modal-gpu-experiment
Pass
Audited by Gen Agent Trust Hub on Apr 19, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [SAFE]: A comprehensive review of the instructions and code examples found no evidence of malicious behavior, prompt injection, or data exfiltration. The skill follows established platform conventions for secure resource management.
- [EXTERNAL_DOWNLOADS]: The skill references standard machine learning packages including
torch,transformers,datasets,accelerate, andwandb. These are well-known industry libraries installed from official registries. References to external documentation and repositories target the vendor's own domains and the official Hugging Face hub, which are considered trusted and well-known services. - [COMMAND_EXECUTION]: The skill includes examples using
subprocess.runto execute training scripts within containerized environments. This behavior is the intended and documented way to manage execution on the Modal platform and does not pose a risk in this context.
Audit Metadata