Data to 7 Oct 2026

Cut the Crap: An Economical Communication Pipeline for LLM-based Multi-Agent Systems

arXiv 2410.02506 ↗First posted 3 Oct 2024cs.MA cs.LGcited by 129

On the map: LLM agents and tool use → General LLM agent frameworks and benchmarks

Abstract (arXiv)

Recent advancements in large language model (LLM)-powered agents have shown that collective intelligence can significantly outperform individual capabilities, largely attributed to the meticulously designed inter-agent communication topologies. Though impressive in performance, existing multi-agent pipelines inherently introduce substantial token overhead, as well as increased economic costs, which pose challenges for their large-scale deployments. In response to this challenge, we propose an economical, simple, and robust multi-agent communication framework, termed textttAgentPrune, which can seamlessly integrate into mainstream multi-agent systems and prunes redundant or even malicious communication messages. Technically, textttAgentPrune is the first to identify and formally define the textitcommunication redundancy issue present in current LLM-based multi-agent pipelines, and efficiently performs one-shot pruning on the spatial-temporal message-passing graph, yielding a token-economic and high-performing communication topology. Extensive experiments across six benchmarks demonstrate that textttAgentPrune (I) achieves comparable results as state-of-the-art topologies at merely 5.6 cost compared to their 43.7, (II) integrates seamlessly into existing multi-agent frameworks with 28.1%~72.8%downarrow token reduction, and (III) successfully defend against two types of agent-based adversarial attacks with 3.5%~10.8%uparrow performance boost.

Key facts. Cited by 129 papers in the map (data through 7 Oct 2026). Placed in General LLM agent frameworks and benchmarks, one of 16 directions in LLM agents and tool use. Builds on 59 of its 79 references that are in the map. First posted on arXiv 3 Oct 2024.

Builds on

59 of 79 references.

Closest ancestors · 15

Same direction, newest first

3 more

Other references · 44

2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,302
2021Training Verifiers to Solve Math Word Problemscited by 9,743
2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,355
2021Evaluating Large Language Models Trained on Codecited by 8,053
2020Measuring Massive Multitask Language Understandingcited by 7,768
2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,601
2023Tree of Thoughts: Deliberate Problem Solving with Large Language Modelscited by 3,155
2023Voyager: An Open-Ended Embodied Agent with Large Language Modelscited by 2,081
2023Improving Factuality and Reasoning in Language Models through Multiagent Debatecited by 1,739
2023Encouraging Divergent Thinking in Large Language Models through Multi-Agent Debatecited by 995
2021Are NLP Models really able to Solve Simple Math Word Problems?cited by 861
2023DSPy: Compiling Declarative Language Model Calls into Self-Improving Pipelinescited by 803

32 more

2017Program Induction by Rationale Generation : Learning to Solve and Explain Algebraic Word Problemscited by 642
2023API-Bank: A Comprehensive Benchmark for Tool-Augmented LLMscited by 574
2023LLM-Blender: Ensembling Large Language Models with Pairwise Ranking and Generative Fusioncited by 550
2022Complexity-Based Prompting for Multi-Step Reasoningcited by 387
2016Solving General Arithmetic Word Problemscited by 353
2023Benchmarking Large Language Models in Retrieval-Augmented Generationcited by 307
2023Deja Vu: Contextual Sparsity for Efficient LLMs at Inference Timecited by 247
2023Exploring Collaboration Mechanisms for LLM Agents: A Social Psychology Viewcited by 221
2023Improving Language Model Negotiation with Self-Play and In-Context Learning from AI Feedbackcited by 202
2019Drawing Early-Bird Tickets: Towards More Efficient Training of Deep Networkscited by 160
2023Progressive-Hint Prompting Improves Reasoning in Large Language Modelscited by 147
2023Self-Taught Optimizer (STOP): Recursively Self-Improving Code Generationcited by 128
2024Simulating Classroom Education with LLM-Empowered Agentscited by 125
2021A Unified Lottery Ticket Hypothesis for Graph Neural Networkscited by 118
2023ToolChain*: Efficient Action Space Navigation in Large Language Models with A* Searchcited by 103
2023Exchange-of-Thought: Enhancing Large Language Model Capabilities through Cross-Model Communicationcited by 78
2023Corex: Pushing the Boundaries of Complex Reasoning through Multi-Model Collaborationcited by 70
2024Evaluating Language Models for Efficient Code Generationcited by 69
2022Parsel: Algorithmic Reasoning with Language Models by Composing Decompositionscited by 62
2023GSLB: The Graph Structure Learning Benchmarkcited by 49
2024Learning Multi-Agent Communication from Graph Modeling Perspectivecited by 47
2024Reasoning in Token Economies: Budget-Aware Evaluation of LLM Reasoning Strategiescited by 44
2023L2MAC: Large Language Model Automatic Computer for Extensive Code Generationcited by 30
2023Large Language Model as a Policy Teacher for Training Reinforcement Learning Agentscited by 30
2023A Review of Cooperation in Multi-agent Learningcited by 27
2024Depending on yourself when you should: Mentoring LLM with RL agents to become the master in cybersecurity gamescited by 20
2023Demystifying Graph Sparsification Algorithms in Graph Properties Preservationcited by 14
2023ANPL: Towards Natural Programming with Interactive Decompositioncited by 13
2024A Simple and Yet Fairly Effective Defense for Graph Neural Networkscited by 12
2024Computational Experiments Meet Large Language Model Based Agents: A Survey and Perspectivecited by 11
2020Learning Multi-Agent Coordination through Connectivity-driven Communicationcited by 10
2024Two Heads Are Better Than One: Boosting Graph Sparse Training via Semantic and Topological Awarenesscited by 10

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 3; 2025: 53; 2026: 73.

129 papers.

Sort:

By area:

2025A-MEM: Agentic Memory for LLM Agentscited by 1,000
2024AFlow: Automating Agentic Workflow Generationcited by 395
2025Multi-agent Architecture Search via Agentic Supernetcited by 164
2025G-Memory: Tracing Hierarchical Memory for Multi-Agent Systemscited by 112
2024G-Designer: Architecting Multi-agent Communication Topologies via Graph Neural Networkscited by 110
2025A Comprehensive Survey in LLM(-Agent) Full Stack Safety: Data, Training and Deploymentcited by 110
2025Multi-Agent Design: Optimizing Agents with Better Prompts and Topologiescited by 106
2025AgenTracer: Who Is Inducing Failure in the LLM Agentic Systems?cited by 105
2025AgentNet: Decentralized Evolutionary Coordination for LLM-based Multi-Agent Systemscited by 86
2025Debate or Vote: Which Yields Better Decisions in Multi-Agent Large Language Models?cited by 74
2025KVFlow: Efficient Prefix Caching for Accelerating LLM-Based Multi-Agent Workflowscited by 56
2025EvoFlow: Evolving Diverse Agentic Workflows On The Flycited by 49

38 more

2025Token Reduction Should Go Beyond Efficiency in Generative Models -- From Vision, Language to Multimodalitycited by 49
2026A Survey of Agentic Reasoning for Large Language Models: Towards Recursively Self-Improving and Collective Agentscited by 47
2025Don't Overthink It: A Survey of Efficient R1-style Large Reasoning Modelscited by 36
2025KVCOMM: Online Cross-context KV-cache Communication for Efficient LLM-based Multi-agent Systemscited by 35
2025AGENTSAFE: Benchmarking the Safety of Embodied Agents on Hazardous Instructionscited by 33
2025Efficient Agents: Building Effective Agents While Reducing Costcited by 32
2025Enabling Agents to Communicate Entirely in Latent Spacecited by 31
2025Two Heads are Better Than One: Test-time Scaling of Multi-agent Collaborative Reasoningcited by 29
2025Adaptive Graph Pruning for Multi-Agent Communicationcited by 28
2025Thought Communication in Multiagent Collaborationcited by 28
2025We Should Identify and Mitigate Third-Party Safety Risks in MCP-Powered Agent Systemscited by 27
2025MAS-ZERO: Designing Multi-Agent Systems with Zero Supervisioncited by 26
2025Large Language Models Miss the Multi-Agent Markcited by 25
2025Graphs Meet AI Agents: Taxonomy, Progress, and Future Opportunitiescited by 23
2025Heterogeneous Swarms: Jointly Optimizing Model Roles and Weights for Multi-LLM Systemscited by 22
2026From Static Templates to Dynamic Runtime Graphs: A Survey of Workflow Optimization for LLM Agentscited by 22
2025When Identity Skews Debate: Anonymization for Bias-Reduced Multi-Agent Reasoningcited by 20
2025Talk Structurally, Act Hierarchically: A Collaborative Framework for LLM Multi-Agent Systemscited by 17
2024FedSSP: Federated Graph Learning with Spectral Knowledge and Personalized Preferencecited by 16
2025Towards High Data Efficiency in Reinforcement Learning with Verifiable Rewardcited by 16
2025Rethinking the Reliability of Multi-agent System: A Perspective from Byzantine Fault Tolerancecited by 16
2026Auditing Multi-Agent LLM Reasoning Trees Outperforms Majority Vote and LLM-as-Judgecited by 16
2025Can Compressed LLMs Truly Act? An Empirical Evaluation of Agentic Capabilities in LLM Compressioncited by 15
2025MASLab: A Unified and Comprehensive Codebase for LLM-based Multi-Agent Systemscited by 14
2025SafeMobile: Chain-level Jailbreak Detection and Automated Evaluation for Multimodal Mobile Agentscited by 14
2025Graph-Augmented Large Language Model Agents: Current Progress and Future Prospectscited by 14
2025BlockA2A: Towards Secure and Verifiable Agent-to-Agent Interoperabilitycited by 14
2025Maestro: Learning to Collaborate via Conditional Listwise Policy Optimization for Multi-Agent LLMscited by 14
2026LatentMem: Customizing Latent Memory for Multi-Agent Systemscited by 12
2025Dynamic Generation of Multi-LLM Agents Communication Topologies with Graph Diffusion Modelscited by 11
2026Toward Efficient Agents: Memory, Tool learning, and Planningcited by 11
2025From What to Why: A Multi-Agent System for Evidence-based Chemical Reaction Condition Reasoningcited by 10
2025Automated Composition of Agents: A Knapsack Approach for Agentic Component Selectioncited by 10
2026Robust Multi-Agent LLMs under Byzantine Faultscited by 10
2025Stochastic Self-Organization in Multi-Agent Systemscited by 9
2026AgentConductor: Topology Evolution for Multi-Agent Competition-Level Code Generationcited by 9
2026Adaptive Collaboration with Humans: Metacognitive Policy Optimization for Multi-Agent LLMs with Continual Learningcited by 9
2025S-DAG: A Subject-Based Directed Acyclic Graph for Multi-Agent Heterogeneous Reasoningcited by 8

50 most cited of 129 All 129 (JSON)

Similar, not cited

Close in wording, no citation either way