physical-ai-defect-image-generation

Pass

Audited by Gen Agent Trust Hub on Jun 24, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides comprehensive workflows for physical AI defect generation on NVIDIA OSMO. All identified network operations and external downloads are directed toward legitimate and well-known services.
  • External Dependencies: The skill downloads standard utilities like yq from official GitHub repositories (e.g., github.com/mikefarah/yq) at runtime. These downloads are documented and necessary for the workflow configuration, originating from well-known sources.
  • Credential Management: The skill utilizes standard practices for handling sensitive tokens, instructing users to provide credentials via environment variables (HF_TOKEN) or the osmo CLI. This avoids hardcoded secrets and relies on the platform's native secret management.
  • Command Execution: Shell commands and scripts are used for environment configuration (e.g., export PATH), dependency staging, and workflow submission. These operations are consistent with the skill's stated purpose of managing cluster-based AI tasks and do not show signs of privilege escalation or unauthorized access.
  • Data Handling: Data setup workflows fetch datasets from established sources such as Hugging Face and public GitHub repositories. The instructions include guidance on license compliance for external datasets (e.g., Roboflow), demonstrating a security-conscious approach to data acquisition.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 24, 2026, 05:54 PM
Security Audit — agent-trust-hub — physical-ai-defect-image-generation