Diagnose, Localize, Align: A Full-Stack Framework for Reliable LLM Multi-Agent Systems under Instruction Conflicts

• URL: http://arxiv.org/abs/2509.23188v1
• Date: Sat, 27 Sep 2025 08:43:34 GMT
• Title: Diagnose, Localize, Align: A Full-Stack Framework for Reliable LLM Multi-Agent Systems under Instruction Conflicts
• Authors: Guancheng Wan, Leixin Sun, Longxu Dou, Zitong Shi, Fang Wu, Eric Hanchen Jiang, Wenke Huang, Guibin Zhang, Hejia Geng, Xiangru Tang, Zhenfei Yin, Yizhou Sun, Wei Wang,
• Abstract summary: Large Language Model (LLM)-powered multi-agent systems (MAS) have rapidly advanced collaborative reasoning, tool use, and role-specialized coordination in complex tasks.<n>However, reliability-critical deployment remains hindered by a systemic failure mode: hierarchical compliance under instruction conflicts.
• Score: 75.20929587906228
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Large Language Model (LLM)-powered multi-agent systems (MAS) have rapidly advanced collaborative reasoning, tool use, and role-specialized coordination in complex tasks. However, reliability-critical deployment remains hindered by a systemic failure mode: hierarchical compliance under instruction conflicts (system-user, peer-peer), where agents misprioritize system-level rules in the presence of competing demands. Moreover, widely used macro-level metrics (e.g., pass@k) obscure these micro-level violations and offer little actionable guidance for remedy. In this work, we present a full-stack, three-stage framework: (1) Diagnose - Contextualized Role Adherence Score (CRAS), a query-wise, context-aware scoring metric that decomposes role adherence into four measurable dimensions; (2) Localize - attention drift analysis revealing that instruction conflicts are resolved by attention heads that are largely concentrated in middle layers; (3) Align - Surgical Alignment of Instruction Layers (SAIL), which installs LoRA only on the localized focal layers and optimizes a token-weighted DPO-style preference objective that credits tokens by their focal attentional contribution. Across standard benchmarks and MAS frameworks, our surgical approach improves instruction hierarchy compliance (e.g., +5.60% with AutoGen on MedQA) without full-model finetuning.

Related papers

• MAS-FIRE: Fault Injection and Reliability Evaluation for LLM-Based Multi-Agent Systems [38.44649280816596]
  We propose MAS-FIRE, a systematic framework for fault injection and reliability evaluation of Multi-Agent Systems.<n>We define a taxonomy of 15 fault types covering intra-agent cognitive errors and inter-agent coordination failures.<n>Applying MAS-FIRE to three representative MAS architectures, we uncover a rich set of fault-tolerant behaviors.
  arXiv  Detail & Related papers  (2026-02-23T13:47:43Z)
• MAS-Orchestra: Understanding and Improving Multi-Agent Reasoning Through Holistic Orchestration and Controlled Benchmarks [86.05918381895555]
  We propose MASOrchestra as a training-time framework that formulates MAS orchestration as a function-calling reinforcement learning problem.<n>In MAS-Orchestra, complex, goal-oriented subagents are abstracted as callable functions, enabling global reasoning over system structure.<n>Our analysis reveals that MAS gains depend critically on task structure, verification protocols, and the capabilities of both orchestrator and subagents.
  arXiv  Detail & Related papers  (2026-01-21T04:57:02Z)
• Beyond Monolithic Architectures: A Multi-Agent Search and Knowledge Optimization Framework for Agentic Search [56.78490647843876]
  Agentic search has emerged as a promising paradigm for complex information seeking by enabling Large Language Models (LLMs) to interleave reasoning with tool use.<n>We propose bfM-ASK, a framework that explicitly decouples agentic search into two complementary roles: Search Behavior Agents, which plan and execute search actions, and Knowledge Management Agents, which aggregate, filter, and maintain a compact internal context.
  arXiv  Detail & Related papers  (2026-01-08T08:13:27Z)
• Explainable and Fine-Grained Safeguarding of LLM Multi-Agent Systems via Bi-Level Graph Anomaly Detection [76.91230292971115]
  Large language model (LLM)-based multi-agent systems (MAS) have shown strong capabilities in solving complex tasks.<n>XG-Guard is an explainable and fine-grained safeguarding framework for detecting malicious agents in MAS.
  arXiv  Detail & Related papers  (2025-12-21T13:46:36Z)
• Metacognitive Self-Correction for Multi-Agent System via Prototype-Guided Next-Execution Reconstruction [58.51530390018909]
  Large Language Model based multi-agent systems excel at collaborative problem solving but remain brittle to cascading errors.<n>We present MASC, a metacognitive framework that endows MAS with real-time, unsupervised, step-level error detection and self-correction.
  arXiv  Detail & Related papers  (2025-10-16T05:35:37Z)
• AgentAsk: Multi-Agent Systems Need to Ask [26.13279490836716]
  Multi-agent systems built on large language models (LLMs) promise enhanced problem-solving capabilities through collaborative division of labor.<n>We propose AgentAsk, a lightweight and plug-and-play clarification module that treats every inter-agent message as a potential failure point and inserts minimally necessary questions to arrest error propagation.<n>AgentAsk consistently improves accuracy and robustness over public multi-agent implementations while keeping overhead minimal, with latency and extra cost all less than 5%.
  arXiv  Detail & Related papers  (2025-10-08T22:36:05Z)
• Vulnerable Agent Identification in Large-Scale Multi-Agent Reinforcement Learning [49.31650627835956]
  Partial agent failure becomes inevitable when systems scale up, making it crucial to identify the subset of agents whose compromise would most severely degrade overall performance.<n>In this paper, we study this Vulnerable Agent Identification (VAI) problem in large-scale multi-agent reinforcement learning (MARL)<n> Experiments show our method effectively identifies more vulnerable agents in large-scale MARL and the rule-based system, fooling system into worse failures, and learning a value function that reveals the vulnerability of each agent.
  arXiv  Detail & Related papers  (2025-09-18T16:03:50Z)
• Visual Document Understanding and Question Answering: A Multi-Agent Collaboration Framework with Test-Time Scaling [83.78874399606379]
  We propose MACT, a Multi-Agent Collaboration framework with Test-Time scaling.<n>It comprises four distinct small-scale agents, with clearly defined roles and effective collaboration.<n>It shows superior performance with a smaller parameter scale without sacrificing the ability of general and mathematical tasks.
  arXiv  Detail & Related papers  (2025-08-05T12:52:09Z)
• Parallelism Meets Adaptiveness: Scalable Documents Understanding in Multi-Agent LLM Systems [0.8437187555622164]
  Large language model (LLM) agents have shown increasing promise for collaborative task completion.<n>Existing multi-agent frameworks often rely on static, fixed roles, and limited inter-agent communication.<n>This paper proposes a coordination framework that enables adaptiveness through three core mechanisms.
  arXiv  Detail & Related papers  (2025-07-22T22:42:51Z)
• MARBLE: A Multi-Agent Rule-Based LLM Reasoning Engine for Accident Severity Prediction [1.3102025155414727]
  Accident severity prediction plays a critical role in transportation safety systems.<n>Existing methods often rely on monolithic models or black box prompting.<n>We propose a multiagent rule based LLM engine that decomposes the severity prediction task across a team of specialized reasoning agents.
  arXiv  Detail & Related papers  (2025-07-07T11:27:49Z)
• Multi-level Value Alignment in Agentic AI Systems: Survey and Perspectives [29.49571891159761]
  Value alignment for agentic AI systems aims to ensure that an agent's goals, preferences, and behaviors align with human values and societal norms.<n>This study comprehensively reviews value alignment in LLM-based multi-agent systems as the representative archetype of agentic AI systems.
  arXiv  Detail & Related papers  (2025-06-11T12:25:38Z)
• SentinelAgent: Graph-based Anomaly Detection in Multi-Agent Systems [11.497269773189254]
  We present a system-level anomaly detection framework tailored for large language model (LLM)-based multi-agent systems (MAS)<n>We propose a graph-based framework that models agent interactions as dynamic execution graphs, enabling semantic anomaly detection at node, edge, and path levels.<n>Second, we introduce a pluggable SentinelAgent, an LLM-powered oversight agent that observes, analyzes, and intervenes in MAS execution based on security policies and contextual reasoning.
  arXiv  Detail & Related papers  (2025-05-30T04:25:19Z)
• Benchmarking LLMs' Swarm intelligence [50.544186914115045]
  Large Language Models (LLMs) show potential for complex reasoning, yet their capacity for emergent coordination in Multi-Agent Systems (MAS) remains largely unexplored.<n>We introduce SwarmBench, a novel benchmark designed to systematically evaluate tasks of LLMs acting as decentralized agents.<n>We propose metrics for coordination effectiveness and analyze emergent group dynamics.
  arXiv  Detail & Related papers  (2025-05-07T12:32:01Z)
• Why Do Multi-Agent LLM Systems Fail? [91.39266556855513]
  We present MAST (Multi-Agent System Failure taxonomy), the first empirically grounded taxonomy designed to understand MAS failures.<n>We analyze seven popular MAS frameworks across over 200 tasks, involving six expert human annotators.<n>We identify 14 unique failure modes, organized into 3 overarching categories, (i) specification issues, (ii) inter-agent misalignment, and (iii) task verification.
  arXiv  Detail & Related papers  (2025-03-17T19:04:38Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.