DOOMQL: Building a Doom-like game engine using SQLite and LLMs

DOOMQL is a functional, Doom-inspired game engine built entirely on top of SQLite. Rather than using a traditional game engine, the project uses SQL queries to manage core mechanics, including player movement, collision detection, enemy behavior, combat, and progression. The engine renders graphics by calculating RGB pixel values directly through a complex recursive common table expression within the database.

The project is implemented as a Python terminal script that generates a local SQLite database to track the game state. By integrating the Datasette Apps plugin, users can visualize the game in a web browser. This setup allows for a custom HTML and JavaScript interface that pulls live data from the database, enabling features like a real-time screen display and a dynamic minimap that updates as the game progresses.

This experiment demonstrates the technical flexibility of SQLite by pushing it beyond its standard role as a data storage tool. By leveraging large language models to write the necessary SQL and interface code, the project shows how complex, interactive systems can be constructed using unconventional database-driven architectures. It serves as a practical example of how modern AI-assisted programming can bridge the gap between database management and real-time graphical applications.