search paths to anchored relative paths only after the dependency is inside the staged asset root.
  • Author extents using USD-computed bounds for local editable meshes.
  • Normalize or generate mesh normals only when the intended smoothing can be inferred or the converter already produced enough topology to compute them deterministically.
  • If adding a root-scale normalization, check for existing root xform ops first. If the root already has meaningful transform ops, preserve order and semantics or use data-space rescaling instead.

Block and report instead of guessing when:

  • The feature failure comes from missing or non-mesh geometry that requires upstream conversion.
  • Unsupported file types require lossy texture/audio/geometry conversion.
  • Winding, normal direction, or origin placement cannot be inferred from local geometry.
  • Rescaling would require editing external layers that are not staged with the asset.
  • A unit wrapper or payload layout causes FET000_CORE or AA.001 to regress.

Validation Handoff

Preserve reports under the staged output directory. If the Physical AI Skill Hub validation commands are available, use:

uv run --python 3.12 validate-usd-minimum <staged-usd> \
  --report <output-root>/minimum-usd-after-fet001.json

uv run --python 3.12 validate-simready-profile <staged-usd> \
  --profile <profile> \
  --profile-version <version> \
  --foundation-root <simready-foundation-root> \
  --foundation-spec-root <simready-foundation-root>/nv_core/sr_specs/docs \
  --report <output-root>/simready-profile-after-fet001.json

Count this skill as successful when FET001_BASE_NEUTRAL passes, even if the full profile still fails on later features such as rigid bodies, multibody physics, or grasp vectors. Report those remaining failures as handoff work for their own skills.

Limitations

  • Do not silently mutate the source asset; work on the requested staged output.
  • Do not hide later profile failures after the selected feature gate passes or fails.
  • Do not invent geometry, metadata, or runtime behavior that conflicts with the asset intent.

Troubleshooting

  • Error: validation tooling is unavailable. Solution: run the narrowest available USD or static check and report the gap.
  • Error: a repair would change asset intent. Solution: stop and ask for direction or stage the smallest reversible edit.
  • Error: later profile gates still fail. Solution: report the next failing feature and hand off to the matching conformance skill.

Resources

  • assets/openai.yaml preserves optional UI metadata for clients that read skill display hints. It is not required for the workflow.
  • references/ contains detailed requirement notes; load only the files needed for the active validation failure.

Summary Format

Report:

FieldMeaning
input_usd_pathOriginal USD path.
output_usd_pathLatest staged/repaired USD path.
profile and profile_versionValidation target.
fet001_versionSelected FET001_BASE_NEUTRAL manifest version.
requirements_repairedRequirement IDs changed by this skill.
requirements_blockedRequirement IDs that need user intent or upstream conversion.
scale_preservedWhether composed physical bounds were preserved after unit repair.
validation_reportPath to the rerun validation report.
next_stepUsually the next failing profile feature.

Keep the user-facing summary short: what was fixed, whether the asset scale was preserved, what FET001 result validation reported, and the first remaining validation gate.

nvidia의 다른 스킬

compileiq-debug
nvidia
Use when something is wrong: Search() hangs, all evaluations return INVALID_SCORE, scores aren't improving, every config returns the same number, ptxas errors…
official
create-github-pr
nvidia
gh CLI를 사용하여 GitHub 풀 리퀘스트를 생성합니다. 사용자가 새 PR을 만들거나, 코드 리뷰를 제출하거나, 풀 리퀘스트를 열고자 할 때 사용합니다. 트리거 키워드 -…
official
diagnose-perf
nvidia
First-responder performance triage for Isaac Sim and Isaac Lab. Identifies bottleneck category (GPU-bound, CPU-bound, VRAM, loading) using nvidia-smi and…
official
eagle3-review-logs
nvidia
Review EAGLE3 pipeline experiment logs from the launcher's experiments/ directory. Summarizes pass/fail status for all 4 tasks, diagnoses failures with root…
official
nemoclaw-maintainer-cross-issue-sweep
nvidia
다른 열린 이슈들을 스캔하여 주어진 PR이 함께 수정하거나 실수로 망가뜨릴 수 있는 이슈를 찾습니다. 인접 수정 기회와 모순 위험을 file:line…과 함께 출력합니다.
official
karpathy-guidelines
nvidia
일반적인 LLM 코딩 실수를 줄이기 위한 행동 지침입니다. 코드 작성, 검토 또는 리팩토링 시 과도한 복잡성을 피하고 정밀한 변경을 위해 사용하세요.
official
fhir-basics
nvidia
에이전트에게 FHIR R4 API의 작동 방식, 사용 가능한 리소스, 검색 매개변수를 사용한 쿼리 방법, 모든 응답 형식을 올바르게 파싱하는 방법을 가르칩니다…
official
underdeclared-agent
nvidia
A helpful assistant agent
official