vision

Pass

Audited by Gen Agent Trust Hub on Sep 2, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses Python's argparse and sys modules to handle CLI interactions. It executes vision analysis by calling a local Python script (vision.py) which processes images and communicates with external AI providers.
  • [DATA_EXPOSURE_AND_EXFILTRATION]: While the skill interacts with external AI APIs (OpenAI, Anthropic, Volcengine, DashScope), these are for the intended purpose of image analysis. It follows safe practices by reading credentials from environment variables (e.g., OPENAI_API_KEY) rather than hardcoding them.
  • [EXTERNAL_DOWNLOADS]: The skill optionally requires the anthropic Python package for specific provider support, which is a well-known package from a trusted service provider.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 2, 2026, 11:56 AM
Security Audit — agent-trust-hub — vision