Self-Evolving Agents

Follow this topic Ask about this topic

Self-evolving agents are systems that use experience from task execution to change how they solve future tasks, rather than relying only on a fixed model or hand-designed workflow. Common approaches store and refine reusable skills or memories, optimize prompts and agent harnesses, train policies with reinforcement learning, or co-evolve task generators, environments, and teams of agents. These methods are tested in settings such as web and GUI use, coding, embodied robotics, scientific workflows, and reasoning benchmarks, with feedback ranging from execution outcomes to verifiers and paired validation. A central challenge is making improvements transfer to new tasks without forgetting earlier abilities or introducing unsafe behavior; reliable evaluation, update verification, and long-term monitoring remain open problems.

Latest papers

427 papers104 papers in the last four weeks, up 148% on the four weeks before. 0.6% of all new papers.

On-Policy Self-Evolution via Failure Trajectories for Agentic Safety Alignment
May 12, 2026Bo Yin, Qi Li, Xinchao WangAgentic RL LLM Safety Alignment
National University of Singapore
EVOCHAMBER: Test-Time Co-evolution of Multi-Agent System at Individual, Team, and Population Scales
May 11, 2026Yaolun Zhang, Tianyi Xu, Shengyu Dai +3Multi-Agent LLM Systems Multi-Agent Collaboration
1Oregon State University · 5AG2AI, Inc. · University of Wisconsin–Madison +2
Evolving-RL: End-to-End Optimization of Experience-Driven Self-Evolving Capability within Agents
May 11, 2026Zhiyuan Fan, Wenwei Jin, Feng Zhang +4Reinforcement Learning Language Model Self-Improvement
1Xiaohongshu Inc. · School of Computer Science, Peking University
EmbodiSkill: Skill-Aware Reflection for Self-Evolving Embodied Agents
May 11, 2026Ruofei Ju, Xinrui Wang, Xin Ding +12LLM Agent Skill Learning Agent Skill Learning
Nanjing University · Institute for AI Industry Research (AIR), Tsinghua University · Huazhong University of Science and Technology +2
MAGE: Multi-Agent Self-Evolution with Co-Evolutionary Knowledge Graphs
May 11, 2026Ruiyi Yang, Zechen Li, Hao Xue +2Agent Memory Continual Learning for LLM Agents
University of New South Wales · The Hong Kong University of Science and Technology (Guangzhou) · Mohamed Bin Zayed University of Artificial Intelligence
Swarm Skills: A Portable, Self-Evolving Multi-Agent System Specification for Coordination Engineering
May 11, 2026Xinyu Zhang, Zhicheng Dou, Deyang Li +10Multi-Agent Coordination Multi-Agent Orchestration
openJiuwen Team & Gaoling School of Artificial Intelligence, Renmin University of China
TacoMAS: Test-Time Co-Evolution of Topology and Capability in LLM-based Multi-Agent Systems
May 10, 2026Chen Xu, Yicheng Hu, Ruizi Wang +4Multi-Agent LLM Systems Multi-Agent System Optimization
1Carnegie Mellon University · University of Science and Technology of China · 3National University of Singapore +1
SimWorld Studio: Automatic Environment Generation with Evolving Coding Agent for Embodied Agent Learning
May 10, 2026Haoqiang Kang, Xiaokang Ye, Yuhan Liu +5Embodied Navigation AI Coding Agents
1UC San Diego · 2New York University
SkillMAS: Skill Co-Evolution with LLM-based Multi-Agent System
May 10, 2026Shuai Pan, Yixiang Liu, Jiaye Gao +7Multi-Agent System Optimization LLM Agent Skill Learning
Shanghai Jiao Tong University · Central South University · OPPO
Do Self-Evolving Agents Forget? Capability Degradation and Preservation in Lifelong LLM Agent Adaptation
May 10, 2026Ye Yu, Xiaopeng Yuan, Haibo Jin +3Continual Learning for LLM Agents LLM Agent Self-Improvement
University of Illinois Urbana-Champaign, IL, USA
Evolutionary Ensemble of Agents
May 9, 2026Zongmin Yu, Liu YangAI Agents for Scientific Discovery Evolutionary Optimization
National University of Singapore
Ace-Skill: Bootstrapping Multimodal Agents with Prioritized and Clustered Evolution
May 9, 2026Feng Xiong, Zengbin Wang, Yong Wang +5Continual Learning for LLM Agents Agent Skill Learning
1AMAP, Alibaba Group · 2CASIA · 3BNU
FlashEvolve: Accelerating Agent Self-Evolution with Asynchronous Stage Orchestration
May 8, 2026Zhengding Hu, Mingge Lu, Zhen Wang +8LLM Agent Orchestration LLM Agent Workflow Optimization
University of California, San Diego · Georgia Institute of Technology
Rethinking Experience Utilization in Self-Evolving Language Model Agents
May 8, 2026Weixiang Zhao, Yingshuo Wang, Yichen Zhang +6Self-Evolving Agents Memory-Augmented Language Model Agents
1Harbin Institute of Technology · 2Huawei Technologies Co., Ltd
GEAR: Genetic AutoResearch for Agentic Code Evolution
May 8, 2026Ahmadreza Jeddi, Minh Ngoc Le, Hakki C. Karaimer +2Evolutionary Optimization Autonomous Experimentation
University of Toronto · Vector Institute · AI Center-Toronto, Samsung Electronics +1
SkillOS: Learning Skill Curation for Self-Evolving Agents
May 7, 2026Siru Ouyang, Jun Yan, Yanfei Chen +13Continual Learning for LLM Agents LLM Agent Skill Learning
University of Illinois Urbana-Champaign · Google Cloud AI Research · Massachusetts Institute of Technology
Safactory: A Scalable Agentic Infrastructure for Training Trustworthy Autonomous Intelligence
May 7, 2026Xinquan Chen, Zhenyun Yin, Shan He +38Agentic RL AI Agent Safety
Knowledge-Graph Paths as Intermediate Supervision for Self-Evolving Search Agents
May 7, 2026Huyu Wu, Jun Liu, Xiaochi Wei +3Multi-Hop QA Reward Shaping
Xiaohongshu Inc., Beijing, China
Autonomous Drift Learning in Data Streams: A Unified Perspective
May 2, 2026Xiaoyu Yang, En Yu, Jie LuConcept Drift Continual Learning
Australian Artificial Intelligence Institute (AAII), Faculty of Engineering and Information Technology, University of Technology Sydney, Australia
Self-Evolving Software Agents
Apr 29, 2026Marco Robol, Paolo GiorginiSoftware Engineering Agents LLM Agent Self-Improvement
University of Trento · Trento, Italy
CurEvo: Curriculum-Guided Self-Evolution for Video Understanding
Apr 29, 2026Guiyi Zeng, Junqing Yu, Yi-Ping Phoebe Chen +3Video QA Curriculum Learning
Huazhong University of Science and Technology Wuhan, China · La Trobe University Melbourne, Australia · Beijing Institute of Computer Technology and Applications Beijing, China
Agentic Harness Engineering: Observability-Driven Automatic Evolution of Coding-Agent Harnesses
Apr 28, 2026Jiahang Lin, Shichun Liu, Chengjun Pan +8AI Coding Agents Agent Harness Optimization
Fudan University · Shanghai Qiji Zhifeng Co., Ltd · Peking University
OxyGent: Making Multi-Agent Systems Modular, Observable, and Evolvable via Oxy Abstraction
Apr 28, 2026Junxing Hu, Tianlong Li, Lei Yu +1Multi-Agent Orchestration AI Agent Monitoring
JD.com Beijing, China
Escher-Loop: Mutual Evolution by Closed-Loop Self-Referential Optimization
Apr 25, 2026Ziyang Liu, Xinyan Guo, Xuchen Wei +2Evolutionary Optimization Self-Improving Agents
Shenzhen X-Institute; Soochow University · Soochow University · Shenzhen X-Institute +2
The Last Harness You'll Ever Build
Apr 22, 2026Haebin Seong, Li Yin, Haoran Zhang +1Meta-Learning Agent Harness Optimization
EvoAgent: An Evolvable Agent Framework with Skill Learning and Multi-Agent Delegation
Apr 22, 2026Aimin Zhang, Jiajing Guo, Fuwei Jia +3Multi-Agent LLM Systems LLM Agent Skill Learning
Focus AI Center, Focus Technology Co., Ltd. · Engineering Research Center of Digital Forensics, Ministry of Education, Nanjing University of Information Science and Technology · Nanjing University of Science and Technology
A Self-Evolving Framework for Efficient Terminal Agents via Observational Context Compression
Apr 21, 2026Jincheng Ren, Siwei Wu, Yizhi Li +8Terminal Agents Long-Horizon Agent Tasks
University of Manchester · 2MAP · 4HKUST(GZ) +2
Forage V2: Knowledge Evolution and Transfer in Autonomous Agent Organizations
Apr 21, 2026Huaqing XieLLM Agent Evaluation Memory-Augmented Agents
Independent Researcher
Agent-World: Scaling Real-World Environment Synthesis for Evolving General Agent Intelligence
Apr 20, 2026Guanting Dong, Junting Lu, Junjie Huang +17Lifelong Learning Agents AI Agent Benchmarks
Renmin University of China · ByteDance Seed
Training LLM Agents for Spontaneous, Reward-Free Self-Evolution via World Knowledge Exploration
Apr 20, 2026Qifan Zhang, Dongyang Ma, Tianqing Fang +5Intrinsic Motivation LLM Agent Self-Improvement
1Tencent · 2The Hong Kong University of Science and Technology (Guangzhou)

May 12, 2026J/K move · Enter open · S save

On-Policy Self-Evolution via Failure Trajectories for Agentic Safety Alignment

Bo Yin, Qi Li, Xinchao Wang

National University of Singapore

Agentic RL LLM Safety Alignment LLM Agent Safety Self-Evolving Agents

Open paper AskPDF arXiv

Tool-using LLM agents fail through trajectories rather than only final responses, as they may execute unsafe tool calls, follow injected instructions, comply with harmful requests, or over-refuse benign tasks despite producing a seemingly safe answer. Existing safety-alignment signals are largely response-level or off-policy, and often incur a safety-utility trade-off: improving agent safety comes at the cost of degraded task performance. Such sparse and single-objective rewards severely limit real-world usability. To bridge this gap, we propose FATE, an on-policy self-evolving framework that transforms verifier-scored failures into repair supervision without expert demonstrations. For each failure, the same policy proposes repair candidates, which are then re-scored by verifiers and filtered across security, utility, over-refusal control, and trajectory validity. This dense trajectory-level information is then used as a supervision signal for agent self-evolution. During this process, we further introduce Pareto-Front Policy Optimization (PFPO), combining supervised warmup with Pareto-aware policy optimization to preserve safety-utility trade-offs. Experiments on AgentDojo, AgentHarm, and ATBench show that FATE improves safety across different models and scales while preserving useful behavior. Compared with strong baselines, FATE reduces attack success rate by 33.5%, harmful compliance by 82.6%, and improves external trajectory-safety diagnosis by 6.5%. These results suggest that failed trajectories can provide structured repair supervision for safer self-evolving agents.