Background

Large language model (LLM)-based agents are rapidly emerging as a major research frontier in Artificial Intelligence (AI). By integrating foundation models with reasoning, planning, memory, tool use, perception, and environmental interaction, LLM-based agents extend conventional language models from passive content generators to autonomous systems capable of pursuing goals and completing complex, long-horizon tasks. Recent advances have demonstrated their potential in software engineering, scientific discovery, information retrieval, data analysis, embodied intelligence, and human-AI collaboration. Meanwhile, multi-agent systems are enabling new forms of distributed problem solving, collaborative decision-making, debate, and collective intelligence. In recognition of the transformative potential of LLM-based agents, FCS is organizing a special issue titled “Frontiers of Large Language Model-Based Agents.” This special issue aims to showcase recent advances, consolidate emerging research directions, and provide forward-looking perspectives on the foundational theories, enabling technologies, system infrastructures, trustworthy evaluation, and real-world applications of large language model-based agents.

Topics of Interest

We invite high-quality submissions presenting original research, comprehensive reviews, practical systems, benchmark studies, and forward-looking perspectives on large language model-based agents. Topics of interest include, but are not limited to:

· Reasoning, planning, long-term memory, context management, tool learning, function calling, retrieval-augmented generation, cognitive architectures, and emergent agent capabilities

· Multi-agent communication, negotiation, role assignment, task orchestration, debate, game-theoretic interaction, collective intelligence, scalability, stability, and collaboration evaluation

· Reinforcement learning and preference optimization for agents, continual learning, self-reflection, self-improvement, experience-driven policy evolution, and mitigation of catastrophic forgetting

· GUI, Web, and operating-system agents, embodied agents, world models, perception–planning–action loops, spatial reasoning, navigation, human–agent collaboration, and human-in-the-loop systems

· Agent workflow orchestration, context engineering, inter-agent communication and interoperability protocols, multi-agent runtimes, scheduling, efficient inference and serving, evaluation platforms, and benchmarks

· Adversarial robustness, prompt-injection defense, least-privilege access control, privacy protection, alignment, controllability, behavioral safety, boundary-violation detection, interpretability, auditability, and process-aware safety evaluation

· AI scientists and autonomous scientific discovery, coding and software engineering agents, data analysis and deep-research agents, and domain-specific agents for finance, healthcare, industry, education, and other real-world applications

 Important Dates

· Submission Deadline: March 31, 2027

· First-round Review Decisions: April 30, 2027

· Deadline for Revision Submissions: May 31, 2027

· Notification of Acceptance: June 30, 2027

· Final Manuscript Due: July 31, 2027

 Guest Editors

- Wayne Xin Zhao, Renmin University, China, zhaoxin.ruc@gmail.com

- Tong Xu, University of Science and Technology of China, China, tongxu@ustc.edu.cn

- Xiaocheng Feng, Harbin Institute of Technology, China, xcfeng@ir.hit.edu.cn

 Online Submission

Please submit your manuscript through the online system at:
http://mc.manuscriptcentral.com/hepfcs
Select manuscript type: “Large Language Model Agents: Foundations, Collaboration, and Trustworthy Evolution”.
Templates and additional submission guidelines are available on the submission website.