I built a prompt engineering game — here's why I think learning by doing beats every tutorial

A developer built a prompt engineering game called The Prompt Game, featuring 10 challenges that require users to apply techniques like zero-shot, few-shot, chain of thought, and role prompting. The game uses a real LLM to respond and a second LLM to evaluate whether the correct technique was used, aiming to teach through hands-on practice rather than theory. Built with Next.js 14, Groq, Supabase, Upstash Redis, and Vercel, it is available for free at thepromptgame.vercel.app.

Every guide on prompt engineering teaches you theory. Read this, memorize that, here are 10 tips. The problem? You never actually write prompts. So I built a game: 10 challenges, each one forces you to use a specific technique — zero-shot, few-shot, chain of thought, role prompting, constraints, and more. You write a prompt → real LLM responds → second LLM evaluates whether you used the right technique. Not just whether the output looks okay — whether your prompt caused it correctly. Stack : Next.js 14, Groq, Supabase, Upstash Redis, Vercel Try it free: thepromptgame.vercel.app https://dev.tourl What technique do you think is hardest to learn hands-on?