tao-run-automl-deft-pipeline
Pass
Audited by Gen Agent Trust Hub on Jun 23, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill coordinates a complex, multi-phase machine learning workflow using official NVIDIA TAO tools and procedures. It uses delegated calls to other trusted skills in the TAO ecosystem.
- [SAFE]: External resource references, such as the NVIDIA Container Registry (nvcr.io), are restricted to official vendor-owned domains and are necessary for the training pipeline's execution environment.
- [SAFE]: The skill uses Python and shell commands for routine state management tasks, such as deep-merging hyperparameter specifications and updating local JSON state files between training phases.
- [SAFE]: The consolidated pre-flight check is a security-conscious design choice that ensures the user can review and approve all proposed operations and configuration across the entire multi-hour pipeline before any side-effecting actions occur.
Audit Metadata