How does this skill handle long prompts that exceed token limits?

It provides specific instructions for using the BREAK token to manage CLIP's 77-token chunks, ensuring that long descriptions aren't silently truncated.

What prompt weight syntax does this skill support?

It supports standard ComfyUI weighting including parentheses for emphasis (e.g., '(word:1.2)'), brackets for de-emphasis, and multiplicative nesting rules.

Does it provide specific advice for Flux models?

Yes, it includes guidance for Flux.1 (Schnell and Dev), emphasizing natural language descriptions, CFG settings of 1.0, and the avoidance of negative prompts.

Can it help with LoRAs and Embeddings?

Absolutely. It covers the correct placement of trigger words for LoRAs and the specific syntax required for calling textual inversion embeddings like 'easynegative'.

ComfyUI Prompt Engineering

Name: ComfyUI Prompt Engineering
Author: artokun

byartokun

•

データサイエンスとML

Optimizes ComfyUI image generation with advanced prompt engineering syntax, model-specific strategies, and CLIP encoding best practices.

This skill equips Claude with deep technical knowledge of ComfyUI's prompting mechanics, ranging from basic CLIP text encoding to advanced weight modifiers and model-specific nuances for SDXL, Flux, and SD3. It enables Claude to assist users in crafting high-quality prompts, troubleshooting common artifacts, managing LoRA triggers, and utilizing advanced features like the BREAK token or textual inversions. By providing domain-specific guidance on token limits and conditioning tensors, it ensures users achieve precise and aesthetically superior visual outputs within their ComfyUI workflows.

主な機能

01Detailed implementation of embeddings and LoRA trigger words

02Best practices for prompt structuring and quality-boosting modifiers

03Comprehensive CLIP text encoding and emphasis weight syntax guidance

04Expert handling of BREAK tokens and the 77-token chunk limit

0511 GitHub stars

06Model-specific strategies for SD 1.5, SDXL, Flux, and SD3.5

ユースケース

01Translating natural language concepts into optimized ComfyUI prompt syntax

02Troubleshooting image artifacts caused by improper weighting or token truncation

03Configuring complex conditioning for multi-model or regional prompting workflows

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add artokun/comfyui-mcp prompt-engineering

For use in Claude.ai and ChatGPT

Download Skill