mindspeed-mm-vlm

Pass

Audited by Gen Agent Trust Hub on May 19, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches core framework repositories and model-specific extensions from vendor-managed GitCode repositories and well-known sources like NVIDIA's GitHub and HuggingFace.
  • [COMMAND_EXECUTION]: The instructions direct the user to perform various shell operations, including repository cloning, package installation, and execution of training scripts via tools like torchrun and bash.
  • [EXTERNAL_DOWNLOADS]: Mentions the use of a local proxy (127.0.0.1:58232) for downloading models from HuggingFace, which is a standard configuration for users in restricted network environments typical of specialized hardware deployments.
Audit Metadata
Risk Level
SAFE
Analyzed
May 19, 2026, 06:27 AM
Security Audit — agent-trust-hub — mindspeed-mm-vlm