tao-train-image-classification

Pass

Audited by Gen Agent Trust Hub on Jun 23, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONCREDENTIALS_UNSAFE
Full Analysis
  • [SAFE]: No malicious patterns such as prompt injection, data exfiltration, or persistence mechanisms were detected. The skill uses signed metadata and adheres to standard machine learning development practices.
  • [EXTERNAL_DOWNLOADS]: The skill references official NVIDIA container images (e.g., nvcr.io/nvidia/tao/tao-toolkit) and uses Weights & Biases (wandb) for experiment tracking. These are trusted sources and legitimate integrations for ML workflows.
  • [COMMAND_EXECUTION]: Shell commands for docker run and TAO toolkit actions are documented. These commands are structured for the specific purpose of model training and are consistent with the author's intent.
  • [CREDENTIALS_UNSAFE]: Configuration files contain an encryption_key parameter. This is used for model security and was found to contain only empty string placeholders, presenting no credential leakage risk.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 23, 2026, 12:10 PM
Security Audit — agent-trust-hub — tao-train-image-classification