Abstract:Recent advancements in Large Language Models (LLMs) have led to a rapid growth of agentic systems capable of handling a wide range of complex tasks. However, current research largely relies on manual, task-specific design, limiting their adaptability to novel tasks. In this paper, we introduce a new research problem: Modularized LLM Agent Search (MoLAS). We propose a modular design space that abstracts existing LLM agent designs into four fundamental modules with uniform IO interface: Planning, Reasoning, Tool Use, and Memory. Building on this design space, we present a novel LLM agent search framework called AgentSquare, which introduces two core mechanisms, i.e., module evolution and recombination, to efficiently search for optimized LLM agents. To further accelerate the process, we design a performance predictor that uses in-context surrogate models to skip unpromising agent designs. Extensive experiments across six benchmarks, covering the diverse scenarios of web, embodied, tool use and game applications, show that AgentSquare substantially outperforms hand-crafted agents, achieving an average performance gain of 17.2% against best-known human designs. Moreover, AgentSquare can generate interpretable design insights, enabling a deeper understanding of agentic architecture and its impact on task performance. We believe that the modular design space and AgentSquare search framework offer a platform for fully exploiting the potential of prior successful designs and consolidating the collective efforts of research community. Code repo is available at <a class="link-external link-https" href="https://github.com/tsinghua-fib-lab/AgentSquare" rel="external noopener nofollow">this https URL</a>.

Design and Realization of Agent Action Schedule Component in Artificial Society Simulation

A Emotion Development Agent Model Based On Occ Model And Operant Conditioning

The Design Of A Network-Structured Artificial Social System Based On Agent-Based Modeling

Modeling Mental States of Social Agent

Agent Architecture and Collaboration for Supply Chain Management

Modular Object Oriented Real-time Simulation Scheduling Service

Memory-based Social Agent's Actions and Mental States Design Model

Modeling Agents Combined With Environment, Mental States And Utility

Design and Implementation of HL A Agent in Distributed Simulation Experiment Management System

LMAgent: A Large-scale Multimodal Agents Society for Multi-user Simulation

Component-Based Software Architecture of Planning and Scheduling Algorithm

Agent-based Modeling of Large-scale Complex Social Interactions.

Intelligent Agent-Assisted Adaptive Order Simulation System in the Artificial Stock Market

An Agent Mechanism For Behavioral Animation

From Individual to Society: A Survey on Social Simulation Driven by Large Language Model-based Agents

AgentSquare: Automatic LLM Agent Search in Modular Design Space

Reusability and Composability Analysis for an Agent-Based Hierarchical Modelling and Simulation Framework.

Feature-Ontology Based Semantic Specification Model For Simulation Component

Design and Implementation of Agent-based Space Equipments Simulation Resources Repository Systems Agent Communication Mechanism

Model Design of Dynamic Workshop Scheduling Based on Multi-Agent System

Multi-Paradigm Artificial Society Modeling and Multi-Agent Simulation Platform Framework