ComfyUI-LLM-Prompt-Tagger

This custom node package adds two nodes for OpenAI-compatible chat-completions APIs:

• LLM Prompt Tags Returns positive_prompt, negative_prompt, and the raw model response text.
• LLM Prompt Conditioning Calls the same API, then encodes the generated prompts into positive and negative conditioning with a CLIP input.

Why two outputs

In ComfyUI, positive and negative prompts are usually handled separately.

Returning two prompt strings is the most flexible design because you can:

• inspect or edit the prompts before encoding
• send them through other string-processing nodes
• connect each one to a separate CLIPTextEncode

The conditioning node is a convenience wrapper when you want a single-step workflow.

API expectations

The node calls an OpenAI-compatible chat/completions endpoint and asks the model to return JSON:

{
  "positive": ["tag 1", "tag 2"],
  "negative": ["tag a", "tag b"]
}

The node then prepends your fixed positive and negative prefixes and removes duplicates.