tao-train-nvpanoptix3d
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill defines a configuration framework for the NVIDIA TAO toolkit's NVPanoptix3D model. Analysis of the documentation, schemas, and metadata reveals no malicious patterns or security risks. The skill follows best practices for containerized machine learning workflows.\n- [EXTERNAL_DOWNLOADS]: The skill references official container images from NVIDIA's registry.\n
- Evidence:
nvcr.io/nvstaging/tao/tao-toolkit-pyt:7.2.0-rc-36-multiarchmentioned inSKILL.mdandreferences/skill_info.yaml.\n- [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it processes external, potentially untrusted data files.\n - Ingestion points: User-provided YAML specifications and dataset metadata files (e.g.,
meta/train.json,meta/val.json) referenced inreferences/skill_info.yaml.\n - Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the provided templates.\n
- Capability inventory: The skill executes shell commands via Docker to run Python modules (
nvidia_tao_pytorch.cv.nvpanoptix3d.entrypoint.nvpanoptix3d) which interact with the file system for reading data and writing results.\n - Sanitization: No specific sanitization or filtering logic is documented in the skill's instructions; it relies on the internal implementation of the TAO toolkit entrypoints.
Audit Metadata