Agent-Kernel: A MicroKernel Multi-Agent System Framework for Adaptive Social Simulation Powered by LLMs

2025-12-01

AI Summary

Existing multi-agent system (MAS) frameworks lack sufficient adaptability, configurability, and scalability to support large-scale, dynamic social simulations powered by large language models (LLMs), where agent count and attributes evolve over time. To address this, we propose a modular, microkernel-based framework designed for adaptive social simulation, adopting a “society-centric” architecture that decouples the system kernel, cognitive models, and physical environment—thereby achieving strict separation between simulation logic and infrastructure. The framework integrates LLM-driven agent behavior modeling, environment interaction, and dynamic population management to significantly enhance reusability and configuration flexibility. Experimental evaluation demonstrates its effectiveness in simulating population dynamics in the classic “Universe 25” scenario and in a heterogeneous, campus-wide simulation involving over 10,000 agents at Zhejiang University. It supports real-time coordination and structural evolution of tens of thousands of agents, establishing a scalable, evolvable foundational architecture for LLM-augmented social simulation.

Technology Category

• Multiagent Systems: Agent-Based Simulation and Emergent Behavior
• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Multi-Agent System (MAS) developing frameworks serve as the foundational infrastructure for social simulations powered by Large Language Models (LLMs). However, existing frameworks fail to adequately support large-scale simulation development due to inherent limitations in adaptability, configurability, reliability, and code reusability. For example, they cannot simulate a society where the agent population and profiles change over time. To fill this gap, we propose Agent-Kernel, a framework built upon a novel society-centric modular microkernel architecture. It decouples core system functions from simulation logic and separates cognitive processes from physical environments and action execution. Consequently, Agent-Kernel achieves superior adaptability, configurability, reliability, and reusability. We validate the framework's superiority through two distinct applications: a simulation of the Universe 25 (Mouse Utopia) experiment, which demonstrates the handling of rapid population dynamics from birth to death; and a large-scale simulation of the Zhejiang University Campus Life, successfully coordinating 10,000 heterogeneous agents, including students and faculty.

Problem

• Addresses limitations in adaptability, configurability, and reliability of existing multi-agent simulation frameworks
• Enables simulation of societies with dynamic agent populations and profiles over time
• Improves code reusability and supports large-scale, heterogeneous agent coordination

Innovation

• Microkernel architecture decouples core functions from simulation logic
• Separates cognitive processes from physical environments and action execution
• Validated through large-scale simulations with dynamic populations and 10,000 agents