pytorch-distributed

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION] (SAFE): The skill includes a Node.js script that launches a Python training process locally. This is standard functionality for distributed training tools and uses safe process spawning techniques without a shell.
  • [EXTERNAL_DOWNLOADS] (SAFE): Referenced URLs are restricted to official PyTorch documentation (pytorch.org), which is a trusted source.
  • [PROMPT_INJECTION] (SAFE): No malicious instructions or bypass attempts were detected. The scripts use localized parameters that do not present a significant risk of indirect injection.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:46 PM