huggingface-zerogpu

Pass

Audited by Gen Agent Trust Hub on Jul 3, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill consists of documentation and reference implementations for the Hugging Face ZeroGPU platform. It provides developers with best practices for handling process isolation, CUDA availability, and hardware-specific constraints.
  • [EXTERNAL_DOWNLOADS]: The documentation references downloading pre-built binary wheels for performance-critical libraries like flash-attention from official GitHub releases and Hugging Face Hub. These are legitimate resources from well-known technology services required for the described ML workflows.
  • [COMMAND_EXECUTION]: The instructions include standard package management commands (e.g., pip, uv) and PyPI metadata queries. These are provided as developer tools for environment configuration and do not involve any malicious or hidden execution patterns.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 3, 2026, 01:07 PM
Security Audit — agent-trust-hub — huggingface-zerogpu