skills/nvidia/skills/ad-model-onboard/Gen Agent Trust Hub

ad-model-onboard

Pass

Audited by Gen Agent Trust Hub on May 18, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches model configuration and code from HuggingFace using the standard huggingface-cli. HuggingFace is a well-known service for AI model distribution, and the download is restricted to non-weight files (code and config) for onboarding purposes.
  • [COMMAND_EXECUTION]: The skill executes local system commands like nvidia-smi to check GPU memory and python -c to verify local library installations. It also runs project-specific scripts such as build_and_run_ad.py for end-to-end model validation, which is consistent with its stated developer purpose.
  • [SAFE]: The skill integrates with the GitHub CLI (gh) to manage pull requests and poll for reviewer feedback. This is a legitimate automation of developer workflows and uses official tooling.
Audit Metadata
Risk Level
SAFE
Analyzed
May 18, 2026, 06:14 AM
Security Audit — agent-trust-hub — ad-model-onboard