Build Games Challenge

Building Super Mario Games with AI 🍄

Nostalgic game devs, if you grew up jumping through pixelated worlds as Mario, you’re going to love this!


This is a modern reinterpretation of the classic Super Mario experience—reimagined in two formats:


one for the web, one for desktop. Both versions were crafted to deliver smooth animations, vibrant graphics, and that familiar, addictive gameplay. 🎮

I created both versions in less than an hour using the Amazon Q Developer CLI, exploring the power of AI to scaffold the game logic, generate the folder structure, and even handle bug fixes automatically.

The Original Plan (That Failed)

I initially asked AI to build a "professional Super Mario game with advanced features." The result?

• 2000+ lines of complicated code


• Confusing file structure


• Features nobody wanted


• A game that wasn't fun to play

Lesson learned: More code ≠ Better game

The AI-Powered Breakthrough

Instead of fighting complexity, I worked with AI to build TWO focused versions:

🌐 HTML Version - The Universal Champion

Perfect for sharing and playing anywhere!

What I asked AI:

"Create a smooth Super Mario game for browsers. Make it mobile-friendly, under 300 lines, with spinning coins and responsive controls."

What AI delivered:

• Instant browser play


• Works on phones and tablets
  


• Smooth animations


• Easy to share with friends

🎮 Pygame Version - The Desktop Powerhouse

Enhanced with professional game development techniques!

What I asked AI:

"Take the HTML version and create a Pygame version with professional animations like squash & stretch effects and enhanced visual polish."

What AI delivered:

• Squash & stretch animations (like Pixar movies!)


• Glowing coin effects


• Smooth physics simulation


• Desktop gaming experience

Game Screenshots - HTML Version

Game Menu

The Game

Winning Scenario

Game Over

Pygame Version

[Desktop app with enhanced animations]

Pygame

AI Development Magic

The AI surprised me with smart suggestions I hadn't thought of:

Automatic mobile support for HTML version:

// AI added touch controls without me asking!

canvas.addEventListener('touchstart', handleTouch);

Professional animation techniques for Pygame:

# AI implemented industry-standard squash & stretch

if mario.jumping:

    mario.scale_y = 0.8  # Squash when jumping

    mario.scale_x = 1.2  # Stretch horizontally

Cross-platform optimization - AI automatically optimized each version for its platform.

Effective AI Prompts I Used

✅ Good Prompts (That Worked)

"Create a Mario game for HTML5 that works on mobile browsers"

"Add professional squash & stretch animations to the Pygame version"  

"Make the coins spin in 3D with a golden glow effect"

❌ Bad Prompts (That Failed)

"Make a complex Mario game with advanced features"

"Add every possible animation and effect"

"Create the most professional game possible"

Key insight: Specific, focused requests work better than vague "make it awesome" prompts.

Time Saved with AI

• 
  HTML Version: 2 hours (vs 8+ hours manually)


• 
  Pygame Version: 3 hours (vs 12+ hours manually)
  


• 
  Bug fixes: AI caught and fixed issues instantly


• 
  Cross-platform adaptation: AI handled platform differences automatically

The Results

Running the Games

HTML Version (Recommended for sharing)

# Just open in any browser

open smooth-mario.html

Pygame Version (Recommended for desktop)

cd pygame-mario

python3 smooth_mario_pygame.py

Easy Launcher

python3 play.py  # Choose your version!

What I Learned

🤖 AI Development Tips

1. 
   Be specific - "mobile-friendly Mario game" works better than "good Mario game"


2. 
   Iterate gradually - Build features one at a time


3. 
   Ask for explanations - Understanding AI's choices helps you learn


4. 
   Platform matters - Different platforms need different approaches

🎮 Game Development Insights

1. 
   Simple can be better - Two focused games beat one complicated mess


2. 
   Platform optimization - HTML for sharing, Pygame for desktop experience


3. 
   User experience first - Fun gameplay > technical complexity


4. 
   Clean code wins - 300 readable lines > 2000 confusing lines

Try It Yourself!

🔗 Play the games: GitHub Repository

🔗 HTML Version: Play in Browser

Want to build your own AI-powered game? Start with a simple, specific prompt and iterate from there!

Credits

• Built using AI assistance for rapid development


• HTML version: Pure JavaScript, no frameworks


• Pygame version: Python with professional animation techniques


• Total development time: 5 hours (would have been 20+ hours manually)

🍄 Some more days until the deadline. Join “Build Games with Amazon Q CLI” to get hands-on agentic coding experience, build games, complete the steps and redeem a cool T-shirt. Get started from here → https://s12d.com/BGWQCLI”

What game will you build with AI next?