The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external data, creating a potential surface for indirect prompt injection attacks.
Ingestion points: The stage_file parameter (file path) and the samples parameter (JSON array) across all tools (set_time_codes, author_xform_samples, author_attribute_samples) represent entry points for untrusted data.
Boundary markers: There are no explicit delimiters or specific instructions within the tools to distinguish between data and potentially embedded instructions.
Capability inventory: The skill possesses the capability to modify the filesystem by writing time-sampled data to USD stages via the pxr runtime.
Sanitization: The implementation relies on the structural validation provided by the pxr (OpenUSD) runtime and the dcc_mcp_openusd wrapper library.

openusd-animation