WebUI Bridge Positive Prompt

The WebUI Prompt Bridge main node is the full workspace; this is the compact version of its positive prompt box. It looks like a plain text input, but it has one trick the standard ComfyUI text field can't do: if you connect model and clip, it will actually load your LoRAs.

That's the feature that makes it worth a node of its own. Write <lora:my-style:0.75> in the prompt, and instead of the tag sitting there as inert text (or worse, being fed to the text encoder and doing nothing), the node resolves the file, applies it to the model and clip chain, and strips the tag from the text that reaches the encoder. It's the same LoRA machinery as the main node, in a small footprint for people who want the prompt flow visible on the canvas.

How it works

The mechanism is worth understanding because it explains the optional inputs. It parses <lora:name:weight> (and <lyco:...> LyCORIS tags) out of your prompt. Then:

• If model and clip are connected, it looks up each LoRA in your models/loras folder, loads it with ComfyUI's own load_lora_for_models, applies it to the chain, and returns the cleaned text plus the LoRA-augmented model and clip. Missing LoRAs raise an error rather than silently continuing.
• If they're not connected, it still detects the tags but can't load anything - it returns your text unchanged and sets lora_info to a warning that LoRAs were found but model/clip weren't fully wired.

It also watches for upstream LoRA loading and skips duplicates, so you don't double-apply a LoRA that an earlier node already loaded.

Inputs and outputs

• positive_prompt (required) - your prompt, tags and LoRA tags included.
• model, clip (optional) - connect these to enable real LoRA loading.
• Outputs: positive_text (cleaned prompt with LoRA tags removed), model and clip (the LoRA-applied chain), and lora_info - a status string telling you exactly what was applied, skipped, or missing.

How to install

Part of the ComfyUI-WebUI-Prompt-Bridge pack. ComfyUI-Manager → ComfyUI WebUI Prompt Bridge, or clone into custom_nodes and restart:

cd ComfyUI/custom_nodes
git clone https://github.com/dianfangsihuo/ComfyUI-WebUI-Prompt-Bridge.git

No models beyond the LoRAs you already have.

Common issues

The #1 gotcha is wiring. If you connect this node but keep feeding the original model into your sampler - which is easy to do by habit - the model output's applied LoRAs never reach generation, and lora_info can't tell you because your sampler simply isn't downstream of the node. Second: the lora_info string isn't just decoration, it's your debugging readout - "Applied: ..." means it loaded, "Missing LoRA(s)" names the file you typo'd, "Skipped upstream" means another node already has it. And a subtle one: the default fail_on_missing_lora behavior in this pack means a missing LoRA errors, which feels aggressive until you've wasted a render on a silently-ignored LoRA - then it feels like the only sane default.