pith. sign in

arxiv: 2506.05364 · v2 · pith:A6S2DVCOnew · submitted 2025-05-26 · 💻 cs.SE

Survey of LLM Agent Communication with MCP: A Software Design Pattern Centric Review

Anjana Sarkar , Soumyendu Sarkar This is my paper

Pith reviewed 2026-05-25 08:31 UTC · model grok-4.3

classification 💻 cs.SE
keywords LLM agentsModel Context Protocolsoftware design patternsmulti-agent communicationMediator patternObserver patternscalabilityreliability
0
0 comments X

The pith

Classical software design patterns enhance reliability and scalability of communication in LLM agent systems using MCP.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This survey examines how established design patterns can organize interactions among LLM-based agents operating under the Model Context Protocol. It traces the move from isolated agents to collaborative multi-agent systems and pinpoints the communication problems that arise during that transition. By revisiting patterns such as Mediator, Observer, Publish-Subscribe, and Broker, the work maps them onto MCP frameworks with conceptual schematics and formal models. These mappings offer concrete structures for managing data flow and varying levels of agent autonomy. The paper applies the approach to financial processing examples and flags remaining security and interoperability questions.

Core claim

The paper claims that classical design patterns supply relevant structures for agent interactions inside MCP-compliant LLM systems. It supplies conceptual schematics and formal models that map communication pathways, examines architectural choices suited to different autonomy levels, and shows how the patterns meet operational needs in real-time financial processing and investment banking while identifying open challenges for scalable multi-agent ecosystems.

What carries the argument

Mapping of Mediator, Observer, Publish-Subscribe, and Broker patterns onto MCP frameworks to structure agent interactions and optimize data flow.

If this is right

  • Architectural variations of the patterns can be selected according to the degree of agent autonomy and overall system complexity.
  • The same mappings support specific operational requirements in domains such as real-time financial processing and investment banking.
  • Structured use of the patterns helps surface and address security risks and interoperability issues in multi-agent LLM setups.
  • The outlined directions point toward more robust and scalable multi-agent LLM ecosystems.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • LLM agents may still need pattern adjustments to accommodate probabilistic outputs and changing context.
  • The same pattern mappings could apply to multi-agent systems outside finance once the core MCP structures are in place.
  • Implementation trials could measure concrete performance gains when the patterns are added to existing MCP agent codebases.

Load-bearing premise

That classical software design patterns developed for traditional systems will translate effectively to LLM-driven agents without significant adaptation for probabilistic outputs or context management.

What would settle it

A side-by-side test of two MCP multi-agent systems, one built with the mapped patterns and one without, that measures whether reliability and scalability differ under identical load and agent count.

read the original abstract

This survey investigates how classical software design patterns can enhance the reliability and scalability of communication in Large Language Model (LLM)-driven agentic AI systems, focusing particularly on the Model Context Protocol (MCP). It examines the foundational architectures of LLM-based agents and their evolution from isolated operation to sophisticated, multi-agent collaboration, addressing key communication hurdles that arise in this transition. The study revisits well-established patterns, including Mediator, Observer, Publish-Subscribe, and Broker, and analyzes their relevance in structuring agent interactions within MCP-compliant frameworks. To clarify these dynamics, the article provides conceptual schematics and formal models that map out communication pathways and optimize data flow. It further explores architectural variations suited to different degrees of agent autonomy and system complexity. Real-world applications in domains such as real-time financial processing and investment banking are discussed, illustrating how these patterns and MCP can meet specific operational demands. The article concludes by outlining open challenges, potential security risks, and promising directions for advancing robust, interoperable, and scalable multi-agent LLM ecosystems.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper is a survey claiming that classical software design patterns (Mediator, Observer, Publish-Subscribe, Broker) can be mapped to LLM agent communication under the Model Context Protocol (MCP) to improve reliability and scalability. It reviews agent evolution, supplies conceptual schematics and formal models for data flow, discusses architectural variations by autonomy level, illustrates applications in real-time finance and investment banking, and outlines open challenges plus security risks.

Significance. If the mappings were shown to hold under LLM-specific conditions, the work would provide a reusable conceptual toolkit for structuring multi-agent systems. The survey format usefully consolidates pattern literature with MCP; the provision of formal models is a creditworthy attempt at rigor. However, the absence of any empirical validation, reproducibility artifacts, or adaptation for non-determinism limits significance to a high-level overview rather than actionable guidance.

major comments (2)
  1. [Abstract] Abstract: the central claim that the patterns 'enhance the reliability and scalability of communication' is load-bearing yet unsupported; the described formal models and schematics presuppose deterministic state transitions and reliable delivery, with no incorporation of LLM stochasticity (temperature, prompt drift, hallucination retry logic, or context truncation).
  2. [Pattern mappings] Section on pattern mappings (Mediator/Observer/Publish-Subscribe/Broker): the transfer of these patterns is presented without modeling variance in agent outputs or context-consistency checks, directly undermining the reliability claims made for MCP-compliant frameworks.
minor comments (2)
  1. The manuscript would benefit from an explicit limitations subsection addressing the deterministic-to-stochastic gap.
  2. Notation in the formal models should be defined consistently with standard software-engineering conventions for message-passing systems.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. We address each major comment below, agreeing where revisions are warranted to better reflect the survey's conceptual scope.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claim that the patterns 'enhance the reliability and scalability of communication' is load-bearing yet unsupported; the described formal models and schematics presuppose deterministic state transitions and reliable delivery, with no incorporation of LLM stochasticity (temperature, prompt drift, hallucination retry logic, or context truncation).

    Authors: We agree the abstract phrasing overstates the claims. The formal models are high-level conceptual mappings assuming reliable delivery. We will revise the abstract to state the patterns 'may enhance' reliability and scalability under suitable conditions, explicitly note the deterministic assumptions, and cross-reference the open challenges section where LLM stochasticity is discussed. revision: yes

  2. Referee: [Pattern mappings] Section on pattern mappings (Mediator/Observer/Publish-Subscribe/Broker): the transfer of these patterns is presented without modeling variance in agent outputs or context-consistency checks, directly undermining the reliability claims made for MCP-compliant frameworks.

    Authors: The mappings are structural analogies at a conceptual level. We concur that output variance and consistency checks are not modeled. We will revise the section to add a dedicated paragraph on these limitations, clarifying that the patterns provide a starting framework rather than guaranteed reliability in non-deterministic settings. revision: yes

Circularity Check

0 steps flagged

No circularity: survey of existing patterns with no derivations or self-referential claims

full rationale

The paper is a literature review that revisits established software design patterns (Mediator, Observer, Publish-Subscribe, Broker) and discusses their relevance to MCP-compliant LLM agent systems. The abstract and structure contain no equations, fitted parameters, predictions derived from inputs, or load-bearing self-citations. Central claims rest on conceptual mapping of prior art rather than any reduction of outputs to the paper's own definitions or fits. No steps match the enumerated circularity patterns.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

This is a survey paper with no new mathematical derivations, empirical claims, or postulated entities. No free parameters, axioms, or invented entities are introduced beyond standard references to existing design patterns.

pith-pipeline@v0.9.0 · 5706 in / 933 out tokens · 30767 ms · 2026-05-25T08:31:25.338487+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Semantic Attacks on Tool-Augmented LLMs: Securing the Model Context Protocol Against Descriptor-Level Manipulation

    cs.CR 2025-12 unverdicted novelty 6.0

    Descriptor-level manipulation in the Model Context Protocol can drive LLMs to unsafe tool selections in up to 36% of cases; a layered defense of integrity checks, auxiliary-LLM vetting, and runtime guardrails reduces ...

  2. When to Answer and When to Defer: A Decision Framework for Reliable Code Predictions

    cs.SE 2026-05 unverdicted novelty 5.0

    Introduces a unified framework integrating uncertainty estimation, calibration, and tool-based abstention for reliable code predictions in language models.

  3. A Comprehensive Survey of Self-Evolving AI Agents: A New Paradigm Bridging Foundation Models and Lifelong Agentic Systems

    cs.AI 2025-08 unverdicted novelty 5.0

    A comprehensive review of self-evolving AI agents that improve themselves over time, organized via a framework of inputs, agent system, environment, and optimizers, with domain-specific and safety discussions.

  4. Red Skills or Blue Skills? A Dive Into Skills Published on ClawHub

    cs.CL 2026-03 unverdicted novelty 4.0

    Analysis of ClawHub shows language-based functional divides in agent skills, with over 30% flagged suspicious and submission-time documentation enabling 73% accurate risk prediction.

Reference graph

Works this paper leans on

53 extracted references · 53 canonical work pages · cited by 4 Pith papers · 1 internal anchor

  1. [1]

    System architecture for agentic large language models, 2024

    Tianjun Zhang. System architecture for agentic large language models, 2024. UC Berkeley Electronic Theses and Dissertations. url: https://escholarship.org/uc/item/5zt3h2r2 30

    work page 2024

  2. [2]

    LLM agents: The ultimate guide

    SuperAnnotate. "LLM agents: The ultimate guide." 2025. Web blog post. Super2025. url: https://www.superannotate.com/blog/llm-agents

    work page 2025

  3. [3]

    Autonomous AI Agents: Leveraging LLMs for Adaptive Decision-Making in Real- World Applications

    Wrick Talukdar. Autonomous AI Agents: Leveraging LLMs for Adaptive Decision-Making in Real- World Applications. 2025. IEEE Computer Society Community Voices

    work page 2025

  4. [4]

    From RAG to Multi-Agent S ystems: A Survey of Modern Approaches in LLM Development

    Aquino, G. D. A. E., Raimundo Gomes and Israel Gondres Torné. 2025 "From RAG to Multi-Agent S ystems: A Survey of Modern Approaches in LLM Development" Preprints. https://doi.org/10.20944/preprints202502.0406.v1

    work page doi:10.20944/preprints202502.0406.v1 2025

  5. [5]

    Beyond Self-Talk: A Communication-Centric Survey of LLM-Based Multi-Agent S ystems

    Yan, Bingyu, Xiaoming Zhang, Litian Zhang, Lian Zhang, Ziyi Zhou, Dezhuang Miao and Chaozhuo Li. “Beyond Self-Talk: A Communication-Centric Survey of LLM-Based Multi-Agent S ystems.” ArXiv abs/2502.14321 (2025): n. pag

    work page arXiv 2025

  6. [6]

    Prompt Engineering Guide

    L LM Agents. Prompt Engineering Guide. 2025. url:https://www.promptingguide.ai/research/llm- agents

    work page 2025

  7. [7]

    What are LLM Agents?

    k2view. What are LLM Agents?. 2025. url:https://www.k2view.com/what-are-llm-agents/

    work page 2025

  8. [8]

    Multi-Agent Collaboration Mechanisms: A Survey of LLMs

    Khanh-Tung Tran, , Dung Dao, Minh-Duong Nguyen, Quoc-Viet Pham, Barry O'Sullivan, and Hoang D. N guyen. "Multi-Agent Collaboration Mechanisms: A Survey of LLMs." (2025)

    work page 2025

  9. [9]

    Building effective agents

    Engineering at Anthropic. Building effective agents. 2024. https://www.anthropic.com/engineering/building-effective-agents

    work page 2024

  10. [10]

    How to think about agent frameworks

    LangChain. How to think about agent frameworks. 2025. url:https://blog.langchain.dev/how-to- think-about-agent-frameworks/

    work page 2025

  11. [11]

    [6] Khanh T ung Tran, Dung Dao, Minh Duong Nguyen, Quoc Viet Pham, Barry O'Sullivan, Hoang D. N guyen. Multi-Agent Collaboration Mechanisms: A Survey of LLMs. 2025. doi: 10. 48550/ARXIV.2501.06322

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  12. [12]

    MCP – A Beginner’s Guide

    S andeep. MCP – A Beginner’s Guide. 2025. url:https://opencv.org/blog/model-context- protocol/

    work page 2025

  13. [13]

    What is Model Context Protocol?

    k2view. What is Model Context Protocol?. 2025. https://www.k2view.com/model-context- protocol/

    work page 2025

  14. [14]

    What Is the Model Context Protocol (MCP) and How It Works

    descope. What Is the Model Context Protocol (MCP) and How It Works. 2025. https://www.descope.com/learn/post/mcp

    work page 2025

  15. [15]

    LLM Agent Frameworks 2025: Guide & Comparison. 2025. url:https://www.chatbase.co/blog/llm-agent-framework-guide

    work page 2025

  16. [16]

    databricks

    Agent system design patterns. databricks. 2025. h ttps://docs.databricks.com/aws/en/generative-ai/guide/agent-system-design-patterns

    work page 2025

  17. [17]

    4 Agentic AI Design Patterns & Real-World Examples

    Cem Dilmegani, Mert Palazoğlu. 4 Agentic AI Design Patterns & Real-World Examples. 2025. url:https://research.aimultiple.com/agentic-ai-design-patterns/

    work page 2025

  18. [18]

    Beyond Self-Talk: A Communication-Centric Survey of LLM-Based Multi-Agent S ystems

    Bingyu Yan, , Xiaoming Zhang, Litian Zhang, Lian Zhang, Ziyi Zhou, Dezhuang Miao, and Chaozhuo Li. "Beyond Self-Talk: A Communication-Centric Survey of LLM-Based Multi-Agent S ystems." (2025). 31

    work page 2025

  19. [19]

    Why Do Multi-Agent LLM Systems Fail?

    Mert Cemri, , Melissa Z. Pan, Shuyi Yang, Lakshya A. Agrawal, Bhavya Chopra, Rishabh Tiwari, Kurt Keutzer, Aditya Parameswaran, Dan Klein, Kannan Ramchandran, Matei Zaharia, Joseph E. Gonzalez, and Ion Stoica. "Why Do Multi-Agent LLM Systems Fail?." (2025)

    work page 2025

  20. [20]

    Multi-agent systems.LangGraph.2025.url:https://langchain- ai.github.io/langgraph/concepts/multi_agent/

    work page 2025

  21. [21]

    The cognitive leap How to reimagine work with AI agents

    Deloitte. The cognitive leap How to reimagine work with AI agents. 2024. rl:https://www2.deloitte.com/content/dam/Deloitte/us/Documents/gen-ai-multi-agents-pov-2.pdf

    work page 2024

  22. [22]

    Exploring the opportunities and challenges of using large language models torepresent institutional agency in land system modelling

    Zeng, Y., C., Brown, J., Raymond, M., Byari, R., Hotz, and M., Rounsevell. "Exploring the opportunities and challenges of using large language models torepresent institutional agency in land system modelling".Earth System Dynamics 16, no.2 (2025): 423–449

    work page 2025

  23. [24]

    MCP broker pattern: Key concepts & applications in modern distributed systems

    Lance Johnson. MCP broker pattern: Key concepts & applications in modern distributed systems. 2025. url:https://www.byteplus.com/en/topic/542191?title=mcp-broker-pattern-key- concepts-applications-in-modern-distributed-systems

    work page 2025

  24. [25]

    Large Language Model Based Multi-Agent System Augmented Complex Event Processing Pipeline for Internet of Multimedia Things

    Zeeshan, Talha, et al. Large Language Model Based Multi-Agent System Augmented Complex Event Processing Pipeline for Internet of Multimedia Things. 2025. doi:10.48550/arXiv.2501.00906 (2025)

    work page doi:10.48550/arxiv.2501.00906 2025

  25. [26]

    Understanding the Model Context Protocol: Architecture. 2025. url:https://nebius.com/blog/posts/understanding-model-context-protocol-mcp-architecture

    work page 2025

  26. [27]

    modemcontextprotocol.Core architecture.2025.url:https://modelcontextprotocol.io/docs/concepts/architecture

    work page 2025

  27. [28]

    geeksforgeeks.Broker Pattern.2025.url:https://www.geeksforgeeks.org/broker-pattern/

    work page 2025

  28. [29]

    Edwin Lisowski.Model Context Protocol (MCP): Solution to AI Integration Bottlenecks.2025.url:https://addepto.com/blog/model-context-protocol-mcp-solution-to-ai- integration-bottlenecks/

    work page 2025

  29. [30]

    Open Protocols for Agent Interoperability Part 1: Inter-Agent Communication on MCP

    Nick Aldridge, Marc Brooker, and Swami Sivasubramanian. Open Protocols for Agent Interoperability Part 1: Inter-Agent Communication on MCP. 2025. AWS Open Source Blog. url:https://aws.amazon.com/blogs/opensource/open-protocols-for-agent-interoperability-part-1- inter-agent-communication-on-mcp/

    work page 2025

  30. [31]

    Chain of Agents: Large language models collaborating on long-context tasks

    Yusen Zhang, Student Researcher, and Ruoxi Sun. Chain of Agents: Large language models collaborating on long-context tasks. 2025. Google Research. url:https://research.google/blog/chain- of-agents-large-language-models-collaborating-on-long-context-tasks/

    work page 2025

  31. [32]

    Open Standards for Real-Time AI Integration – A Look at MCP

    Sijie Guo. Open Standards for Real-Time AI Integration – A Look at MCP. 2025. url:https://streamnative.io/blog/open-standards-real-time-ai-mcp

    work page 2025

  32. [33]

    A survey of agent interoperability protocols: Model Context Protocol (MCP), Agent Communication Protocol (ACP), Agent-to-Agent Protocol (A2A), and Agent Network Protocol (ANP)

    Abul Ehtesham, , Aditi Singh, Gaurav Kumar Gupta, and Saket Kumar. "A survey of agent interoperability protocols: Model Context Protocol (MCP), Agent Communication Protocol (ACP), Agent-to-Agent Protocol (A2A), and Agent Network Protocol (ANP)." (2025)

    work page 2025

  33. [34]

    Bringing Memory to AI: A Look at A2A and MCP-like Technologies Across Platforms

    Ofir Yakobi,Shir Sadon. Bringing Memory to AI: A Look at A2A and MCP-like Technologies Across Platforms. 2025. url:https://orca.security/resources/blog/bringing-memory-to-ai-mcp-a2a-agent- context-protocols/ 32

    work page 2025

  34. [35]

    Spheron Network.Building Your First Hierarchical Multi-Agent System.2025.url:https://blog.spheron.network/building-your-first-hierarchical-multi-agent-system

    work page 2025

  35. [36]

    Stuti Dhruv, Pawan Pawar.How to Build Multi Agent AI System.2025.https://www.aalpha.net/blog/how-to-build-multi-agent-ai-system/

    work page 2025

  36. [37]

    Isabelle Nguyen,Tanay Soni.Understanding the Model Context Protocol (MCP).2025.url:https://www.deepset.ai/blog/understanding-the-model-context-protocol-mcp

    work page 2025

  37. [38]

    Yingxuan Yang, Huacan Chai, Shuai Shao, Yuanyi Song, Siyuan Qi, Renting Rui, & Weinan Zhang. (2025). AgentNet: Decentralized Evolutionary Coordination for LLM-based Multi-Agent Systems

    work page 2025

  38. [39]

    Khanh-Tung Tran, Dung Dao, Minh-Duong Nguyen, Quoc-Viet Pham, Barry O'Sullivan, & Hoang D. Nguyen. (2025). Multi-Agent Collaboration Mechanisms: A Survey of LLMs

    work page 2025

  39. [40]

    Zhao Wang, Sota Moriyama, Wei-Yao Wang, Briti Gangopadhyay, & Shingo Takamatsu. (2025). Talk Structurally, Act Hierarchically: A Collaborative Framework for LLM Multi-Agent Systems

    work page 2025

  40. [41]

    Multi-Agent System Patterns in Financial Services: Architectures for Next-Generation AI Solutions

    Hasan Mehdi,Alfredo Castillo. Multi-Agent System Patterns in Financial Services: Architectures for Next-Generation AI Solutions. 2025. url:https://community.aws/content/2uDxjoo105xRO6Q7mfkogmOYTVp/multi-agent-system- patterns-in-financial-services-architectures-for-next-generation-ai-solutions

    work page 2025

  41. [42]

    What is Model Context Protocol and how to leverage it in the fintech industry?

    Gonzalo Sanchez. What is Model Context Protocol and how to leverage it in the fintech industry?. 2025. url:https://prometeoapi.com/en/blog/model-context-protocol-fintech

    work page 2025

  42. [43]

    He, P., Lin, Y., Dong, S., Xu, H., Xing, Y., & Liu, H. (2025). Red-Teaming LLM Multi-Agent Systems via Communication Attacks. Preprint

    work page 2025

  43. [45]

    Review of LLM based Control

    Soumyendu Sarkar, Sahand Ghorbanpour, Ricardo Luna Gutierrez, Antonio Guillen-Perez, Vineet Gundecha, Ashwin Ramesh Babu, Avisek Naug. Review of LLM based Control. (2025). url: https://www.researchgate.net/publication/390334927_Review_of_LLM_based_Control

    work page arXiv 2025

  44. [46]

    LLM Enhanced Bayesian Optimization for Scientific Applications like Fusion

    Sahand Ghorbanpour, Ricardo Luna Gutierrez, Vineet Gundecha, Desik Rengarajan, Ashwin Ramesh Babu, Soumyendu Sarkar. LLM Enhanced Bayesian Optimization for Scientific Applications like Fusion. (2024). url: https://openreview.net/forum?id=VnxFb2VG4E

    work page 2024

  45. [47]

    N-critics: Self-refinement of large language models with ensemble of critics

    Mousavi, S., Guti’errez, R.L., Rengarajan, D., Gundecha, V., Babu, A.R., Naug, A., Guillen-Perez, A., Sarkar, S. N-critics: Self-refinement of large language models with ensemble of critics. (2023). ArXiv abs/2310.18679

    work page arXiv 2023

  46. [48]

    Informed Tree of Thought: Cost-efficient Problem Solving with Large Language Models

    Sajad Mousavi, Desik Rengarajan, Ashwin Ramesh Babu, Sahand Ghorbanpour, Vineet Gundecha, Avisek Naug, Soumyendu Sarkar. Informed Tree of Thought: Cost-efficient Problem Solving with Large Language Models. (2024). url: https://openreview.net/forum?id=EJNkaV27yB

    work page 2024

  47. [49]

    Imitation Guided Automated Red Teaming

    Sajad Mousavi, Desik Rengarajan, Ashwin Ramesh Babu, Vineet Gundecha, Antonio Guillen, Ricardo Luna Gutierrez, Avisek Naug, Sahand Ghorbanpour, Soumyendu Sarkar. Imitation Guided Automated Red Teaming. (2024). url: https://openreview.net/forum?id=STIC40U1es

    work page 2024

  48. [50]

    Suchow, Rong Liu, Zhenyu Cui, Zhaozhuo Xu, Denghui Zhang, Koduvayur Subbalakshmi, Guojun Xiong, Yueru He, Jimin Huang, Dong Li, & Qianqian Xie

    Yangyang Yu, Zhiyuan Yao, Haohang Li, Zhiyang Deng, Yupeng Cao, Zhi Chen, Jordan W. Suchow, Rong Liu, Zhenyu Cui, Zhaozhuo Xu, Denghui Zhang, Koduvayur Subbalakshmi, Guojun Xiong, Yueru He, Jimin Huang, Dong Li, & Qianqian Xie. (2024). FinCon: A Synthesized LLM Multi-Agent System with Conceptual Verbal Reinforcement for Enhanced Financial Decision Making. 33

    work page 2024

  49. [51]

    TradingAgents: Multi-Agents LLM Financial Trading Framework

    Yijia Xiao, , Edward Sun, Di Luo, and Wei Wang. "TradingAgents: Multi-Agents LLM Financial Trading Framework." (2025)

    work page 2025

  50. [53]

    LLM-Powered AI Agent Systems and Their Applications in Industry

    Guannan Liang, , and Qianqian Tong. "LLM-Powered AI Agent Systems and Their Applications in Industry." (2025)

    work page 2025

  51. [54]

    Agent Communication Language Definition: Understanding Its Role in Multi-Agent Systems

    Alexander De Ridder. Agent Communication Language Definition: Understanding Its Role in Multi-Agent Systems. 2025. url:https://smythos.com/ai-agents/ai-agent-development/agent- communication-language-definition/

    work page 2025

  52. [55]

    AgentNet: Decentralized Evolutionary Coordination for LLM-based Multi-Agent Systems

    Yingxuan Yang, , Huacan Chai, Shuai Shao, Yuanyi Song, Siyuan Qi, Renting Rui, and Weinan Zhang. "AgentNet: Decentralized Evolutionary Coordination for LLM-based Multi-Agent Systems." (2025)

    work page 2025

  53. [56]

    INVESTORBENCH: A Benchmark for Financial Decision-Making Tasks with LLM-based Agent

    Li, Haohang, Yupeng, Cao, Yangyang, Yu, Shashidhar Reddy, Javaji, Zhiyang, Deng, Yueru, He, Yuechen, Jiang, Zining, Zhu, Koduvayur, Subbalakshmi, Guojun, Xiong, Jimin, Huang, Lingfei, Qian, Xueqing, Peng, Qianqian, Xie, and Jordan, Suchow. "INVESTORBENCH: A Benchmark for Financial Decision-Making Tasks with LLM-based Agent".preprint (2024)

    work page 2024