computer-vision

Pass

Audited by Gen Agent Trust Hub on Aug 10, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to perform a directory listing (ls) on specific local paths (e.g., ~/.claude/skills/) to detect which companion skills are already installed. This check is used to recommend additional relevant tools to the user based on the skill's recommendations.
  • [EXTERNAL_DOWNLOADS]: The documentation and code samples reference downloading pretrained model weights and libraries from established, well-known services and organizations including PyTorch, Ultralytics, and Hugging Face.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 10, 2026, 02:05 PM
Security Audit — agent-trust-hub — computer-vision