Metadata-Version: 2.1
Name: comfyui-moss-toolkit
Version: 0.0.1
Summary: MOSS-Image toolkit: a one-node DiT + text encoder (bf16, INT8 W8A8 or fp8) + VAE loader, image-to-image (a text encode that puts the page through Qwen3.5's vision tower, and a MODEL patch that loads the i2i LoRA and feeds the page as a clean VAE reference latent on RoPE's reserved r axis, isolated for colorize v3), an Anima caption -> MOSS layout-JSON prompt node, and a caption helper with the prompt each i2i LoRA trained on (colorize with the page's OCR lines, sanitize's fixed caption). Works under KSampler and the MOSS Spectrum KSampler. Needs the moss-image node.
Home-page: https://github.com/sorryhyun/ComfyUI-MOSS-Toolkit
Requires-Python: >=3.10
