2026-05-19 New AI Skills - Country, City, and Landmark Image Generators

Canonical version: 2026-05-19 New AI Skills - Country, City, and Landmark Image Generators.

A school teacher needed hero images for a school fair: countries, cities, landmarks. I spent an afternoon turning the prompts I converged on into two reusable AI skills and pushed them to GitHub so anyone can grab them.

Repo: https://github.com/dsebastien/countries-cities-landmark-images-skills

The two skills

Two skills, both MIT-licensed, both driving the same tiny Node wrapper around Google's Gemini 3 Pro Image (a.k.a. Nano Banana Pro):

• travel-country-poster — one blended cinematic golden-hour photo per country or city, with iconic landmarks naturally fused into a single frame and the national flag billowing pole-less across the upper-left sky, in an editorial travel-magazine style.
• world-landmark-globe — a hero image of planet Earth from low orbit, with 3D miniature landmarks rising from each continent at their real geographic locations.

The prompts took most of the time. Each skill is a SKILL.md per use case, plus a shared CLI script (scripts/nano-banana.mjs) and a folder of example outputs you can check before paying for any API call.

Why share this

1. The prompts went through dozens of iterations across countries with very different landmarks (Belgium, India, Japan, USA, France), and the results look consistent.
2. The skills are portable: copy the folder into any Claude Code skills directory to use them.
3. A teacher needed these images. Open-sourcing the skills costs me nothing and can save other teachers an afternoon.

How to use them

git clone https://github.com/dsebastien/countries-cities-landmark-images-skills.git
cd countries-cities-landmark-images-skills
export GOOGLE_BANANA_API_KEY="your-key-here"

node scripts/nano-banana.mjs generate "<prompt from the SKILL.md template>" \
  --aspect 16:9 --size 2K \
  --model gemini-3-pro-image-preview \
  --output out/paris.jpg

The CLI also supports edit (image-to-image with a reference), so you can iterate — "make the flag smaller, keep everything else" — without restarting from scratch.

Heads-up on cost. Gemini 3 Pro Image is paid, and each 2K image costs a few cents. Don't batch a hundred countries without checking the pricing page first.

Announced here

I posted about it on X: https://x.com/dSebastien/status/2056816834767470877

If you remix the prompts for your own country/city/theme, send the result; I'd love to see what comes out.

References

• Repo: https://github.com/dsebastien/countries-cities-landmark-images-skills
• Announcement: https://x.com/dSebastien/status/2056816834767470877

Related

• Claude Code
• Claude Code Skills
• Nano Banana Pro
• AI image generation models

About Sébastien

Ready to get to the next level?