Evolve as a Team: Collaborative Self-Evolution for LLM-based Multi-Agent Systems

2026-05-28

AI Summary

This work addresses the limitations of existing large language model (LLM)-based multi-agent systems in complex, long-horizon tasks, where execution failures hinder self-improvement and entangled execution-communication trajectories impede effective experience reuse. To overcome these challenges, the authors propose the Meta-Team framework, which introduces, for the first time, a post-task collaborative reflection mechanism grounded in full execution context tracking. This enables multi-scale co-evolution spanning individual behaviors, collaborative strategies, and team organizational structures. Experimental results across six long-horizon benchmark tasks demonstrate that Meta-Team significantly outperforms single-agent systems, handcrafted multi-agent baselines, and current evolutionary approaches, thereby validating its superior reliability and scalability.

Abstract

LLM-based multi-agent systems (MAS) have emerged as an effective paradigm for complex and long-horizon tasks. However, in real-world tasks, MAS often exhibit various failures during execution and such failures are difficult to eliminate during design. This motivates experience-driven MAS evolution, where a system improves based on its own execution experience. Yet such evolution is challenging because MAS experience is prolonged and intricate, interleaving multiple agents' execution chains and communication messages, which makes it difficult to identify what should be improved. To address this challenge, we propose Meta-Team, an experience-driven MAS evolution framework based on collaborative self-evolution. Meta-Team preserves the execution context of each agent and coordinates post-task communication, enabling agents to exchange distributed evidence for evolution. Building on this design, Meta-Team conducts multi-scale self-evolution, transforming execution experience into reusable improvements to agent behaviors, inter-agent coordination, and team-level organization. Across six long-horizon agent benchmarks, Meta-Team consistently outperforms single-agent systems, hand-crafted MAS, and prior MAS evolution methods; further analyses demonstrate that Meta-Team enables more reliable and scalable MAS self-evolution.

Problem

• multi-agent systems
• self-evolution
• execution experience
• LLM-based agents
• system failure

Innovation

• collaborative self-evolution
• multi-agent systems
• experience-driven evolution
• execution context preservation
• multi-scale improvement