Mad House — Usborne Creepy Computer Games

2026-05-24 · Source: Simon Willison's Weblog · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering, Gaming & Interactive Media · Depth: Intermediate, quick

Summary

Mad House — Usborne Creepy Computer Games" is a newly released retro text-based escape game, inspired by the 1983 Usborne book "Creepy Computer Games." Players navigate a shifting house by controlling near and far doorways using keyboard controls (X/C and N/M) or on-screen buttons, aiming to align three exits before being caught by footsteps. The game features authentic CRT phosphor aesthetics, including scanlines and a pulsing HUD that tracks time and door alignment. This interactive version was created by feeding the original book's PDF into Claude, which then generated the JavaScript and HTML code based on a prompt requesting a mobile-friendly, retro-aesthetic recreation.

Key takeaway

For creative technologists exploring retro game development or modernizing classic content, you should consider large language models like Claude. This approach allows for rapid prototyping and implementation of interactive experiences directly from static source material, such as old computer books. Focus your prompt engineering on specific aesthetic requirements, like CRT phosphor effects and scanlines, to ensure an authentic and engaging user experience that resonates with nostalgia.

Key insights

AI can rapidly transform static legacy content into interactive, retro-styled digital experiences.

Principles

• AI can convert legacy content into interactive experiences
• Authentic retro aesthetics enhance nostalgic user engagement

Method

Feed a legacy game book PDF into an LLM like Claude, then prompt for a vanilla JavaScript/HTML recreation, specifying mobile-friendliness and desired retro aesthetic.

In practice

• Use LLMs to revive old game concepts from scanned books
• Implement CRT effects (scanlines, phosphor) for retro UI

Topics

• Retro Gaming
• Text-based Games
• AI-assisted Development
• Claude
• Usborne Books
• JavaScript
• Prompt Engineering

Best for: AI Engineer, Software Engineer, Creative Technologist

Related on AIssential

• Open Thread 421 — Genie 3 creates persistent, explorable AI-generated worlds by leveraging extensive spatial data and advanced world memory.
• The Commodore 64 and ZX Spectrum have been turned into retrofuturistic handhelds — New handhelds revive 1980s computer aesthetics for dedicated retro gaming, prioritizing design over raw emulation power.
• Character.AI expands storytelling with Books experience — AI platforms are evolving to offer immersive narrative experiences, raising both entertainment potential and significant user safety concerns.
• Turbo Pascal 3.02A, deconstructed — AI models can effectively deconstruct and interpret legacy binary executables into human-readable code.
• I Made GPT-5.5 Build a C64 Shoot ’Em Up. The Hard Part Wasn’t the Code. — AI excels at implementation velocity, but human expertise remains crucial for architectural and cultural design decisions.
• Once and for all - What Clawdbot Actually Is and Why It's Not Claude Code — Claude Code is for deep development work, while Clawdbot is for continuous personal automation.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Simon Willison's Weblog.