Recursive Game Creator tops GameCraft-Bench with experience-oriented agentic loop
An arXiv paper introduces Recursive Game Creator, a recursive agentic game-development harness built around four components — Designer, Builder, Player and Reviewer — that aims to push playable prototypes toward genuinely entertaining games. The authors report an overall score of 77.89 on GameCraft-Bench, a strict task success rate of 53.2% on GameASG-Bench (a 34.1% improvement over the same-model baseline), and the highest mean runtime-check pass rate among compared methods at 93.4%. They also report a user study showing longer playtime and higher ratings, and say code is coming soon.
It shifts evaluation of agentic game generation from program correctness toward player experience, and uses programmatic interfaces to collect gameplay trajectories at scale to reduce evaluation bias — a useful reference for future game-agent research.
Computer Science > Artificial Intelligence
arXiv:2610.08621 (cs)
[Submitted on 6 Oct 2026]
Title:Recursive Game Creator: An Agentic Product-Level Experience-Oriented Game Harness
Authors:Jiajun Chen, Haoyu Wu, Mingda Jia, Xihui Liu
View a PDF of the paper titled Recursive Game Creator: An Agentic Product-Level Experience-Oriented Game Harness, by Jiajun Chen and 3 other authors