Interactive 3D Miniature City
A live 3D miniature of Manhattan in your browser: fly to eight landmarks, switch day and night, and orbit over the traffic, pedestrians and boats.
Bring a city and the landmarks it should feature. The Skill returns a downloadable Skill (meta prompt, lessons, verification) and a runnable reference scene.
How it works
The meta prompt goes to your coding agent together with any reference images.
The agent starts from the reference scene, replaces the street plan, the landmark models and the copy, and compares overview, landmark and night screenshots each round.
Verify headlessly and build one self-contained HTML file.
Frequently asked questions
Does it call any model or need a key?
No. The Skill gives your agent a prompt, lessons and a reference scene; the scene itself makes no network calls.
Why is the demo sandboxed?
The frame has an opaque origin and no network access, so the scene can’t read anything from this page or reach the internet.
Is the map accurate?
No. The street plan and landmark positions are artistically compressed to fit a tabletop; it is not a survey.
Can I make another city?
Yes. Describe the city and its landmarks in the slots; your agent replaces the street plan, the landmark models and the copy together, not just the labels.
Does it work on mobile?
Phones get the low tier automatically and a compact layout with tap targets, and the scene honours reduced motion. It has been verified under touch emulation, not on real devices.
Why is there no online generation?
Building a new scene needs a model key inside the sandbox, which the backend doesn’t support yet. Run the Skill with your own agent for now.