MAS-on-the-Fly: In-Context Structural Adaptation of LLM-Based Multi-Agent Systems
cs.MA, cs.AI
Submitted: 2026-02-14
Updated: 2026-08-31
License: http://creativecommons.org/licenses/by-nc-nd/4.0/
The gist: Large Language Model (LLM)-based multi-agent systems (MAS) have emerged as a promising paradigm for complex tasks.
Terminology
Abstract
Large Language Model (LLM)-based multi-agent systems (MAS) have emerged as a promising paradigm for complex tasks. However, existing works often rely on manual designs or "one-size-fits-all" automation and lack adaptability after deployment. We study in-context structural adaptation, where structured experience conditions both query-dependent system generation and execution-time reconfiguration without updating LLM parameters. We introduce MASFly, which realizes this adaptation through two complementary mechanisms. First, a retrieval-augmented SOP instantiation mechanism retrieves and adapts successful collaboration patterns to construct a query-specific MAS. Second, an experience-enhanced process supervision mechanism uses a dedicated Watcher agent to monitor execution against prior failure experience and reconfigure the system upon abnormal behavior. Experiments demonstrate that MASFly achieves state-ofthe-art performance, including a 61.7% success rate on TravelPlanner, with strong task adaptability and robustness.
Sources
- FlowReasoner: Reinforcing Query-Level Meta-Agents
- DeepSeek-V3 Technical Report
- GPT-4 Technical Report
- Program Synthesis with Large Language Models
- ChatDev: Communicative Agents for Software Development
- Evaluating Large Language Models Trained on Code
- AgentSquare: Automatic LLM Agent Search in Modular Design Space
- AgentVerse: Facilitating Multi-Agent Collaboration and Exploring Emergent Behaviors
- Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters
- Multi-Agent Collaboration: Harnessing the Power of Intelligent LLM Agents
- Agent KB: Leveraging Cross-Domain Experience for Agentic Problem Solving
- MAS-GPT: Training LLMs to Build LLM-based Multi-Agent Systems
- HumanEval Pro and MBPP Pro: Evaluating Large Language Models on Self-invoking Code Generation
- Tent: Fully Test-time Adaptation by Entropy Minimization
- EvoAgent: Towards Automatic Multi-Agent Generation via Evolutionary Algorithms
- G-Memory: Tracing Hierarchical Memory for Multi-Agent Systems
- Multi-agent Architecture Search via Agentic Supernet
- Agent Workflow Memory
- G-Designer: Architecting Multi-agent Communication Topologies via Graph Neural Networks
- SwarmAgentic: Towards Fully Automated Agentic System Generation via Swarm Intelligence
Related papers
- Highway Congestion Reduction through Reinforcement Learning Based Eulerian Headway Control
- You Only Align Once: Propagating Cooperative Behaviors in Multi-Agent Systems through Seed Agents
- Deny Without Disabling: Authorization-Paired Evaluation and Control for Multi-Agent Systems
- MA-SAPO: Multi-Agent Reasoning for Score-Aware Prompt Optimization
- PeroMAS: A Multi-agent System of Perovskite Material Discovery
- StitchCUDA: An Automated Multi-Agents End-to-End GPU Programing Framework with Rubric-based Agentic Reinforcement Learning