ARIA: Runtime Inspection and Amendment for Multi-Agent Systems

Key points

• In LLM-based multi-agent systems (MAS), a local mistake can propagate as other agents build on it, turning a correctable error into a failed task.1
• Our key intuition is to turn execution-verified repairs into reusable supervision, moving repair search into data construction and training so that a learned model can intervene at runtime without repeatedly testing alternatives.1
• From these trajectories, we construct the ARIA dataset, to our knowledge the first large-scale dataset pairing failed multi-agent trajectories with execution-verified repairs.1
• Then, we develop the ARIA framework for MAS-agnostic runtime inspection and amendment, designed to integrate with any MAS that exposes execution trajectories and supports continuation replacement.1

Sources (1)

• [1]ARIA: Runtime Inspection and Amendment for Multi-Agent Systems
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 14, 06:27 PM

  We present Agent Runtime Inspection and Amendment (ARIA), a framework for learning when and how to repair ongoing multi-agent executions.

  In LLM-based multi-agent systems (MAS), a local mistake can propagate as other agents build on it, turning a correctable error into a failed task.

Extractive summary: sentences quoted from the sources.