tao-train-centerpose
Pass
Audited by Gen Agent Trust Hub on Jun 30, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides legitimate instructions for 6-DoF pose estimation model training and deployment using the NVIDIA TAO toolkit.
- [PROMPT_INJECTION]: No malicious prompt injection patterns were found; all instructions are consistent with the intended machine learning workflow.
- [CREDENTIALS_UNSAFE]: No hardcoded credentials were detected. Fields for encryption keys in configuration templates use safe placeholders like 'tlt_encode' or empty strings.
- [EXTERNAL_DOWNLOADS]: External container images are referenced from the official NVIDIA registry (nvcr.io), which is a well-known service from the vendor.
Audit Metadata