Pixie: Code-Level Mechanic Generation for Game Designers

Authors

  • Michael Cook King's College London

DOI:

https://doi.org/10.1609/aiide.v21i1.36824

Abstract

Procedural generation has been applied to many kinds of game content including levels, puzzles and narrative. Although PCG researchers have tackled the generation of rules and mechanics, these systems typically operate on bespoke codebases designed for mechanic invention tasks. In this paper we present Pixie, a system for generating and testing game mechanics that can be installed in any Unity project and configured using a simple annotation process to focus its scope and objectives. We describe the system's design and operation, demonstrate its ability to generate mechanics for several open-source Unity games, and provide an initial discussion of developer perspectives on the content it generates and its usefulness as a design companion.

Downloads

Published

2025-11-07

How to Cite

Cook, M. (2025). Pixie: Code-Level Mechanic Generation for Game Designers. Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment, 21(1), 206–215. https://doi.org/10.1609/aiide.v21i1.36824