WeiLin Positive Prompt To AutoLoras

This is the pack's LoRA-loader node with the negative half cut off, and honestly, that's enough for a lot of workflows. It takes a model, a CLIP, and a positive prompt; scans the prompt for <lora:...> tags; loads and applies each LoRA; and hands you a patched model plus ready-made positive conditioning. No separate LoraLoader node, no separate CLIPTextEncode - one node, prompt in, conditioning out.

Why this variant exists

The two-in-one version (WeiLinComfyUIPromptToLoras) carries a negative box too, but here's the thing the pack never advertises: its negative side doesn't parse LoRA tags anyway. The negative gets encoded as plain text, tag or not. So if you keep your negative prompt elsewhere - another conditioning branch, a separate text encode - the only real difference between this node and the two-in-one is that you're not dragging an extra empty box around. That's exactly what this one is for.

How it works

Same engine as the full version. It regex-scans your positive for <lora:...> tags, strips them (plus trailing commas) out of the text, and for each one resolves the filename against your ComfyUI/models/loras folder, loads the file, and patches model and CLIP with ComfyUI's own LoRA machinery. Then it encodes the cleaned prompt into a CONDITIONING.

The strength syntax is the A1111-style form:

• <lora:filename.safetensors:0.3:0.4> - 0.3 model strength, 0.4 CLIP strength
• <lora:filename.safetensors:0.4> - both 0.4
• <lora:filename.safetensors> - both 1.0

Use the filename exactly as it appears in your loras folder, extension included. And remember LoRAs are architecture-bound - an SDXL LoRA won't do anything on a Flux or modern LLM-encoded checkpoint, no matter how neatly this node loads it.

Inputs and outputs

• model (MODEL) and clip (CLIP) - from your checkpoint loader.
• positive (multiline string).
• Outputs: model and 正向条件 CONDITIONING (Chinese output name, constant across languages).

Wire model and the conditioning into a KSampler alongside your negative conditioning from wherever you're keeping it.

Installing

ComfyUI Manager → search WeiLin-ComfyUI-prompt-all-in-one → install → restart, or:

cd ComfyUI/custom_nodes
git clone https://github.com/weilin9999/WeiLin-ComfyUI-prompt-all-in-one

First launch auto-installs gradio and ruamel.yaml - let the startup log settle.

The honest take

It's a thin variant of a niche idea, so set expectations: if you stack many LoRAs with careful per-entry control, rgthree's Power Lora Loader is the better tool - toggles, trigger words, CivitAI info - and this node's tag-parsing approach is really a comfort blanket for A1111 habits. Where it wins is convenience when you copy a LoRA prompt off a model page and just want it to run. Same caveat as the whole pack: it's deprecated, frozen at 3.6.9.1, with the author pointing to the incompatible successor WeiLin-Comfyui-Tools. Works today; not a foundation for the future.