minimax: base64 video/audio reference loader nodes

minimax: base64 video/audio reference loader nodes

lora_from_url: log matched target modules per lora (bind diagnostic)

h3 nsfw lora mirror job

lora mirror job

minimax: pin huggingface_hub<1.0 (transformers compat)

minimax: comfy smoke test before handler start

minimax: ship boot log on ERR too

minimax: boot log shipping to HF for debuggability

minimax: install into /opt/venv python (comfy runtime env)

minimax: upgrade torch to cu130 for int8_convrot

minimax worker: lora_from_url.py

minimax worker: minimax_start.sh

Add minimax h3 mirror job script

handler: support image_format/webp/quality and guidance alias for flowcore client compat

Add runpod serverless worker handler (qwen-image-turbo + lora-by-url)