Camel: Communicative agents for" mind" exploration of large language model society
Open accessThe DOI matches its Crossref record; 29 of the 29 references it cites were found in a registry or an index.
Abstract
The rapid advancement of chat-based language models has led to remarkable progress in complex task-solving. However, their success heavily relies on human input to guide the conversation, which can be challenging and time-consuming. This paper explores the potential of building scalable techniques to facilitate autonomous cooperation among communicative agents, and provides insight into their "cognitive" processes. To address the challenges of achieving autonomous cooperation, we propose a novel communicative agent framework named role-playing. Our approach involves using inception prompting to guide chat agents toward task completion while maintaining consistency with human intentions. We showcase how role-playing can be used to generate conversational data for studying the behaviors and capabilities of a society of agents, providing a valuable resource for investigating conversational language models. In particular, we conduct comprehensive studies on instruction-following cooperation in multi-agent settings. Our contributions include introducing a novel communicative agent framework, offering a scalable approach for studying the cooperative behaviors and capabilities of multi-agent systems, and open-sourcing our library to support research on communicative agents and beyond: https://github.com/camel-ai/camel.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“The development of medical AI agents can be approached through two primary methodologies: graphical user interface (GUI)-based frameworks and coding-based development.110,111,112”
From A foundational architecture for AI agents in healthcare (2025) · cites this paper as [111]
Citations
Cites
The publisher hasn’t deposited this paper’s reference list.
Cited by
Items 101–200 of 369
ESG Reporting Lifecycle Management with Large Language Models and AI Agents
Hoang, T., Klymenko, M., Xu, X., Pan, S., Ding, Y., Tang, X., Yang, Z., Shi, J., & Lo, D. (2026) · IEEE International Conference on Software Architecture Companion (ICSA-C) · doi:10.1109/icsa-c68850.2026.00014 · Cited by 11
“In addition, our proposed agentic ESG lifecycle adopts two main principles observed in advanced agentic AI systems: (1) agents specialize in specific tasks, and (2) agents exchange outputs to solve problems and improve performance [8], [18].”
Preacher: Paper-to-Video Agentic System
Liu, J., Yang, L., Luo, H., Wang, F., Li, H., & Wang, M. (2025) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51701.2025.01591 · Cited by 15
“Compared to single-agent approaches [1, 51], agen-tic systems harness collective intelligence and specialized expertise, enabling them to address complex challenges, including advanced programming tasks [11, 21, 30] and planning in physical environments [8, 17, 22, 46].”
Collaborative intelligence in AI: Evaluating the performance of a council of AIs on the USMLE
Shaikh, Y., Jeelani-Shaikh, Z. A., Jeelani, M., Javaid, A., Mahmud, T., Gaglani, S. M., Gibbons, M., Cheema, M., Cross, A., Livingston, D., et al. (2025) · PLOS Digital Health · doi:10.1371/journal.pdig.0000787 · Cited by 7
LLM-SAP: Large Language Models Situational Awareness-Based Planning
Wang, L., & Zhong, H. (2024) · IEEE International Conference on Multimedia and Expo workshops · doi:10.1109/icmew63481.2024.10645429 · Cited by 14
“Such collaboration leverages the individual strengths of each agent [28, 29, 30].”
BackgroundEnhancing math reasoning ability of large language models via computation logic graphs
Zhao, D., Han, D., Wu, J. X., He, Z., Ning, B., Yuan, Y., Li, Y., Wang, C., & Song, S. (2025) · Knowledge-Based Systems · doi:10.1016/j.knosys.2025.113905 · Cited by 18
Wang, J., Cheng, Y., Lin, D., Leong, C. T., & Li, W. (2023) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2023.emnlp-main.72 · Cited by 40
Advances in edge-cloud collaboration and evolution for large-small models
Wang, Y., Shen, T., Zhang, S., Wu, F., Zhao, Z., Cai, H., Lyu, C., Ma, L., Yang, C., & Wu, F. L. (2024) · Journal of Image and Graphics · doi:10.11834/jig.240011 · Cited by 12
Flows: Building Blocks of Reasoning and Collaborating AIPreprint
Josifoski, M., Klein, L., Peyrard, M., Baldwin, N., Li, Y., Geng, S., Schnitzler, J. P., Yao, Y., Wei, J., Paul, D., et al. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2308.01285 · Cited by 41
EDA-Debugger: An LLM-Based Framework for Automated EDA Runtime Issue Resolution
Li, J., Wong, S.-Z., Wan, G.-W., Wang, X., & Yang, J. L. (2025) · International Symposium on Quality Electronic Design (ISQED) · doi:10.1109/isqed65160.2025.11014463 · Cited by 15
“With the introduction of role-playing functions in agent communication [24], a comprehensive LLM-based multi-agent collaboration framework has demonstrated increased efficiency in handling complex tasks.”
DialogLab: Authoring, Simulating, and Testing Dynamic Human-AI Group Conversations
Hu, E., Chen, Y., Li, M., Phadnis, V., Xu, P., Qian, X., Olwal, A., Kim, D., Heo, S., & Du, R. (2025) · ACM Symposium on User Interface Software and Technology (UIST) · doi:10.1145/3746059.3747696 · Cited by 18
“Frameworks such as AutoGen [102] and Camel [58] demonstrate how multiple AI agents can collaborate autonomously on complex tasks through defined functional roles (e.g., Planner, Critic, Executor).”
Difficulty-Aware Hybrid Routing for Cost-Efficient Long-Context LLM Inference
Gu, Z., Xiao, Z., & Lin, O. (2026) · International Conference on Intelligent Communications and Computing (ICICC) · doi:10.1109/icicc71012.2026.11637729 · Cited by 5
Wang, N., Zou, D., Xie, H., & Qin, S. J. (2026) · Computers and Education Artificial Intelligence · doi:10.1016/j.caeai.2026.100653 · Cited by 4
SCREEN: A Benchmark for Situated Conversational Recommendation
Lin, D., Wang, J., Leong, C. T., & Li, W. (2024) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3664647.3681651 · Cited by 11
Signalling and social learning in swarms of robots
Cazenille, L., Toquebiau, M., Lobato‐Dauzier, N., Loi, A., Macabre, L., Aubert-Kato, N., Genot, A. J., & Bredèche, N. (2025) · Philosophical Transactions of the Royal Society A Mathematical Physical and Engineering Sciences · doi:10.1098/rsta.2024.0148 · Cited by 14
AssistEditor: Multi-Agent Collaboration for GUI Workflow Automation in Video Creation
Gao, D., Hu, S., Bai, Z., Lin, K. Q., & Shou, M. Z. (2024) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3664647.3684998 · Cited by 12
Mitigating Structural Hallucination in Large Language Models with Local DiffusionPreprint
Kiritani, K., & Kayano, T. (2024) · Research Square · doi:10.21203/rs.3.rs-4678127/v1 · Cited by 5
OASIS: Open Agent Social Interaction Simulations with One Million Agents
Yang, Z., Zhang, Z., Zheng, Z., Jiang, Y., Gan, Z., Wang, Z., Ling, Z., Chen, J., Ma, M., Dong, B., et al. (2024) · arXiv · doi:10.48550/arxiv.2411.11581 · Cited by 3
“Our agent module is based on large language models, and the core features of the agent module are inherited from CAMEL (Li et al., 2023).”
Self-citationSimulating Human-like Daily Activities with Desire-driven Autonomy
Wang, Y., Chen, Y., Zhong, F., Ma, L., & Wang, Y. (2024) · arXiv · doi:10.48550/arxiv.2412.06435 · Cited by 2
The Multi-Agentization of a Dual-Arm Nursing Robot Based on Large Language Models
Fang, C., Yue, X., Zhao, Z., & Guo, S. (2025) · Bioengineering · doi:10.3390/bioengineering12050448 · Cited by 4
“Studies, such as Camel [26] and Chateval [27], have demonstrated that multi-agent systems based on LLMs significantly outperform single-agent systems in terms of performance.”
Adaptive-TOD: An LLM-driven and adaptive agent for diverse interaction modes
Li, P., Yang, Q., Xu, S., Li, X., Li, Z., Chen, W., Liu, Y., Guo, T., Tang, J., & Wen, Y. (2025) · Neurocomputing · doi:10.1016/j.neucom.2025.130991 · Cited by 4
Exocortex Network for AI-Augmented Human-Led Scientific Expedition
Tsai, E. H. R., & Yager, K. G. (2025) · Photon Science · doi:10.1021/photonsci.5c00009 · Cited by 4
MOFReasoner: think like a scientist—a reasoning large language model via knowledge distillation
Bai, X., Zheng, Z., Zhang, X., Wang, H.-T., Yang, R., & Li, J. (2026) · Digital Discovery · doi:10.1039/d5dd00429b · Cited by 2
EpidemIQs: Prompt-to-Paper LLM Agents for Epidemic Modeling and Analysis
Samaei, M. H., Sahneh, F. D., Cohnstaedt, L. W., & Scoglio, C. (2026) · IEEE Transactions on Artificial Intelligence · doi:10.1109/tai.2026.3666830 · Cited by 9
“Two agent types are inspired by the generative agent concept [35] and further advanced by frameworks such as CAMEL [36], AutoGen [37], and Metagpt [38], and are designed to enable autonomous scientific reasoning and investigation.”
Quantifying Privacy Leakage in Multi-Agent LLMs: A Unified Theoretical and Empirical Analysis
Biswas, M., Yu, W., Wang, Q., & Liao, W. (2025) · IEEE International Conference on Big Data (BigData) · doi:10.1109/bigdata66926.2025.11401523 · Cited by 4
Kraidia, I., Qaddara, I., Almutairi, A., Alzaben, N., & Belhouari, S. B. (2026) · Scientific Reports · doi:10.1038/s41598-026-42705-7 · Cited by 18
“Several frameworks have formalized these interaction paradigms, including AutoGen 15 , CAMEL 16 , AgentVerse 17 , and MetaGPT 18 , which explore role-based communication, task decomposition, and iterative consensus-building among LLM agents.”
Does Socialization Emerge in AI Agent Society? A Case Study of Moltbook
Li, M., Li, X., & Zhou, T. (2026) · ACM Conference on AI and Agentic Systems (CAIS) · doi:10.1145/3786335.3813123 · Cited by 30
“Subsequent research explored structured interaction among multiple LLM agents, primarily aiming to improve task performance through coordinated discussion, role-based collaboration, and workflow orchestration [7, 9, 12, 18, 20, 24, 29, 30, 52, 53, 57].”
Designing Artificial Identity: The Identity Design Framework and Research Agenda
Bransky, K., Sweetser, P., Holthaus, P., Laban, G., Winkle, K., Akalın, N., Ashok, A., Lee, J., Khot, R., Bejarano, A., et al. (2026) · Designing Interactive Systems Conference · doi:10.1145/3800645.3812962 · Cited by 5
“…the group level, while role assignment and norm-following can be implemented through in-context learning and fine-tuning of LLMs [97, 131], future research can look at how agents can actively express group membership, adapt to dynamic roles, and interact fluidly with humans (e.g., [78, 130, 162]).”
Ying, T., Wang, L., Hu, P., & Ding, T. (2026) · International Conference on Electronic Technology and Information Science (ICETIS) · doi:10.1109/icetis70504.2026.11633437 · Cited by 7
A survey on mathematical reasoning and optimization with large language models
Forootani, A., Aliabadi, D. E., & Thrän, D. (2026) · Intelligent Systems with Applications · doi:10.1016/j.iswa.2026.200712 · Cited by 32
-, S. B., Ranjan, S., Adhitya, M., Suri, V., Hassija, V., Chamola, V., & Hussain, A. (2026) · Cognitive Computation · doi:10.1007/s12559-026-10619-1 · Cited by 10
“Other works emphasize evaluation benchmarks or multi-agent coordination in isolation [24, 27, 38, 39].”
InfluentialLi, H., Gong, R., & Jiang, H. (2025) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v39i1.32017 · Cited by 24
MemEngine: A Unified and Modular Library for Developing Advanced Memory of LLM-based Agents
Zhang, Z., Dai, Q., Chen, X., Li, R., Li, Z., & Dong, Z. (2025) · Companion Proceedings of the ACM on Web Conference 2025 · doi:10.1145/3701716.3715299 · Cited by 2
“Specifically, AutoGen , MetaGPT [3], CAMEL [4], AgentScope [2], LangChain , AgentLite [5], CrewAI , AutoGPT , and AgentVerse [1] are prominent open-source libraries for building LLM-based AI agent systems.”
Generative AI in Embodied Systems: System-Level Analysis of Performance, Efficiency and Scalability
Wan, Z.-S., Qian, J., Du, Y.-H., Jabbour, J., Du, Y.-L., Zhao, Y., Raychowdhury, A., Krishna, T., & Reddi, V. J. (2025) · IEEE International Symposium on Performance Analysis of Systems and Software · doi:10.1109/ispass64960.2025.00013 · Cited by 5
“The communication module accesses relevant data from memory (e.g., environmental maps, task progress, agent states, and past interactions) and generates appropriate messages for communication with other agents in multi-agent systems [22]–[24].”
Prompt Sapper: LLM-Empowered Software Engineering Infrastructure for AI-Native ServicesPreprint
Xing, Z., Huang, Q. Y., Cheng, Y., Zhu, L., Lu, Q., & Xu, X. (2023) · doi:10.36227/techrxiv.23298488 · Cited by 2
Zhang, G., Li, G., Li, H., Su, Y., & Li, Y. (2024) · doi:10.2196/preprints.65470 · Cited by 2
Self-citationLeong, H. Y., & Wu, Y. (2025) · SSRN Electronic Journal · doi:10.2139/ssrn.5276004 · Cited by 14
Symphony: Edge-powered Decentralized Multi-agent Framework for Autonomous Co-evolving IntelligencePreprint
Wang, J., Feng, Y., Ai, L., & Shi, B. (2025) · SSRN Electronic Journal · doi:10.2139/ssrn.5361139 · Cited by 2
Branda, F., Ahmed, M. M., Ciccozzi, M., Guzzi, P. H., & Scarpa, F. (2026) · Briefings in Bioinformatics · doi:10.1093/bib/bbag245 · Cited by 2
CitySim: Modeling Urban Behaviors and City Dynamics with Large-Scale LLM-Driven Agent Simulation
Bougie, N., & Watanabe, N. (2025) · arXiv · doi:10.48550/arxiv.2506.21805 · Cited by 8
“Numerous studies have pointed out that after being empowered by LLMs, agents gain the ability to reason, plan, and interact through natu-ral language (Gao et al., 2024; Wei et al., 2022; Bougie and Watanabe, 2025; Li et al., 2023; Park et al., 2023a; Bougie and Watanabe, 2024).”
MedSentry: Understanding and Mitigating Safety Risks in Medical LLM Multi-Agent Systems
Chen, K., Zhen, T., Wang, H., Liu, K., Li, X., Huo, J., Yang, T., Xu, J., Dong, W., & Gao, Y. (2025) · arXiv · doi:10.48550/arxiv.2505.20824 · Cited by 32
“However, this framework does not consider multi-agent collaborative benchmarks involving LLMs acting as both patients and healthcare providers [35, 36].”
A Review of Multi-Agent AI Systems for Biological and Clinical Data Analysis
Spieser, J., Balapour, A., Meller, J., Patra, K. C., & Shamsaei, B. (2026) · Methods and Protocols · doi:10.3390/mps9020033 · Cited by 5
“These provide templates for common MAS architectures (e.g., collaborative coding agents, conversational QA teams) [13].”
Flora-RAG: Enhancing Conversational AI with Retrieval Augmented Generation for Floriculture
Shelley, S., Kaur, P., Aggarwal, G., Kaushal, A., & Dutta, M. K. (2025) · International Conference on Educational Technology Management (ICETM) · doi:10.1109/icetm63734.2025.11051979 · Cited by 1
Prosocial behavior in Large Language Models: Value alignment and affective mechanisms
Liu, H., Lei, Y., & Wu, Z. (2025) · Science China Technological Sciences · doi:10.1007/s11431-025-2977-0 · Cited by 8
Jiang, T., Wu, S., Yang, S., Yuan, X., Gu, L., & Wu, X. (2025) · Expert Systems with Applications · doi:10.1016/j.eswa.2025.130496 · Cited by 4
EMRDCM : An experience-enhanced multi-role debate method based on compromise mechanisms
Zhang, Y., Feng, J., Yiming, X., Wang, Z., Yang, T., & Wang, X. (2026) · Journal of Intelligent Information Systems · doi:10.1007/s10844-026-01022-1 · Cited by 1
Yu, J. S., & Yao, Y. (2026) · Springer eBooks · doi:10.1007/978-981-95-4632-9_5 · Cited by 3
SIA: A Framework for Context-Aware Intent Clarification in Speech-Driven Immersive Analytics
Song, H., Whitley, K., Krokos, E., & Varshney, A. (2026) · International Conference on Intelligent User Interfaces · doi:10.1145/3742413.3789063 · Cited by 3
Fang, W., Li, L., Shi, K., Yang, Y., & Zhang, J. (2026) · International World Wide Web Conference (WWW) · doi:10.1145/3774904.3792976 · Cited by 2
RPGAgent: Driving Coherent Story-to-Play Generation with an LLM-Based Multi-Agent System
Zhang, S., Xiao, Y., Ma, R., & Leung, C.-S. (2026) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3772318.3790326 · Cited by 6
“To address these complexities, a particularly promising paradigm has emerged: the organization of LLMs into multi-agent systems, where specialized agents collaborate synergistically to tackle multifaceted tasks [23, 37].”
InfluentialKhan, M. S., Khan, M. J., Sharif, M., & Yasmin, S. (2026) · IEEE Access · doi:10.1109/access.2026.3683900 · Cited by 1
“Specialized systems such as CAMEL [13], ChatDev [14], and Generative Agents [3] demonstrate impressive emergent behaviors in specific domains such as software development and social simulation.”
Yue, L., Ko, C.-Y., Chen, P.-Y., Di, S., & Pan, S. (2026) · Frontiers in Artificial Intelligence · doi:10.3389/frai.2026.1820375 · Cited by 8
“Recent advances in large language models (LLMs) have enabled agents that can interleave reasoning with tool use (Yao et al., 2022; Schick et al., 2023) and collaborate via multi-agent conversation and orchestration frameworks (Wu et al., 2023; Li et al., 2023; Hong et al., 2023).”
Huang, N. (2025) · International Conference on Big Data Economy and Information Management (BDEIM) · doi:10.1145/3800000.3800146 · Cited by 1
“We integrate a Dynamic Knowledge Graph Evolution mechanism that continuously updates entity relationships while ensuring temporal validity [7].”
Is My Vision-Language Data in Your AI? Membership Inference Test (MINT) Demo 2
DeAlcala, D., Mancera, G., Fiérrez, J., Morales, A., Tolosana, R., & Vera-Rodríguez, R. (2026) · IEEE Annual International Computer Software and Applications Conference (COMPSAC) · doi:10.1109/compsac69091.2026.00323 · Cited by 10
Internet of Agentic AI: Incentive-Compatible Distributed Teaming and Workflow
Yang, Y.-T., & Zhu, Q. (2026) · International Symposium on Modeling and Optimization in Mobile, Ad Hoc, and Wireless Networks (WiOpt) · doi:10.23919/wiopt71098.2026.11568240 · Cited by 9
Agentic AI Safety: A Structured Review of Open Problems and Their Regulatory Anchoring
Valenta, T., Rozinek, O., & Horálek, J. (2026) · AI · doi:10.3390/ai7080298 · Cited by 2
Constructive Specification for Plug-and-Play Learnware Agents
Liu, J.-D., Zhao, Z.-C., Sun, H., Wu, L.-X., Zhang, H., Wang, P., Zhao, M., Chu, X., Yan, S., Zhu, Y., et al. (2026) · ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) · doi:10.1145/3770855.3817996 · Cited by 2
Wu, Z., Xiao, F., Li, S., Peng, J., Li, H., & Li, H. (2026) · Building Simulation · doi:10.1007/s12273-026-1467-3 · Cited by 1
“The results of this model have been experimentally verified, and both models were constructed based on applicable standard methods in the literature (Wang et al. 2024a; Zhang et al. 2025a) and have also been verified in different climate zones (Li et al. 2023b; Li et al. 2024a; Li et al. 2026).”
Intelligent Agents for Smart Agriculture: Architectures, Applications, and Future Challenges
Tao, W., Sang, Q., Chen, C., & Mao, Q. (2026) · Agriculture · doi:10.3390/agriculture16171808 · Cited by 2
Mitrović, S., Giuffrida, V., Rosselli, M., Bosetti, M., & Salani, M. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-39395-1_23 · Cited by 1
ConfidenceCal: Enhancing LLMs Reliability through Confidence Calibration in Multi-Agent Debate
Bai, Y. (2024) · International Conference on Big Data and Information Analytics (BigDIA) · doi:10.1109/bigdia63733.2024.10808396 · Cited by 10
KAOS: Large Model Multi-agent Operating System
Zhuo, Z., Li, R., Liu, K., Zou, H., Li, K., Yu, J., Sun, T., & Wu, Q. (2025) · Communications in computer and information science · doi:10.1007/978-981-96-1809-5_27 · Cited by 10
Understanding Agent Scaling in LLM-Based Multi-Agent Systems via Diversity
Yang, Y., Qu, C., Wen, M., Shi, L., Wen, Y., Zhang, W., Wierman, A., & Gu, S. (2026) · arXiv · doi:10.48550/arxiv.2602.03794 · Cited by 1
On the Resilience of LLM-Based Multi-Agent Collaboration with Faulty Agents
Huang, J.-T., Zhou, J., Jin, T., Zhou, X., Chen, Z., Wang, W., Yuan, Y., Lyu, M. R., & Sap, M. (2024) · arXiv · doi:10.48550/arxiv.2408.00989 · Cited by 1
“We select six multi-agent collaboration systems that represent three classical human organizational structures: Linear (Hong et al., 2024; Dong et al., 2024), Flat (Li et al., 2023; Wang et al., 2024c), and Hierarchical (Chen et al., 2024b; Liang et al., 2024).”
Effective Strategies for Asynchronous Software Engineering Agents
Geng, J., & Neubig, G. (2026) · arXiv · doi:10.48550/arxiv.2603.21489 · Cited by 23
“Early frameworks such as CAMEL [Li et al., 2023] and Generative Agents [Park et al., 2023] established the foundation for communicative interaction, which was later structured into a natural language communication pipeline from ChatDev [Qian et al., 2024a].”
Hippocampus: An Efficient and Scalable Memory Module for Agentic AI
Li, Y., Cao, L., Ahmed, F., Sharma, P., & Li, B. (2026) · arXiv · doi:10.48550/arxiv.2602.13594
"Theater of Mind" for LLMs: A Cognitive Architecture Based on Global Workspace Theory
Shang, W. (2026) · arXiv · doi:10.48550/arxiv.2604.08206
“Role-playing methodologies, such as CAMEL [9], alongside communicative frameworks like AutoGen [10] and MetaGPT [11], facilitate decentralized task execution through sequential message passing.”
Chen, Y. (2026) · arXiv · doi:10.48550/arxiv.2606.14923 · Cited by 2
“A growing body of work places LLM agents in cooperative and competitive settings [14, 8, 18, 17], including repeated economic games [1], negotiation [13], and multi-agent debate in which agents cross-check one another [6, 10].”
Verifying Intent and Harm: A Unified Defense Against LLM-Generated Threats
Thota, P., Lei, Y., Thangaraj, S., Jonnalagadda, S. R., & Nilizadeh, S. (2026) · arXiv · doi:10.48550/arxiv.2606.26377
“Multi-agent frameworks have demonstrated strong performance in collaborative reasoning, debate, coding, and planning [28], [29], [30], [31], [32], [33].”
Memory for Autonomous LLM Agents:Mechanisms, Evaluation, and Emerging Frontiers
Du, P. (2026) · arXiv · Cited by 1
TokenDance: Scaling Multi-Agent LLM Serving via Collective KV Cache Sharing
Bian, Z., Wu, F., Zhang, C., Dong, H., Liang, Y., & Zhuo, Y. (2026) · arXiv · doi:10.48550/arxiv.2604.03143
SEMA-RAG: A Self-Evolving Multi-Agent Retrieval-Augmented Generation Framework for Medical Reasoning
Huang, Y., Chen, R., & Cheng, J. (2026) · Findings of the Association for Computational Linguistics: ACL 2026 · doi:10.18653/v1/2026.findings-acl.917
“Multi-agent systems enhance complex-task solving through role specialization and coordination (Guo et al., 2024; Zong et al., 2024; Tran et al., 2025) (e.g., CAMEL (Li et al., 2023), MetaGPT (Hong et al., 2024)).”
MAPLE-Guard: Memory-Aware Link Enforcement Against Memory-Link Poisoning in Multi-Agent Systems
Xiong, W., Zhou, Y., Wang, J., Gu, S., Tang, B., Li, Z., Xiong, F., Wen, Y., & Wen, M. (2026) · arXiv · doi:10.48550/arxiv.2608.00426 · Cited by 2
“LLM-based MAS coordinate agents through roles, workflows, and communication [15, 22, 34, 49].”
Shen, X., Liu, Y., Dai, Y., Wang, Y., Miao, R., Tan, Y., Pan, S., & Wang, X. (2025) · arXiv · doi:10.48550/arxiv.2505.23352 · Cited by 6
“Agents powered by large language models (LLMs) (Li et al., 2023; Wang et al., 2024a; Xi et al., 2025) have demonstrated strong performance across tasks like reasoning (Yao et al., 2023), code generation (Zhang et al., 2024b), and complex decision-making (Guo et al., 2024b).”
WebAgent-R1: Training Web Agents via End-to-End Multi-Turn Reinforcement Learning
Wei, Z., Yao, W., Liu, Y., Zhang, W., Lu, Q., Qiu, L., Yu, C., Xu, P., Zhang, C., Yin, B., et al. (2025) · arXiv · doi:10.48550/arxiv.2505.16421
“LLMs have demonstrated promising agentic capabilities, such as breaking down complex tasks into manageable subgoals and reasoning over long horizons (Zhou et al., 2022; Huang et al., 2022; Madaan et al., 2022; Li et al., 2023a,b; Wu et al., 2024; Liu et al., 2025b; Chu et al., 2025).”
Multi-agent Architecture Search via Agentic Supernet
Zhang, G., Niu, L., Fang, J., Wang, K., Bai, L., & Wang, X. (2025) · arXiv · doi:10.48550/arxiv.2502.04180
EvoFlow: Evolving Diverse Agentic Workflows On The Fly
Zhang, G., Chen, K., Wan, G., Chang, H., Cheng, H., Wang, K., Hu, S., & Bai, L. (2025) · arXiv · doi:10.48550/arxiv.2502.07373
MS-Eval: Benchmarking Sub-Agent Delegation and Orchestration in MCP-Intensive Working Scenarios
Ding, B., Wang, Q., Ruan, L., Lu, W., Ge, Q., Fang, R., & Lin, T. (2026) · arXiv · doi:10.48550/arxiv.2610.22869
Decentralized Multi-Agent Systems with Shared Context
Mao, Y., Gu, J., Chauhan, A., Zhang, Q., Kang, H., & Mirhoseini, A. (2026) · arXiv · doi:10.48550/arxiv.2606.10662
“Early frameworks such as CAMEL (Li et al., 2023) instantiate role-playing agents to study autonomous cooperation.”
Liu, Z., Li, G., Shi, B., Zhang, X., Ge, J., Wu, Y., & Lyu, H. (2024) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn60899.2024.10650172 · Cited by 3
“Furthermore, CAMEL [5], leveraging role-playing prompts, guides agents in conducting conversation tasks, ensuring their alignment with human intentions.”
BackgroundSelf-citationLeveraging Generative AI for Accelerating Enterprise Application Development: Insights from ChatGPT
Rajbhoj, A., Sant, T., Somase, A., & Kulkarni, V. (2024) · Asia-Pacific Software Engineering Conference (APSEC) · doi:10.1109/apsec65559.2024.00052
“[23].”
The Survey for a Hierarchical Resource Management Framework Enabled with LLM-based Agents
Chan, D., Wu, Y., Bai, D., Mak, T., Chan, T., XinXu, & Qiu, T. (2024) · IEEE Future Networks World Forum (FNWF) · doi:10.1109/fnwf63303.2024.11028788
Large Language Model Based Agents for Generating Industrial Dashboards
Kimmel, R., Pfeiffer, J., Zhang, J., & Wortmann, A. (2025) · Lecture notes in production engineering · doi:10.1007/978-3-031-86893-1_57
Towards the Versioning of LLM-Agent-Based Software
Liu, C., Zhang, L., Xu, X., Guo, W., & Liu, Y. (2025) · ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE) · doi:10.1145/3696630.3728714 · Cited by 4
ASTRA: Agentic Steerability and Risk Assessment Framework
Hazan, I., Mathov, Y., Shtar, G., Bitton, R., & Mantin, I. (2026) · Lecture notes in networks and systems · doi:10.1007/978-3-032-32732-1_21 · Cited by 3
EvoRL: Evolution of Agentic Workflows Using Reinforcement Learning
Attota, D. C., & Xie, Y. (2025) · IEEE International Conference on Big Data (BigData) · doi:10.1109/bigdata66926.2025.11401641
“For instance, CAMEL [21] introduced role-playing, assigning agents specific personas to guide their actions during task-oriented dialogues.”
Agentic AI Framework for Low-Resource Essay Evaluation via Scoring, Explanation, and Debate
Thapa, S., Rauniyar, K., Shiwakoti, S., Adhikari, S., Rashid, J., Kim, J., & Naseem, U. (2025) · IEEE International Conference on Big Data (BigData) · doi:10.1109/bigdata66926.2025.11402339 · Cited by 1
“Frameworks like AutoGen [31], CAMEL [11], and MetaGPT [6] demonstrated how multiple LLMs, assigned different roles, could communicate and coordinate to complete complex tasks collectively.”
Enhancing Role-Playing Language Models via Similar-Scene Generation
Yan, H. (2026) · International Conference on Electrical Automation and Artificial Intelligence (ICEAAI) · doi:10.1109/iceaai68945.2026.11442520
“Meanwhile, Li et al. [21] propose CAMEL, which investigates how to enhance collaboration among chat agents via inception prompting, guiding them to accomplish tasks while preserving human intent.”
Construction and Application of University Agent Space Based on Ecosystem Theory
Wang, M., Fang, H., Fang, W., & Zeng, Z. (2025) · International Conference on Artificial Intelligence and Teacher Education (ICAITE) · doi:10.1109/icaite68636.2025.11442537
“CAMEL, the first multi-agent framework, has shown the advantages of multi-agent system in collaborative learning, natural language communication and so on [9].”
Improving Cooperation in Multi-agent Deep Reinforcement Learning for Mobile Robotics
Toquebiau, M. (2024) · theses.fr (ABES) · doi:10.70675/3385d0bbzc66cz4fceza9e0z8696ff61ff4e
Foundation Vision-Language models
Laurençon, H. (2025) · theses.fr (ABES) · doi:10.70675/ec0311a0zb6a4z4249zb9fbz01bf51a02c7c · Cited by 1
PoliSim@CHI 2026: LLM Agent Simulation for Policy
Li, Y., Deng, W. H., Zhou, X., Klyman, K., Yu, C., Shi, Y., Vincent, N., Zhang, A. X., Sap, M., Das, S., et al. (2026) · CHI Conference Extended Abstracts on Human Factors in Computing Systems (CHI EA) · doi:10.1145/3772363.3778738
“these agents exhibit emergent social phenomena such as coalition formation, information diffusion, and collective decision-making [4, 25, 48, 58, 59].”
Zhang, Z., Yi, X., Dai, Z., Yu, X., Zhang, S., Liu, B., Cao, J., & Zhao, H. (2026) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3772318.3791926 · Cited by 2
“…demand predefined agent roles and processes, operate through closed internal dialogues [52], offer limited transparency or opportunities for user intervention, and often produce static outputs with limited opportunities for users to refine or steer plans once execution begins [34, 51, 61].”
Zeng, R., Zhang, X., Miao, D., & Qi, Y. (2026) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp55912.2026.11462454
LLM-based multi-agent systems for code generation: A multi-vocal literature review
Rasheeda, Z., Waseema, M., Kemella, K.-K., Saari, M., & Abrahamsson, P. (2026) · Journal of Systems and Software · doi:10.1016/j.jss.2026.113142 · Cited by 3
Zablith, F., & Abbas, J. (2026) · Information Systems Frontiers · doi:10.1007/s10796-026-10738-y · Cited by 1
Wu, J. (2026) · International Symposium on Big Data and Applied Statistics (ISBDAS) · doi:10.1109/isbdas69350.2026.11484104
Temporal Knowledge Graph-Driven Multi-Agent Collaborative Retrieval Augmented Generation Framework
Niu, Y., Ai, Y., Wan, X., Gao, S., Zhao, Y., Kuang, J., & Liu, J. (2025) · China Automation Congress (CAC) · doi:10.1109/cac67268.2025.11487625
“Later frameworks such as ReAct [16] and CAMEL [17] formalize Planner, Tool-Caller, and Critic roles that converse in thought-action loops, markedly boosting multi-step reasoning and tool use.”
HOLA: Hierarchical OODA-LLM Agent Architecture With Commander-Like Cognition
Kuang, X., Zeng, D., Li, G., Yang, F., & Li, X. (2026) · IEEE Transactions on Games · doi:10.1109/tg.2026.3689572 · Cited by 1
Self-citationLLM-Based Multi-Agent Method for Compliant Generation of Electrical Bonding Processes
Zhang, Q., & Ling, W. (2026) · International Conference on Generative Artificial Intelligence and Digital Media · doi:10.1145/3803730.3803749
Metadata
- Authors
- Li, G., Hammoud, H., Itani, H., Khizbullin, D., & Ghanem, B.
- Venue
- Advances in Neural Information Processing Systems 36
- Published
- 2023
- Pages
- 51991-52008
- ISSN
- 1049-5258 (ISSN-L)
- OpenAlex
- W7133220405
- Cited by
- 314
- References
- 0
- Citations by year
- 2026: 197 · 2025: 77 · 2024: 32 · 2023: 8
- FWCI
- 78.60
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Licence
- cc-by
- Locations
- neural information processing systems (publishedVersion)
- arXiv (Cornell University) (submittedVersion, cc-by)
- arXiv (Cornell University) (cc-by)
- Funders
- King Abdullah University of Science and Technology
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Social Sciences › Arts and Humanities › History and Philosophy of Science › Cybernetics and Technology in Society
- Physical Sciences › Computer Science › Artificial Intelligence › Multi-Agent Systems and Negotiation
- Keywords
- large language models, multi-agent systems, agent communication, emergent behavior, collective intelligence
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.