nv-reason-cxr
Pass
Audited by Gen Agent Trust Hub on Jul 8, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns or security risks were identified. The skill correctly implements security best practices, such as providing explicit medical disclaimers and limitations.
- [EXTERNAL_DOWNLOADS]: The skill connects to Hugging Face (
huggingface.co,*.hf.space) to download model weights or perform remote API inference. These are standard operations for an AI model wrapper and target well-known, trusted services. - [COMMAND_EXECUTION]: Static analysis flagged the use of
subprocess.run()intests/test_nv_reason_cxr.py; however, this is used exclusively for unit testing the runner script and does not pose a runtime risk to users. - [DYNAMIC_EXECUTION]: Static analysis flagged
eval()and__import__()inscripts/run_nv_reason_cxr.py. These are false positives or benign in context: - The
.eval()calls are PyTorch model methods used to switch the neural network into evaluation mode. - The
__import__()call is used within a setup utility to check if required packages are properly installed before running inference. - [REMOTE_CODE_EXECUTION]: The skill manages dependencies using standard package managers (pip) targeting official registries. No unverified or suspicious remote script execution patterns were found.
Audit Metadata