Comfyui LLM Prompt Enhancer — AI skill for Claude Code

ComfyUI LLM Prompt Enhancer

A ComfyUI node that rewrites your prompt with an LLM before it reaches the CLIP encoder. You type something short, pick a style, and the node hands your image model a fuller prompt to work with.

It takes a CLIP input and returns conditioning plus the enhanced prompt as text, so you can drop it straight into an existing workflow or chain it with other nodes.

What it does

• Five providers: OpenAI, Anthropic, Google, OpenRouter, and Ollama for local models
• Two output formats:
  • descriptive writes flowing sentences describing the image
  • tags writes comma separated SDXL and Danbooru style tags
• 47 enhancement styles across 9 categories, from photorealistic to ukiyo-e to vaporwave
• Model picker per provider, so you can trade cost against quality without editing code
• Runs fully local through Ollama if you would rather not send prompts to an API
• Falls back to your original prompt if the API call fails, so a bad key or a rate limit doesn't break the run
• Works with Flux and Stable Diffusion, including SDXL and custom fine-tunes

Example gallery

Two Cloudinary collections with more than fits here:

• Node usage examples shows the node wired into ComfyUI workflows
• Generated image examples shows output across different styles

Anime Style

Bauhaus Style

Charcoal Drawing