Computer Vision Helper

Fail

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: CRITICAL
Full Analysis
  • [EXTERNAL_DOWNLOADS] (SAFE): The skill utilizes legitimate ML libraries such as torch, clip, and ultralytics. Methods like clip.load() and YOLO() download pre-trained weights from trusted provider repositories.
  • [DATA_EXFILTRATION] (SAFE): Image data is processed locally or sent to the official OpenAI API as part of the intended gpt-4-vision-preview functionality. No sensitive file access or unauthorized network calls were found.
  • [COMMAND_EXECUTION] (SAFE): All code snippets are pure Python and do not attempt to execute shell commands or spawn sub-processes.
  • [SAFE] (SAFE): Investigation of the 'box.co' security alert confirms it is a false positive. The scanner likely flagged the code line conf = box.conf[0].item() due to the substring 'box.co'. There is no actual URL or network request directed to the box.co domain.
Recommendations
  • Contains 1 malicious URL(s) - DO NOT USE
Audit Metadata
Risk Level
CRITICAL
Analyzed
Feb 17, 2026, 05:19 PM