nv-segment-ct-finetune
Pass
Audited by Gen Agent Trust Hub on Aug 31, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill performs its intended medical imaging functions without security risk. All network activity is directed to trusted vendor repositories (NVIDIA on GitHub and Hugging Face) for necessary model components.
- [COMMAND_EXECUTION]: Shell commands are used for environment setup and running the MONAI training bundle. These are securely handled using sys.executable and shlex.quote, preventing command injection.
- [DYNAMIC_EXECUTION]: Runtime generation of configuration files is used to parameterize the training process, which is a standard feature of the MONAI framework and operates within the skill's local scope.
- [INDIRECT_PROMPT_INJECTION]: The skill validates user-supplied datasets and datalists through NIfTI header checks and anatomical audits, ensuring that processed data conforms to expected medical imaging formats.
Audit Metadata