The Agent Skills Directory

[SAFE]: The skill is entirely instructional, providing patterns for using the @spaces.GPU decorator and managing ML workloads. There are no indications of malicious intent or hidden code.\n- [EXTERNAL_DOWNLOADS]: The skill refers to trusted sources for documentation and pre-built binaries, including official Hugging Face and PyTorch resources, which is standard for specialized ML hardware environments.\n- [COMMAND_EXECUTION]: Includes helper commands for dependency resolution and metadata inspection that use established tools (uv, curl) and target whitelisted domains like PyPI. These examples are transparent and for developer utility.\n- [DATA_EXFILTRATION]: No mechanisms for sensitive data access or unauthorized exfiltration were identified. All network-related instructions pertain to legitimate package management and documentation access.

huggingface-zerogpu