minimax-h3

Pass

Audited by Gen Agent Trust Hub on Aug 17, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADS
Full Analysis
  • [SAFE]: The skill serves as an educational and technical reference for configuring the MiniMax H3 model. It provides structured guidance on prompt formatting, including the use of integrated multimodal descriptions and specific dialogue tags like <d> for quality control. The content is strictly instructional and contains no obfuscation, credential harvesting, or malicious command patterns.
  • [EXTERNAL_DOWNLOADS]: The documentation identifies several external resources for downloading model weights and custom nodes from GitHub and HuggingFace. These sources include the official Comfy-Org repositories and recognized community contributors (e.g., kijai, larryvrh, and drbaph). These references are informational, targeting established platforms for AI model distribution, and do not involve automated script execution.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 17, 2026, 07:37 AM
Security Audit — agent-trust-hub — minimax-h3