A while back I built a Hangman demo on my website where two LLM agents play the game by guessing letters against each other.
This game actually started as a practice project for my AI-102 cert. A couple of months after passing, I rewrote it into a container using SignalR and built a frontend for it on my site to see it in action - already using an AI assistant to help, though back then it was chat-only, which made things a bit more complex.
I just took it a step further: using Claude Code, I extended the game so a human player can jump in and guess too. Instead of just watching two AI agents play, you can now play alongside them.
A quick peek under the hood
- The backend runs as a container in Azure Container Apps
- The frontend is served through a Cloudflare Worker
- The two LLM agents also run via Cloudflare
- The container is deployed with a GitHub Actions workflow
Want to give it a try yourself? Play a round and let me know what you think. 