omniverse-cad-to-simready

omniverse-cad-to-simready is an orchestrator skill from NVIDIA Omniverse that coordinates the end-to-end pipeline from a source CAD or other asset to a simulation-ready (SimReady) USD asset or package. It solves the problem of stitching together the many discrete stages required to make an asset usable in physics simulation: conversion to USD, material and physics assignment, SimReady conformance, validation, optional rendering, and packaging. Rather than acting as a single monolithic runner, it is documentation-driven and routes work through nested stage references, invoking installed stage scripts directly when deterministic execution is needed.

The workflow centers on a preflight step that installs or verifies local upstream checkouts, writes a `cad-to-simready-preflight.json` manifest, and exports `PHYSICAL_AI_PREFLIGHT_MANIFEST` and `PHYSICAL_AI_REQUIRE_PREFLIGHT=1` for downstream references. A `property_assignment_intent` value (run, or skip for conversion-only) drives routing; when set to run, the first operational action is to verify or deploy Content Agents (OVRTX renderer plus Material and Physics service containers, and optionally the Texture Agent) exported through `CONTENT_AGENTS_*_BASE_URL`. Ordered stages include identify-asset-context, convert-to-usd, validate-usd-minimum, Content Agents material/physics/texture assignment, simready-conform-profile with FET repairs, validation gates (omni-asset-validate, omni-asset-validate-geometry, omni-asset-validate-physics, simready-validate), ovrtx-render-service, and packaging via assemble-package-source, nv-core-package-sample, and nv-core-package-sample-validation. Conversion-only is a valid minimal scope. NVIDIA-backed source conversion requires an installed usd-convert-cad checkout; if missing, the skill preserves the blocked report and its install_hint rather than substituting another converter.

It targets engineers building physical-AI and robotics simulation content on Linux or macOS. Requirements include Python 3.12 and uv, and for managed Content Agents deployment an NVIDIA_API_KEY from build.nvidia.com, Docker with the NVIDIA Container Toolkit, and an NVIDIA GPU. Already-running endpoints may instead use explicit endpoint variables plus usage tokens such as NGC_API_KEY, NVCF_API_KEY, or CONTENT_AGENTS_*_TOKEN. If required authentication is missing or services cannot be made healthy, the skill blocks and reports readiness instead of proceeding.