A solid wrapper around OpenAI's gpt-image-2 that gets you generating, editing, and inpainting images from the command line or inside Claude Code. Ships with 162 curated prompts across research figures, UI mockups, typography, and product shots, which is genuinely helpful if you're staring at a blank prompt. The CLI is fast (uvx run without installing), handles multi-reference edits and mask-based inpainting, and has sensible quality presets so you can draft cheap and render final at high. Python SDK examples are clear. If you're prototyping interfaces, generating diagram placeholders, or need repeatable image workflows in agent tasks, this is more practical than copying curl commands every time.
npx skills add https://github.com/aradotso/trending-skills --skill gpt-image-2-skill