Concept Image Splitter is an agent skill by Tommy Geoco built from Dive Radio E4, where he showed how he built Backyard Designers from AI-generated concepts. It breaks a concept image into pieces you can build without shipping one giant picture.
What it does
- Lists every visible element, back to front, and flags extras the image model added.
- Sorts each element into tile it, code it in CSS, extract it, make it a component, live text, or cut.
- Writes the follow-up prompts to strip content and pull clean assets, with a chroma key fallback.
- Drafts a build brief for your coding agent with the asset list and each treatment.
- Lists speed checks like size variants, modern formats, and lazy loading, without guessing scores.
When to use it
Use it when you have a UI concept from GPT Image, Midjourney, or a flat comp and want a real, fast page built from it.
How to get it
The skill is free on GitHub, in Tommy Geoco's Dive Radio skills repo. Install the whole library with npx skills add tommygeoco/dive-radio-skills. The same skills also live in the Dive Radio Skills Library at dive.radio/skills.
more skills
Concept Image Splitter is an agent skill by Tommy Geoco built from Dive Radio E4, where he showed how he built Backyard Designers from AI-generated concepts. It breaks a concept image into pieces you can build without shipping one giant picture.
What it does
- Lists every visible element, back to front, and flags extras the image model added.
- Sorts each element into tile it, code it in CSS, extract it, make it a component, live text, or cut.
- Writes the follow-up prompts to strip content and pull clean assets, with a chroma key fallback.
- Drafts a build brief for your coding agent with the asset list and each treatment.
- Lists speed checks like size variants, modern formats, and lazy loading, without guessing scores.
When to use it
Use it when you have a UI concept from GPT Image, Midjourney, or a flat comp and want a real, fast page built from it.
How to get it
The skill is free on GitHub, in Tommy Geoco's Dive Radio skills repo. Install the whole library with npx skills add tommygeoco/dive-radio-skills. The same skills also live in the Dive Radio Skills Library at dive.radio/skills.
more skills
Concept Image Splitter is an agent skill by Tommy Geoco built from Dive Radio E4, where he showed how he built Backyard Designers from AI-generated concepts. It breaks a concept image into pieces you can build without shipping one giant picture.
What it does
- Lists every visible element, back to front, and flags extras the image model added.
- Sorts each element into tile it, code it in CSS, extract it, make it a component, live text, or cut.
- Writes the follow-up prompts to strip content and pull clean assets, with a chroma key fallback.
- Drafts a build brief for your coding agent with the asset list and each treatment.
- Lists speed checks like size variants, modern formats, and lazy loading, without guessing scores.
When to use it
Use it when you have a UI concept from GPT Image, Midjourney, or a flat comp and want a real, fast page built from it.
How to get it
The skill is free on GitHub, in Tommy Geoco's Dive Radio skills repo. Install the whole library with npx skills add tommygeoco/dive-radio-skills. The same skills also live in the Dive Radio Skills Library at dive.radio/skills.

