Hi everyone,
We just published a set of three tutorials on building CesiumJS applications with AI coding agents, and I wanted to share them here and get your feedback.
All three are agent-agnostic. The setup steps cover Claude Code, OpenAI Codex, GitHub Copilot, and Cursor, and they focus on the prompts and iteration patterns that get an agent producing solid CesiumJS.
Build a CesiumJS App with AI: Start here. This walks you through setting up your agent, scaffolding a Vite + React + TypeScript project with vite-plugin-cesium, and getting your first globe viewer running. It also covers installing our CesiumJS Agent Skills (more on those below).
Build a Philadelphia Landmark Tour with CesiumJS Using AI: Build a 3D flyover tour across six Philadelphia landmarks using Google Photorealistic 3D Tiles, with a UI overlay for navigation and smooth camera flights between locations. There’s also a section on customizing it further (keyboard navigation, auto-play, restyling the UI) and an optional deploy step.
Build a Flight Simulator with CesiumJS Using AI: Build a simple flight sim with a tracking third-person camera, keyboard throttle/pitch/roll controls, and a minimap with a flight trail.
One piece worth calling out on its own: the CesiumJS Agent Skills repo. It’s a set of curated domain skills covering the CesiumJS API that give your agent accurate, up-to-date knowledge of our patterns and best practices, so it writes noticeably better code. The skills follow the open Agent Skills standard, so they work whether you’re in Claude Code or any other harness. Even if you don’t follow the tutorials, these are useful to drop into any CesiumJS project where you’re working with an agent.
We’d love your feedback. If you build something with these, share it in this thread, and let us know what you’d like to see next. More app types? Different data sources? Deeper coverage of specific APIs in the skills? We’re listening.
Happy building,
Austin