Agent Native

simready-foundation-conform-fet-001-standard

This skill repairs USD assets to conform to the SimReady Minimal Standard (FET_001_STANDARD) specification after basic validation passes.

by NVIDIAOfficial publisherNot reviewed

Identity proven by GitHub's verified organisation badge (NVIDIA, NVIDIA-NeMo)

Agents
Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, OpenCode, claude.ai and API
License
Apache-2.0 (permissive)
Pin
60b8bddpinned Oct 9, 2026
Context cost
about 3,100 tokenswhen the skill loads
GitHub stars
95on nvidia/simready-foundation
Scripts
None found

What it does

“Use for repairing SimReady Minimal assets: units, upAxis, defaultPrim, hierarchy, mesh quality, extents, and origin placement.”
The author's description, from SKILL.md

Generated summary from the skill's own text

This skill repairs USD assets to conform to the SimReady Minimal Standard (FET_001_STANDARD) specification after basic validation passes. It fixes units, axis orientation, hierarchy, mesh quality, extents, and origin placement while preserving physical scale and asset intent.

When to use it

  • Asset fails FET_001_STANDARD validation after passing FET_000_STANDARD
  • Need to repair USD hierarchy, units, or mesh geometry issues
  • Conforming SimReady assets to Minimal or Standard profile requirements
  • Handling converter output with non-meter units or missing metadata

When not to

  • Asset still fails FET_000_STANDARD core validation
  • Repair requires upstream CAD or DCC conversion
  • Changes would alter fundamental asset intent or geometry

Install

These commands install from the publisher's repository and need no account.

skills CLI

npx skills add nvidia/simready-foundation --skill simready-foundation-conform-fet-001-standard

Installs from the repository's default branch, which may have changed since the commit we checked.

GitHub CLI 2.90 or later

gh skill install nvidia/simready-foundation simready-foundation-conform-fet-001-standard@60b8bddb086f43c8cc394f50635b4e65b0ec36fd

Installs the commit we checked (60b8bdd).

Copy by hand

Download the folder at 60b8bdd and put it in .claude/skills/simready-foundation-conform-fet-001-standard for Claude Code and in .agents/skills/simready-foundation-conform-fet-001-standard for Codex, Cursor, Gemini CLI, GitHub Copilot and OpenCode.

Skill folder on GitHub at 60b8bdd

Claude Code plugin

/plugin install nvidia-skills@claude-plugins-official

The publisher's plugin in Claude's official marketplace. It may hold more than this skill, and we have not pinned it.

Copy for your agent

A prompt for any coding agent. It installs this skill from GitHub at the pinned commit, copies the files as they are, and runs nothing unless you ask.

Prompt

Install the agent skill "simready-foundation-conform-fet-001-standard" by NVIDIA.

Source, pinned to commit 60b8bddb086f43c8cc394f50635b4e65b0ec36fd:
https://github.com/nvidia/simready-foundation/tree/60b8bddb086f43c8cc394f50635b4e65b0ec36fd/skills/simready-foundation-conform-fet-001-standard

1. Download that folder at exactly this commit. Do not use the default branch.
2. Copy every file unchanged, keeping the folder structure, into .claude/skills/simready-foundation-conform-fet-001-standard/ for Claude Code and .agents/skills/simready-foundation-conform-fet-001-standard/ for Codex, Cursor, Gemini CLI, GitHub Copilot and OpenCode. Create the folders if they do not exist.
3. Do not edit the files, and do not run any script in the folder unless I ask you to.
4. When you are done, list the files you copied and where you put them.

Listing and checks: https://www.agentnative.dev/skills/nvidia/simready-foundation-conform-fet-001-standard

SKILL.md at the pin

---
name: simready-foundation-conform-fet-001-standard
description: "Use for repairing SimReady Minimal assets: units, upAxis, defaultPrim, hierarchy, mesh quality, extents, and origin placement."
license: Apache-2.0
metadata:
  author: "Shaad Boochoon <sboochoon@nvidia.com>"
  tags:
    - simready
    - conformance
    - minimal
---


# SimReady Conform FET_001_STANDARD Minimal

## Purpose
Use this workflow skill after FET_000_STANDARD passes and the next profile validation gate is `FET_001_STANDARD`. It brings an existing USD asset toward the Minimal/Standard 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 FET_001_STANDARD 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 FET_001_STANDARD manifest selected by the profile:

- `nv_core/tiers/simready_foundation_tier_core/simready/foundation/tier_core/features/FET_001_STANDARD-0.1.0.json`
- `nv_core/tiers/simready_foundation_tier_core/simready/foundation/tier_core/features/FET_001_STANDARD-1.0.0.json`
- `nv_core/tiers/simready_foundation_tier_core/simready/foundation/tier_core/features/FET_001_STANDARD-1.0.1.json`
- `nv_core/tiers/simready_foundation_tier_core/simready/foundation/tier_core/features/FET_001_STANDARD.md`

Profiles may select `FET_001_STANDARD@0.1.0`, `FET_001_STANDARD@1.0.0`, or `FET_001_STANDARD@1.0.1`. 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 FET_001_STANDARD validation report or inspection identifies matching failures.

## Inputs

Collect these before editing:

| Input | Requirement |
|---|---|
| `usd_asset` | Required `.usd`, `.usda`, `.usdc`, or unpacked USD-family asset to repair. |
| `output_root` | Required or inferred folder for staged assets and reports. |
| `simready_profile` | Profile being validated, such as `Robotics-Prop`. |
| `profile_version` | Profile version, if supplied by the user or validation command. |
| `fet001_version` | Optional explicit `FET_001_STANDARD` version. Default to latest checked-in version `1.0.1` when neither the user nor the selected profile/report specifies a version. |
| `validation_report` | Preferred JSON or markdown report from the failing profile/feature validation gate. |
| `source_asset` | Original CAD, DCC, URDF, MJCF, or conversion input path, used for provenance. |
| `physical_size_policy` | Whether to preserve physical size. Default to preserve size. |

## Instructions

Use this checklist when changing the repository:

1. Confirm the input asset exists and that FET_000_STANDARD is already passing on the same staged asset. If Core still fails, hand back to `simready-foundation-conform-fet-000-standard` first.
2. Select the FET_001_STANDARD version:
   - Use the version explicitly requested by the user when present.
   - Otherwise use the exact `FET_001_STANDARD` version selected by the profile or validation report.
   - Otherwise default to latest checked-in version `1.0.1`.
3. Load the selected manifest and `references/fet001-requirements.md`. Only repair requirements present in that selected manifest.
4. Parse the validation report and filter to `FET_001_STANDARD` failures. Do not repair later profile features in this skill.
5. Create a staged output folder under `output_root`; do not mutate the source unless the user explicitly asks for in-place repair.
6. Run or inspect minimum USD metadata: stage openability, default prim, root prims, `upAxis`, `metersPerUnit`, composed bounds, mesh/geometry counts, authored references, and used layers.
7. Apply deterministic fixes in this order, skipping requirement IDs that are not in the selected manifest:
   - 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.008`, `VG.014`, `VG.027`, `VG.028`, and `VG.029`.
   - Origin placement fixes for `VG.025` only when the asset placement convention is clear.
8. Rerun the same profile validation gate, or the narrowest available FET_001_STANDARD validation gate for the selected version.
9. Summarize the stage as passed, failed, skipped, or blocked. Stop when the selected FET_001_STANDARD version passes or the next FET_001_STANDARD failure requires upstream conversion, source geometry, or user intent.

## Version Selection

Default to `FET_001_STANDARD@1.0.1` for open-ended conformance requests such as "repair FET_001_STANDARD" or "make this Minimal compliant." Use `@1` or `@2` only when the user, profile, validation report, or migration task explicitly selects that older contract.

| Selected Version | Manifest | Conformance Target |
|---|---|---|
| `0.1.0` | `FET_001_STANDARD-0.1.0.json` | Original Minimal visual-presence contract: atomic asset, declared units/up axis, meter scale, imageable geometry, and default prim. |
| `1.0.0` | `FET_001_STANDARD-1.0.0.json` | Mesh-quality Minimal contract: version 0.1.0 retained requirements plus Z-up, mesh representation, extents, topology, winding, normals, single root, xformable root, and origin placement. |
| `1.0.1` | `FET_001_STANDARD-1.0.1.json` | Latest Minimal Standard contract: version 1.0.0 plus `VG.008` cached extent conformance. |

If an older report uses legacy feature versions, map them before loading the manifest:

| Legacy Version | Current Version |
|---|---|
| `0.1.0` | `1` |
| `1.0.0` | `2` |
| `1.0.1` | `3` |

## 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:

| Strategy | Use When | Notes |
|---|---|---|
| Re-export from converter with meter units | Conversion options are available and reliable. | Best long-term fix because geometry, metadata, and derived data are generated consistently. |
| Rescale staged USD data | All 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 USD | The asset root has no existing authored xform ops and the selected FET_001_STANDARD 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 wrapper | The package/validator contract allows a local USD payload reference without violating Core or atomic-asset rules. | Verify `FET_000_STANDARD` 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 change | The 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:

```text
Repair FET_001_STANDARD minimal stage failures on a staged USD asset.
```

Expected result summary:

```text
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 `FET_000_STANDARD` 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:

```bash
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 `FET_001_STANDARD` 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:

| Field | Meaning |
|---|---|
| `input_usd_path` | Original USD path. |
| `output_usd_path` | Latest staged/repaired USD path. |
| `profile` and `profile_version` | Validation target. |
| `fet001_version` | Selected `FET_001_STANDARD` manifest version; default is `1.0.1` unless explicitly overridden. |
| `requirements_repaired` | Requirement IDs changed by this skill. |
| `requirements_blocked` | Requirement IDs that need user intent or upstream conversion. |
| `scale_preserved` | Whether composed physical bounds were preserved after unit repair. |
| `validation_report` | Path to the rerun validation report. |
| `next_step` | Usually the next failing profile feature. |

Keep the user-facing summary short: what was fixed, whether the asset scale was preserved, what FET_001_STANDARD result validation reported, and the first remaining validation gate.

From nvidia/simready-foundation at 60b8bdd, shown as plain text under its license: Apache-2.0 (permissive). The license text is in skills/simready-foundation-conform-fet-001-standard/SKILL.md.

View SKILL.md on GitHub at 60b8bdd

Files

4 files, 24 KB in total. Each file has a sha256 in the pin's manifest; the first 12 characters are shown.

Skill folder on GitHub at 60b8bdd
All 50 skills from NVIDIA

More from NVIDIA

  • error-explainer

    This skill helps the agent understand and explain compiler and runtime errors.

    Claude Code, Codex, Cursor +4 · about 66 tokens · pinned 6 days ago

  • order-lookup

    This skill retrieves and modifies individual orders in an e-commerce system.

    Claude Code, Codex, Cursor +4 · about 100 tokens · pinned 6 days ago

  • debug-openshell-cluster

    Diagnoses OpenShell gateway deployment problems across Docker, Podman, Kubernetes, and VM compute platforms.

    Claude Code, Codex, Cursor +4 · about 19.1K tokens · pinned 6 days ago

  • nemo-automodel-model-onboarding

    This skill guides adding or modifying model architecture support in NeMo AutoModel, covering discovery of model type from config, implementation of custom…

    Claude Code, Codex, Cursor +3 · about 5.6K tokens · pinned 6 days ago

  • nvflare-convert-lightning

    Converts existing PyTorch Lightning training code into an NVFLARE federated learning job by patching the trainer with the Lightning Client API, configuring…

    Contains scripts

    Claude Code, Codex, Cursor +3 · about 3K tokens · pinned 6 days ago

  • nvflare-convert-pytorch

    Converts existing plain PyTorch training code into NVFLARE federated learning jobs using the Client API for model exchange, local validation, and job export.

    Contains scripts

    Claude Code, Codex, Cursor +3 · about 2.5K tokens · pinned 6 days ago

All Coding workflow skills
  • planning

    This skill orchestrates multi-workflow projects with dependencies and shared data tables by creating a task graph in the orchestrator via the create-tasks tool.

    by n8nOfficial publisher

    Claude Code, Codex, Cursor +3 · about 2.4K tokens · pinned 6 days ago

  • progressive-building

    This skill implements incremental workflow development for agents building or extending workflows.

    by n8nOfficial publisher

    Claude Code, Codex, Cursor +3 · about 1.4K tokens · pinned 6 days ago

  • workflow-builder

    Builds and edits n8n workflows using the TypeScript SDK.

    by n8nOfficial publisher

    Claude Code, Codex, Cursor +3 · about 14.2K tokens · pinned 6 days ago

  • remotion-upgrade

    This skill upgrades Remotion and its related packages to the latest stable version.

    by RemotionOfficial publisher

    Claude Code, Codex, Cursor +3 · about 460 tokens · pinned 6 days ago

  • remotion-upgrade

    This skill upgrades Remotion and related packages across a project, automatically detecting the package manager and workspaces.

    by RemotionOfficial publisher

    Claude Code, Codex, Cursor +3 · about 460 tokens · pinned 6 days ago

  • final-release-review

    Audits Python SDK release candidates or plans against the previous release, comparing diffs for regressions, API changes, and version compatibility.

    by OpenAIOfficial publisher

    Contains scripts

    Claude Code, Codex, Cursor +3 · about 5.4K tokens · pinned 6 days ago

The skills listed here are third-party software, collected from public GitHub repositories as a catalog of recent skills and provided as is. Agent Native does not write or control them and is not responsible for any harm, loss or damage caused by installing or using them. Labels and checks describe what we saw at one commit and are not a guarantee. Read a skill's files before you run it. Terms of use