This is a router skill that picks the right image model from RunComfy's catalog based on what you're actually trying to do. It knows when to use GPT Image 2 for typography, Seedream 5 for photoreal portraits, Klein 4B for sub-second iteration, or any of the other 11+ models it supports. The skill ships prompting patterns and exact CLI invocations for both text-to-image and image-to-image workflows. Useful if you're already on RunComfy and tired of manually picking between FLUX variants, Nano Banana tiers, ByteDance models, and open-weights options. It's opinionated about defaults (Klein 9B) and avoids (don't use Klein for text, don't use Klein 4B for finals), which saves you from reading model cards every time.
npx skills add https://github.com/agentspace-so/runcomfy-agent-skills --skill ai-image-generation