Simon Willison Codes a Game With One Prompt to Claude
A blogger proved that 'vibe coding' can produce a surprisingly polished Three.js game in just one conversation.

An AI program that builds complete web applications from plain-language descriptions, generating code, graphics, sound, and interactivity automatically.
Prompting a Game Into Existence
Simon Willison turned to Claude Opus 5.5 last week with a deceptively simple request: build a Three.js game called Tree Appreciator where players compare two algorithmically generated trees and appreciate the differences between them.
What happened next was less engineering and more alchemy. The model spent 3 hours and 7 minutes processing before delivering a complete, playable game complete with music, animation, and achievements.
Willison — known for his experiments at the intersection of AI and creative tools — admitted to finding the result \"somewhat embarrassingly\" good given his own recent talks about the unpredictability of AI generation. The game, he noted, is \"quite compelling to play.\"
The Vibe Coding Philosophy
The project falls squarely into what Willison has been calling \"vibe coding\" — the practice of describing desired outcomes in natural language and letting an AI model figure out the implementation details.
\"I've found that telling Opus 5.5 'make sure it is fun' works uncomfortably well,\" Willison wrote on his blog on October 14th. The implication is clear: when you give an AI model creative direction rather than technical specifications, it sometimes produces something that feels genuinely delightful.
The resulting game asks players to spot differences between two procedurally generated trees. While the concept is straightforward, the execution — with its polished audio, smooth animations, and achievement system — demonstrates what happens when powerful language models take creative direction seriously.
Vibe coding turns a text description into a finished game
- Claude Opus 5.5AI model · wrote all code and assets
- Three.jsruns 3D graphics in the browser
- Tree Appreciatorthe game made · compare two trees
- Procedural generationtrees created by rules, not drawn by hand
- Simon Willisonblogger · gave the prompt
From Simon Willison’s October 14th blog post; terms defined in story
What This Means for Creators
The experiment raises interesting questions about the future of game development. If a single prompt can produce a finished, animated game with sound and interactivity, does the barrier to entry for creating interactive experiences continue to shrink?
Willison's work suggests the answer is yes — but with caveats. The process took over three hours of computation, and the result, while impressive, required him to guide it with natural language rather than write traditional code.
For now, the project stands as a proof of concept: AI models can take creative briefs and deliver functional, even charming, interactive experiences. Whether this represents the future of game design or merely a clever party trick remains to be seen.
Key facts
- Claude Opus 5.5 spent 3 hours and 7 minutes generating the game
- The game features procedurally generated trees with music and animations
- Simon Willison used the iPhone app's chat mode, not Claude Code
- Achievements and spot-the-difference mechanics are included
- The project demonstrates 'vibe coding' — creating through natural language prompts
Sources
Every fact in this story comes from the source above.
If this story concerns you or your company, or uses something that belongs to you, ask for a correction or for it to be taken down. How requests work
Keep reading
The weekThis Week in AI New Models Flood In and Agents Get SaferHerald of AI
This Week in AI New Models Flood In and Agents Get Safer
Big open-weight releases, speed breakthroughs, and a push to make AI agents more reliable and secure.
Business and rulesMirendil raises $1 billion to use AI agents for faster model developmentHerald of AI
Mirendil raises $1 billion to use AI agents for faster model development
A new startup founded by former Google DeepMind and Anthropic researchers says its automated platform lets small teams build large AI models quickly and cheaply.
Apps and productsJ.Crew launches an AI stylist powered by Anthropic's Claude modelsHerald of AI
J.Crew launches an AI stylist powered by Anthropic's Claude models
Shoppers in the US can now describe an occasion and get a full outfit from J.Crew — all using Claude instead of hunting for each piece one by one.