Game balancing is a longstanding challenge requiring repeated playtesting, expert intuition, and extensive manual tuning. We introduce RuleSmith, the first framework that achieves automated game balancing by leveraging the reasoning capabilities of multi-agent LLMs. It couples a game engine, multi-agent LLMs self-play, and Bayesian optimization operating over a multi-dimensional rule space. As a proof of concept, we instantiate RuleSmith on CivMini, a simplified civilization-style game containing heterogeneous factions, economy systems, production rules, and combat mechanics, all governed by tunable parameters. LLM agents interpret textual rulebooks and game states to generate actions, to conduct fast evaluation of balance metrics such as win-rate disparities. To search the parameter landscape efficiently, we integrate Bayesian optimization with acquisition-based adaptive sampling and discrete projection: promising candidates receive more evaluation games for accurate assessment, while exploratory candidates receive fewer games for efficient exploration. Experiments show that RuleSmith converges to highly balanced configurations and provides interpretable rule adjustments that can be directly applied to downstream game systems. Our results illustrate that LLM simulation can serve as a powerful surrogate for automating design and balancing in complex multi-agent environments.
LLM self-play + Bayesian optimization can automatically balance asymmetric strategy games from natural-language rulebooks.
Figure 1: Overview of RuleSmith. Multi-agent LLMs perform zero-shot self-play using solely the rule book under parameterized rule sets to automatically optimize asymmetric strategy games and other rule-driven systems.
Executable self-play from natural-language rulebooks. We show that multi-agent LLMs can perform zero-shot self-play in an executable, asymmetric strategy game from natural-language rulebooks and structured game states, producing legal and verifiable actions without training.
Multi-agents pipeline for automated game balancing. We present a general framework that integrates multi-agent LLMs self-play with Bayesian optimization and acquisition-based adaptive sampling to automatically adjust rule parameters and achieve balanced outcomes in asymmetric strategic environments.
Comprehensive empirical validation. Through systematic evaluations on CivMini across different model sizes (2B, 8B) and faction configurations, we demonstrate that RuleSmith consistently achieves near-balanced outcomes (50% ± 5% win rates), with interpretable parameters that transfer across evaluation settings.
Method Overview
We consider balancing an asymmetric, parameterized, turn-based strategy game by optimizing its rule parameters so that two roles (Empire and Nomads) achieve approximately equal win rates when controlled by LLM agents. RuleSmith uses two LLM agents to play the game from a natural-language rulebook; a Bayesian optimizer with acquisition-based adaptive sampling searches the rule space, allocating more evaluation games to promising candidates. Continuous proposals are discretized to valid rule configurations (using the precision above) before evaluation.
Figure 2: Overview of the RuleSmith method. We represent CivMini as a parameterized rule space θ. Given a candidate θ_t, two LLM agents (Empire and Nomads) play N_t self-play games, producing a balance loss L(θ). A Bayesian optimizer maintains a surrogate g(θ) and selects new candidates by maximizing an acquisition function. The number of games N_t is adaptively set by Expected Improvement; continuous proposals are mapped to discrete rulesets via D(·) before evaluation.
The Game: CivMini
CivMini is a 7×7 grid, turn-based asymmetric game. Empire has Farmers (gather only) and Soldiers (combat only); Nomads have Cavalry (combat + move 2 cells/turn, gain resources by killing). Each turn, each unit takes one action: GATHER, MOVE, BATTLE, PRODUCE_RESOURCE, PRODUCE_UNIT, or PASS. Win: destroy the opponent’s city, or else highest score at turn limit (score = weighted resources + battles won + surviving units).
12 tunable parameters (RuleSmith optimizes these for balance):
Parameter
Range
Precision
Economy
Initial resources
[2, 10]
integer
Empire farmer gather
[1, 5]
integer
Nomads kill resource gain
[1, 10]
integer
Combat
Empire damage
[1, 5]
integer
Nomads damage
[1, 5]
integer
Empire soldier HP
[4, 16]
integer
Nomads cavalry HP
[4, 16]
integer
Production
Empire unit cost
[2, 10]
integer
Nomads unit cost
[2, 10]
integer
Scoring
Resource weight
[0.1, 0.5]
0.1
Battle weight
[1, 5]
integer
Unit weight
[1, 5]
integer
Experimental Results
Optimized winning chances under different training (rows) and evaluation (columns). E = Empire, N = Nomads. Each cell: Empire wins | Nomads wins. Near-balanced (50% ± 5%) in bold. Model sizes: 2B, 8B.
Train \ Eval
E2B vs N2B
E2B vs N8B
E8B vs N2B
E8B vs N8B
E2B vs N2B
48 | 52
32 | 68
27 | 73
55 | 45
E2B vs N8B
81 | 19
47 | 53
91 | 9
75 | 25
E8B vs N2B
37 | 63
6 | 94
52 | 48
29 | 71
E8B vs N8B
53 | 47
24 | 76
81 | 19
51 | 49
Ablation on optimization methods. Random Search and (1+1)-ES use fixed N=64 games per iteration. BO with adaptive sampling uses N ∈ [16, 64]. Win rates as Empire | Nomads.
Random Search
(1+1)-ES
BO (adaptive)
13 | 87
26 | 74
51 | 49
BO (N=16)
BO (N=32)
BO (N=64)
34 | 66
61 | 39
48 | 52
Ablation on game designs. RuleSmith achieves balanced win rates across map sizes and turn limits (turns in parentheses).
5×5 (16)
7×7 (16)
9×9 (32)
11×11 (32)
53 | 47
51 | 49
48 | 52
51 | 49
Game Rollout Visualizations
Representative CivMini games under balanced parameters (InternVL3.5-8B for both factions).
Nomad high-score victory (16 turns). The Nomads expanded by producing six cavalry units and overwhelmed the Empire's defense. The Empire attempted to hold a defensive line with farmers and soldiers but was gradually eliminated, resulting in a board dominated by Nomad cavalry.Empire defense victory (16 turns). The Empire successfully defended against the Nomads' early aggression. Although the Nomads deployed four cavalry units to threaten the top-left, the Empire established a defense with three soldiers. After heavy attrition, the Empire secured the win through economic dominance, utilizing Farmer units to generate resources throughout the battle.Rapid Nomad conquest (6 turns). Nomad Cavalry Unit 0 executed consecutive strikes on the Empire's city from turns 3 to 6, causing sustained damage. This aggression resulted in the destruction of the Empire's city and a rapid win.Empire score victory (16 turns). The Empire successfully defended against a direct Nomad attack on their city. The Nomads rushed the top-left and began attacking on turn 5, but the soldiers systematically eliminated the cavalry. The Empire retained two soldiers and two farmers, securing a higher score through unit survival and resource accumulation.
Citation
If you find our work useful, please cite:
@article{zeng2026rulesmith,
title = {Rulesmith: multi-agent llms for automated game balancing},
author = {Zeng, Ziyao and Liu, Chen and Liu, Tianyu and Wang, Hao and Sun, Xiatao and Yang, Fengyu and Liu, Xiaofeng and Fan, Zhiwen},
journal = {arXiv preprint arXiv:2602.06232},
year = {2026}
}
