skills/nvidia/skills/nv-segment-ctmr/Gen Agent Trust Hub

nv-segment-ctmr

Pass

Audited by Gen Agent Trust Hub on Aug 31, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The script scripts/run_ctmr.py executes system commands via subprocess.run to interact with Git and the MONAI bundle CLI. These calls are part of the core functionality to track versions and perform segmentation inference.\n- [EXTERNAL_DOWNLOADS]: The skill fetches resources from the vendor's official GitHub repository (NVIDIA-Medtech/NV-Segment-CTMR) and Hugging Face model repository (nvidia/NV-Segment-CTMR). These are recognized vendor assets and well-known services.\n- [INDIRECT_PROMPT_INJECTION]: The skill has a data ingestion surface through processing medical imaging volumes and external configuration files.\n
  • Ingestion points: NIfTI image volumes and JSON metadata files are read in scripts/run_ctmr.py.\n
  • Boundary markers: Strict argument validation is present in the wrapper script, though data content is not explicitly sanitized for text instructions.\n
  • Capability inventory: The skill is capable of subprocess execution, file writes to specified output directories, and network access for repository cloning.\n
  • Sanitization: Input modality and file paths are validated before execution.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 31, 2026, 04:37 PM
Security Audit — agent-trust-hub — nv-segment-ctmr