Learning Timeline
Key Insights
The Power of Agentic Coding
Models like GLM-5 are capable of handling complex logic such as 'procedural generation' (automatically building infinite levels) in a single prompt without requiring any follow-up prompts.
Visual Improvement Tips
While the initial result is already impressive, you can add follow-up prompts to further refine visual quality, add more complex enemy types, or fine-tune the character's physics system.
Prompts
One-Shot Platformer Game Generation
Target:
GLM-5
Create a 2D platformer game similar to Super Mario. Make it look amazing. Put everything in a standalone HTML file. Use publicly available assets, models, everywhere, smooth physics, two enemy types, bouncing slimes, and flying bats with wing animations, stomping mechanics. Include coin collection, a life system, and infinite procedurally generated levels.
Step by Step
Building a 2D Platformer Game with GLM-5
- Access the platform or interface that uses the GLM-5 AI model.
- Enter a highly specific prompt into the chat box (refer to the Prompt Card for the full text).
- Click the 'Send' or 'Generate' button to begin the agentic coding process.
- Wait for the AI to finish generating the entire application code in a single standalone file.
- Copy the generated code and save it as a new file in .html format (e.g., mario_game.html).
- Open the .html file using your preferred web browser.
- Click the 'Start Game' button on the game interface to begin playing.
- Use your mouse to 'zoom in' and 'zoom out' on the game display if necessary.
- Test the game mechanics, such as collecting coins, stomping on enemies, and controlling character movement using the keyboard.
