skills/rimagination/h3lite/h3lite/Gen Agent Trust Hub

h3lite

Pass

Audited by Gen Agent Trust Hub on Aug 25, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the Python subprocess module across several scripts (h3_doctor.py, h3_status.py, h3_fastpath.py, h3_generate.py) to perform environment diagnostics and media verification. These operations are strictly limited to well-defined tasks:
  • Calling nvidia-smi to detect GPU capabilities and VRAM usage.
  • Invoking the local Python interpreter to probe for installed libraries like torch and comfy_kitchen.
  • Using ffmpeg and ffprobe to verify the metadata and quality of generated video files.
  • Executing internal helper scripts within the skill's own directory.
  • [EXTERNAL_DOWNLOADS]: The documentation provides guidance for downloading model weights and ComfyUI extensions from established platforms including GitHub (Comfy-Org), Hugging Face (MiniMaxAI), and ModelScope. It also mentions Baidu Netdisk for pre-configured component sets common in specific user communities. All downloads are user-initiated or guided by clear instructions for local environment setup.
  • [DATA_EXPOSURE]: Network activity is restricted to communicating with the local ComfyUI server at 127.0.0.1:8188. The scripts do not access sensitive system files (e.g., SSH keys, cloud credentials) or perform unauthorized exfiltration of user data.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 25, 2026, 11:40 AM
Security Audit — agent-trust-hub — h3lite