tao-train-segformer

Pass

Audited by Gen Agent Trust Hub on Jun 30, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill does not exhibit any malicious behavior or safety violations. It is a legitimate wrapper for the NVIDIA TAO Toolkit.
  • [COMMAND_EXECUTION]: The skill specifies commands to be executed within Docker containers, such as segformer train and segformer export. These are documented vendor-provided tools for model lifecycle management.
  • [DATA_EXPOSURE]: While the skill templates include an encryption_key field with the value tlt_encode, this is a publicly documented default for the TAO Toolkit and is not a private credential.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 30, 2026, 05:48 PM
Security Audit — agent-trust-hub — tao-train-segformer