RuleSmith: Multi-Agent LLMs for Automated Game Balancing

Ziyao Zeng, Chen Liu, Tianyu Liu, Hao Wang, Xiatao Sun, Fengyu Yang, Xiaofeng Liu, Zhiwen Fan · 2026

Read RuleSmith: Multi-Agent LLMs for Automated Game Balancing by Ziyao Zeng, Chen Liu, Tianyu Liu, Hao Wang, Xiatao Sun, Fengyu Yang, Xiaofeng Liu, Zhiwen Fan (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-06232.txt.

Description

Balancing a strategy game by hand takes endless playtesting and gut feel. Ziyao Zeng and 13 co-authors present RuleSmith, a framework where language model agents play against each other to tune game rules automatically. A small civilization-style test game, CivMini, has two uneven factions and a dozen tunable numbers.

Agents read the rulebook and the game state, propose moves, and the engine checks them, giving noisy win rate estimates. A Bayesian search then spends more games on promising rule settings. Repeated runs found near-even win rates, checked over hundreds of fresh games. Other search methods plugged into the same setup, and small human studies suggested the balance transfers to people.

Catalog record

CREATORZiyao Zeng, Chen Liu, Tianyu Liu, Hao Wang, Xiatao Sun, Fengyu Yang, Xiaofeng Liu, Zhiwen Fan
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE46 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-06232

Browse the full library · About the Tower of Babel