Skill-MAS: Evolving Meta-Skill for Automatic Multi-Agent Systems

2026-06-17

AI Summary

This work addresses the limitations of existing large language model (LLM)-driven multi-agent systems, which struggle to accumulate reasoning experience and rely on training paradigms constrained by smaller models, hindering scalability to state-of-the-art LLMs. The authors propose Skill-MAS, a novel framework that decouples experience retention from parameter updates for the first time. It introduces an evolvable Meta-Skill abstraction to capture high-level coordination capabilities and establishes a closed-loop optimization mechanism through multi-trajectory replay, selective reflection, and hierarchical contrastive analysis, enabling non-parametric experience accumulation and generalization. Evaluated across four complex benchmarks and four mainstream LLMs, Skill-MAS demonstrates substantial performance gains with high cost efficiency, while the evolved Meta-Skills exhibit strong robustness and transferability across tasks and models.

Abstract

Large Language Model (LLM)-based automatic Multi-Agent Systems (MAS) generation has become a crucial frontier for tackling complex tasks. However, existing methods face a dilemma between model capability and experience retention. Inference-time MAS leverages frozen frontier LLMs but repeats identical searches without learning from past experience. Conversely, Training-time MAS internalizes experience via gradient updates but is constrained by the low capability ceiling of smaller models, and is hard to scale to large frontier LLMs. To bridge this gap, we propose Skill-MAS, a novel third path that decouples experience retention from parametric updates by conceptualizing the high-level orchestration capability as an evolvable Meta-Skill. Skill-MAS refines this architectural knowledge through a closed optimization loop: (1) Multi-Trajectory Rollout samples a behavioral distribution for each task under the current Meta-Skill; and (2) Selective Reflection adaptively selects priority tasks and applies hierarchical contrastive analysis to distill systemic experience into generalizable, strategy-level principles. Extensive experiments across four complex benchmarks and four distinct LLMs demonstrate that Skill-MAS not only achieves remarkable performance gains but also maintains a favorable cost-performance trade-off. Further analysis reveals that the evolved Meta-Skills are highly robust and exhibit strong transferability across unseen tasks and different LLMs.

Problem

• Multi-Agent Systems
• Large Language Models
• Experience Retention
• Model Capability
• Automatic MAS Generation

Innovation

• Meta-Skill
• Multi-Agent Systems
• Experience Retention
• Closed-loop Optimization
• Transferability