ComfyUI-PromptLora

ComfyUI-PromptLora

PromptLora is a ComfyUI node that combines a text prompt editor, a LoRA browser, and CLIP text encoding in one place. Select LoRAs from a visual library instead of adding a separate loader node for each one.

Features

• Browse LoRAs with image previews, folder filters, search, and favorites.
• Insert or remove LoRAs directly in the prompt, using tags such as <lora:example:0.8>.
• Automatically add available trigger words when selecting a LoRA.
• Adjust LoRA weights from the node without reopening the manager.
• View available LoRA metadata and edit activation text, preferred weight, and notes.
• Apply selected LoRAs to the model and CLIP, then produce positive conditioning.

Installation

1. Download the repository ZIP and extract it to ComfyUI/custom_nodes/ComfyUI-PromptLora/, or clone the repository into ComfyUI/custom_nodes/.
2. Restart ComfyUI and refresh your browser.
3. Find PromptLora in the conditioning category.

No additional Python packages are required beyond a working ComfyUI installation.

How to use

Connect the nodes as follows:

Checkpoint Loader (MODEL) ──> PromptLora (MODEL) ──> KSampler (model)
Checkpoint Loader (CLIP)  ──> PromptLora (CLIP)
                             PromptLora (CONDITIONING) ──> KSampler (positive)

Enter your positive prompt in PromptLora. Open the LoRA Manager from the node to choose LoRAs. Clicking a LoRA adds its tag and any available trigger words to the prompt; clicking it again removes it. You can also type tags manually, for example:

portrait, <lora:example:0.8>, detailed lighting

Negative prompts: use a separate CLIP Text Encode node connected to the KSampler's negative input. PromptLora supplies positive conditioning only.

LoRA previews and metadata

Preview images are read from files next to each .safetensors LoRA, with matching names (for example example.preview.png, example.png, or example.webp).

When present, metadata may be read from the LoRA's safetensors header and nearby .json, .metadata.json, .civitai.info, or .sq_lora.json files. These additional files are optional. Saving edits in the LoRA information window writes a matching .json file next to the LoRA, so that folder must be writable.

Favorites are stored in the browser and do not change the LoRA model files.

Notes

• The node type is PromptLora so existing workflows using that type remain compatible.
• LoRA tags must refer to LoRA files available in ComfyUI's configured LoRA directories.
• No Civitai extension or other LoRA manager is required.