ovphysx-usd-authoring
Pass
Audited by Gen Agent Trust Hub on Sep 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill involves authoring and potentially reading USD files (.usda), which are complex data formats. While the skill's instructions are safe, processing USD content from untrusted external sources could theoretically be used for indirect prompt injection via metadata or scene descriptions. This is a general risk associated with rich data formats handled by agents.
- [DYNAMIC_EXECUTION]: The skill provides Python snippets for registering PhysX-specific 'codeless schemas' using the Plug.Registry().RegisterPlugins() method. The code calls ovphysx.codeless_schema_paths() to retrieve plugin paths at runtime. This is the standard, documented procedure for extending the OpenUSD runtime with specialized vendor schemas and represents intended functionality within the NVIDIA Omniverse ecosystem.
Audit Metadata