Build a browser racing game from one prompt
A lone agent turned one prompt into a playable kart racer, then kept testing and fixing it while acting as designer, developer, and QA.
- What they did
- The user gave Codex a detailed request for a Three.js voxel kart racer, along with image-generation and web game development tools. Codex then built the game by itself, switching between design, coding, and testing roles as it played through the project. It was also sent a rotating set of generic follow-up prompts, including checks for missing pieces, feature requests, testing, and bug fixes.
- How it went
- The final game included multiple racers, eight tracks, and space-bar items. The process used more than 7 million tokens. The source does not say how good the finished game was, and earlier versions were kept so the changes can be reviewed.
- Tips
- Give the agent a clear game concept plus a requirement to self-test by playing. Before anything that changes files or removes code, ask it to pause and show the planned change for approval first.
Try it yourself with Codex
Build a browser-based [game genre] in [project folder] using [engine or framework], with [number] players or racers, [number] maps or levels, and [items, abilities, or power-ups]. First outline the main systems you will need. Then implement the game, play it yourself to test it, and fix any bugs, balance issues, or missing features you find. Before you delete files, overwrite major code paths, submit anything, or make any irreversible change, stop and ask for my approval. When you are done, give me a short QA summary, how to run it, and a list of what you changed.
Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.
3 more reports of this
Get 3 agent uses like this every morning
With the day's top AI news. Free, and one click to unsubscribe.
Related agent uses
Use Codex to shape game architecture and animation blending
A game builder had Codex handle core architecture and improve animation transitions by reading engine docs and wiring in merging logic.
Pit three agents against a browser shooter build
Three coding agents independently built the same multiplayer shooter, letting the team compare their strengths and weak spots.
Run multiple game builds in parallel
One user had four Grok 4.20 agents create different games at the same time and finish them within a day.
Use Gemini to generate a match-3 game from one prompt
A single Gemini prompt produced a working Candy Crush-style game and a share link.
Use an AI to referee a D&D 5e dungeon
They ran a browser D&D crawl with three characters while the agent checked rules text and monster stats from separate sources.
Use Cursor to build a small game collection fast
A data scientist turned a rough idea into a Python tkinter app with four minigames using ChatGPT and Cursor agent mode.