DOOMQL
Simon Willison · Simon Willison · 2026-07-13
Peter Gostev used GPT-5.6 Sol to build DOOMQL, a Doom-like game in which SQLite is the game engine handling all movement, collision, combat, and pixel rendering via a recursive CTE ray tracer, which Simon Willison then extended with a real-time Datasette visualization layer built with Claude Fable 5.
Appears in
Extraction
Topics: ai-assisted-programmingsqlitecreative-codinggenerative-aidatasette
Claims
- DOOMQL implements a full ray tracer in SQLite using a single recursive CTE SQL query, driving all game logic through SQL rather than imperative code.
- GPT-5.6 Sol was used to build the core DOOMQL game engine.
- Simon Willison extended DOOMQL with a Datasette Apps plugin that renders the game's frame_pixels view as a live HTML canvas refreshed once per second, with a minimap added via a follow-up prompt to Claude Fable 5.
Key quotes
DOOMQL started with a deliberately unreasonable question: what if SQLite were the game engine, not merely the place where a game stores data?
The result is a small, original Doom-like game in which SQL owns movement, collision, enemies, combat, progression and every RGB pixel on screen.