tao-train-centerpose
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill utilizes official NVIDIA container images from the well-known registry
nvcr.io, which is consistent with the vendor's own infrastructure. - [SAFE]: All documented operations involve standard TAO Toolkit workflows, such as model training, evaluation, and TensorRT engine generation via the
centerposeCLI inside isolated Docker containers. - [SAFE]: The skill includes a digital signature bundle (
skill.oms.sig) to ensure the integrity and authenticity of the skill files. - [SAFE]: No patterns of data exfiltration, credential theft, prompt injection, or malicious persistence were identified during the analysis.
Audit Metadata