tao-train-image-classification
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: Extensive analysis across all threat categories, including prompt injection, data exfiltration, and persistence mechanisms, revealed no malicious patterns or vulnerabilities.
- [EXTERNAL_DOWNLOADS]: The skill references container images hosted on nvcr.io (NVIDIA Container Registry). These are official, well-known resources necessary for the machine learning workflows described in the skill.
- [COMMAND_EXECUTION]: The skill provides instructions for running specific TAO Toolkit actions (such as train, evaluate, and export) within Docker containers. These commands are strictly relevant to the skill's primary purpose and use standard configuration-driven execution patterns.
Audit Metadata