tilegym-cutile-python

Pass

Audited by Gen Agent Trust Hub on Jun 12, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No malicious patterns or security vulnerabilities were detected in the skill instructions or provided scripts.
  • [EXTERNAL_DOWNLOADS]: The skill fetches configuration and source code from official repositories at runtime to support its core functionality. Evidence: It clones official repositories from github.com/NVIDIA/TileGym and github.com/pytorch/pytorch to provide context and reference implementations.
  • [COMMAND_EXECUTION]: The skill includes an automated validation loop that involves executing generated Python code to verify kernel correctness. Evidence: The SKILL.md instructions specify writing generated code to a file and running it via python .py to check for compilation and numerical errors.
  • [SAFE]: Static analysis flags for eval() are verified as false positives. Evidence: The detected eval() calls in the example scripts (e.g., examples/normalization/group_norm.py) are calls to the standard PyTorch nn.Module.eval() method, which is used for setting models to evaluation mode and does not present a security risk.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 12, 2026, 11:36 PM
Security Audit — agent-trust-hub — tilegym-cutile-python