TCP-MCP: Landscape-Guided Co-Evolution of Prompts and Communication Topologies for Multi-Agent Systems

2026-05-26

AI Summary

This work addresses the challenge of jointly optimizing task performance, communication overhead, and structural complexity in multi-agent systems, where prompt design and communication topology are typically optimized in isolation. The authors propose TCP-MCP, a novel framework that unifies prompts and communication topology into a single genomic representation and co-evolves them through a collaborative evolutionary process. By incorporating landscape probing during initialization and Pareto-front diagnostics, the method adaptively balances accuracy, token cost, and topological complexity. Evaluated on DeepSeek-V3.2, TCP-MCP achieves state-of-the-art results of 82.66%, 89.96%, and 96.61% accuracy on MMLU-Pro, MMLU, and GSM8K, respectively—significantly outperforming automated graph-generation baselines and reducing token consumption by up to 5.69× compared to debate-based systems.

Abstract

Effective multi-agent systems cannot be designed by selecting prompts or communication graphs in isolation. Agent behavior depends on the information an agent receives, while the usefulness of a communication edge depends on how the receiving agent interprets and uses that information. We propose \textbf{TCP-MCP} (Topology-Coupled Prompting for Multi-Agent Collaborative Problem-Solving), a co-evolution framework that searches agent prompts and communication topologies as a unified genome. TCP-MCP uses an initialization-time landscape probe to calibrate early search behavior, and then relies on Pareto-front diagnostics to adapt exploration under three objectives: task performance, token cost, and structural complexity. Using the same DeepSeek-V3.2 backbone across all methods, TCP-MCP achieves 82.66\%, 89.96\%, and 96.61\% accuracy on MMLU-Pro, MMLU, and GSM8K, respectively. Across the three benchmarks, it consistently outperforms automated graph-generation baselines and achieves competitive accuracy relative to debate-style systems, while using up to 5.69$\times$ fewer tokens than those systems at the reported operating points. These results show that jointly evolving prompts and communication structure provides a practical route to cost-aware and task-adaptive multi-agent system design in controlled evaluations.

Problem

• multi-agent systems
• prompt optimization
• communication topology
• co-evolution
• task-adaptive design

Innovation

• co-evolution
• prompt optimization
• communication topology
• Pareto-front adaptation
• multi-agent systems