Assemble Your Crew: Automatic Multi-agent Communication Topology Design via Autoregressive Graph Generation

2025-07-24

AI Summary

Existing multi-agent systems (MAS) rely on predefined agent sets and hardcoded interaction topologies, lacking the capability to autonomously generate task-adaptive collaboration structures. Method: We propose the first conditional autoregressive graph generation framework for MAS design, formulating it as an end-to-end graph generation task conditioned on natural language task descriptions. Our approach jointly generates the number of agents, their role types, and inter-agent communication links—bypassing rigid template-based paradigms. It supports dynamic topology construction, integration of extensible role libraries, and task-driven role evolution. Results: Evaluated on six benchmark tasks, our method achieves state-of-the-art performance, with significantly improved inference efficiency: token consumption reduced by 23%–41%. The system demonstrates enhanced scalability and customization capability, enabling flexible, task-specific MAS configuration without manual architectural engineering.

Technology Category

• Multiagent Systems: Agent Communication
• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Natural Language Processing: Generation

Application Category

• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Search and Retrieval-Augmented AI: Agentic search
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Multi-agent systems (MAS) based on large language models (LLMs) have emerged as a powerful solution for dealing with complex problems across diverse domains. The effectiveness of MAS is critically dependent on its collaboration topology, which has become a focal point for automated design research. However, existing approaches are fundamentally constrained by their reliance on a template graph modification paradigm with a predefined set of agents and hard-coded interaction structures, significantly limiting their adaptability to task-specific requirements. To address these limitations, we reframe MAS design as a conditional autoregressive graph generation task, where both the system composition and structure are designed jointly. We propose ARG-Designer, a novel autoregressive model that operationalizes this paradigm by constructing the collaboration graph from scratch. Conditioned on a natural language task query, ARG-Designer sequentially and dynamically determines the required number of agents, selects their appropriate roles from an extensible pool, and establishes the optimal communication links between them. This generative approach creates a customized topology in a flexible and extensible manner, precisely tailored to the unique demands of different tasks. Extensive experiments across six diverse benchmarks demonstrate that ARG-Designer not only achieves state-of-the-art performance but also enjoys significantly greater token efficiency and enhanced extensibility. The source code of ARG-Designer is available at https://github.com/Shiy-Li/ARG-Designer.

Problem

• Automated design of multi-agent communication topologies
• Overcoming template-based limitations in agent interaction structures
• Joint optimization of system composition and collaboration structure

Innovation

• Autoregressive graph generation for MAS design
• Dynamic agent and role selection from pool
• Conditional topology creation via task query