CARD: Towards Conditional Design of Multi-agent Topological Structures

2026-03-01

AI Summary

This work addresses the limitations of existing multi-agent systems, which typically rely on fixed or statically learned communication topologies that struggle to adapt to dynamic environmental changes such as model updates or tool modifications. To overcome this, the authors propose the CARD framework, which explicitly incorporates environmental signals into the communication graph generation process for the first time. By integrating a conditional variational graph encoder with environment-aware optimization and leveraging large language models, CARD enables runtime-adaptive construction of communication topologies. Furthermore, the AMACP protocol is introduced to support real-time structural adjustments. Evaluated on HumanEval, MATH, and MMLU benchmarks, CARD significantly outperforms both static-topology baselines and prompt-engineering approaches, demonstrating superior accuracy and cross-scenario robustness.

Technology Category

• Multiagent Systems: Agent Communication
• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Agentic search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Large language model (LLM)-based multi-agent systems have shown strong capabilities in tasks such as code generation and collaborative reasoning. However, the effectiveness and robustness of these systems critically depend on their communication topology, which is often fixed or statically learned, ignoring real-world dynamics such as model upgrades, API (or tool) changes, or knowledge source variability. To address this limitation, we propose CARD (Conditional Agentic Graph Designer), a conditional graph-generation framework that instantiates AMACP, a protocol for adaptive multi-agent communication. CARD explicitly incorporates dynamic environmental signals into graph construction, enabling topology adaptation at both training and runtime. Through a conditional variational graph encoder and environment-aware optimization, CARD produces communication structures that are both effective and resilient to shifts in model capability or resource availability. Empirical results on HumanEval, MATH, and MMLU demonstrate that CARD consistently outperforms static and prompt-based baselines, achieving higher accuracy and robustness across diverse conditions. The source code is available at: https://github.com/Warma10032/CARD.

Problem

• multi-agent systems
• communication topology
• dynamic environments
• LLM-based agents
• topology adaptation

Innovation

• conditional graph generation
• adaptive multi-agent communication
• dynamic topology design
• environment-aware optimization
• LLM-based multi-agent systems