RCR-Router: Efficient Role-Aware Context Routing for Multi-Agent LLM Systems with Structured Memory

2025-08-06

AI Summary

Existing multi-agent LLM systems rely on static or full-context routing, leading to excessive token consumption, redundant memory usage, and poor adaptability across multi-turn interactions. To address these limitations, we propose RCR-Router—the first dynamic context routing framework jointly conditioned on agent roles and task progression stages. It incorporates structured shared memory, lightweight semantic relevance scoring, iterative context integration, and dynamic subset selection under explicit token budget constraints. Crucially, we introduce Answer Quality Score (AQ-Score), an output-aware metric enabling fine-grained evaluation of generated responses. Extensive experiments on HotPotQA, MuSiQue, and 2WikiMultihop demonstrate that RCR-Router reduces token consumption by up to 30% while maintaining or improving answer accuracy. These results validate the method’s effectiveness, robustness, and scalability in complex multi-hop reasoning tasks.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Multiagent Systems: Multiagent Learning
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

Multi-agent large language model (LLM) systems have shown strong potential in complex reasoning and collaborative decision-making tasks. However, most existing coordination schemes rely on static or full-context routing strategies, which lead to excessive token consumption, redundant memory exposure, and limited adaptability across interaction rounds. We introduce RCR-Router, a modular and role-aware context routing framework designed to enable efficient, adaptive collaboration in multi-agent LLMs. To our knowledge, this is the first routing approach that dynamically selects semantically relevant memory subsets for each agent based on its role and task stage, while adhering to a strict token budget. A lightweight scoring policy guides memory selection, and agent outputs are iteratively integrated into a shared memory store to facilitate progressive context refinement. To better evaluate model behavior, we further propose an Answer Quality Score metric that captures LLM-generated explanations beyond standard QA accuracy. Experiments on three multi-hop QA benchmarks -- HotPotQA, MuSiQue, and 2WikiMultihop -- demonstrate that RCR-Router reduces token usage (up to 30%) while improving or maintaining answer quality. These results highlight the importance of structured memory routing and output-aware evaluation in advancing scalable multi-agent LLM systems.

Problem

• Dynamic role-based memory routing for multi-agent LLMs
• Reducing token usage while maintaining answer quality
• Structured memory and output-aware evaluation in multi-agent systems

Innovation

• Dynamic role-aware memory selection for agents
• Lightweight scoring policy guides memory routing
• Iterative output integration for context refinement