vision-analysis

Pass

Audited by Gen Agent Trust Hub on Jun 22, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill's primary function is to facilitate image analysis using the MiniMax_understand_image tool, which is a legitimate vision capability.
  • [SAFE]: Setup instructions include the use of uvx to run the minimax-coding-plan-mcp package. This is a standard and expected method for deploying MCP servers within an agent's environment.
  • [SAFE]: All external URLs and API hosts point to the official MiniMax platform (minimaxi.com), presenting no risk of phishing or unauthorized data redirection.
  • [SAFE]: The skill provides template prompts for various analysis modes (OCR, UI review, etc.) that guide the model's behavior without attempting to bypass safety filters or ignore system instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 22, 2026, 02:17 PM
Security Audit — agent-trust-hub — vision-analysis