GenEngine

Live Demo

Inspiration

I've always found game development intimidating. The steep learning curves and complex setups felt like a barrier that made me scared to even start. I wanted to build a tool that makes game development fun, fast, and accessible for everyone, especially students and solo builders who might feel the same way. The goal was to go from a simple idea to a playable game in minutes, not days.

What it does

GenEngine is a live, AI-powered game development environment where you can create, edit, and play games using natural language. The workflow is simple and interactive:

Prompt in the Director Panel: On the left, you type a prompt like, "Create the game Pong."
See Instant Results: Google Gemini (gemini-2.5-flash) generates the complete JavaScript code in the Monaco editor on the right (the Blueprint Panel). Simultaneously, the game becomes playable in the central Live Canvas.
Iterate and Refine: The process is continuous. A follow-up prompt like, "make the ball go faster," will instantly update the code and the live game. An explanation of the changes made by the AI appears right above the code.
Ask AI: If you don't understand a piece of code, you can use the "Ask AI" feature to get a detailed explanation of its purpose and function.
Revert Changes: A "Reset to Last" button at the bottom acts like a simple version control, letting you safely undo a change that didn't work out.

Challenges I ran into

Deploying the project presented significant challenges, particularly in making the Google Gemini API work seamlessly with the Cloudflare Workers backend. Architecting the live editor was also a major hurdle, creating a sandboxed environment where generated code could execute safely and update the game in real-time without freezing the app required a lot of trial and error.

Accomplishments I’m proud of

Honestly, just getting a complex, full-stack application like this fully operational and deployed was a huge accomplishment. I'm especially proud of the clean and intuitive UI/UX, which makes the creative process feel natural. The best part is that the prompt-to-play loop is genuinely fun and addicting (I'll admit I spent a couple of hours just playing around with the games I was creating).

What I learned

This project was a deep dive into several key technologies. I learned a ton about the Cloudflare ecosystem (Workers, Pages, R2) and how to manage a serverless backend. I also gained practical experience with the Google Gemini API, from crafting effective prompts to handling its output. Finally, I learned the intricacies of building an environment for live code execution and displaying it on screen.

What’s next for GenEngine

Accounts & Authentication: Allowing users to save and manage their projects.
AI Image Generation: Adding the ability to generate and incorporate art assets.
Collaboration: A real-time, multi-user mode for building games with friends.
Code Refinements: Improving the robustness and quality of the AI-generated code.