Self-Improving Agents

Follow this topic Ask about this topic

Self-improving agents study how systems can learn from their own actions and outcomes to perform better on later tasks, from coding and web use to robotics and long-horizon games. Common approaches turn successes and failures into reusable memories or executable skills, revise prompts and agent harnesses, or update model parameters through fine-tuning and reinforcement learning. Work increasingly uses repeated trials, held-out tasks, execution-based checks, and external evaluators to test whether improvements transfer rather than merely exploit a score. The central tension is that feedback and self-authored tests can be unreliable, so genuine progress must be balanced against regression, reward hacking, safety risks, and the cost of verification.

Latest papers

171 papers57 papers in the last four weeks, up 171% on the four weeks before. 0.3% of all new papers.

Practice Makes Unsafe: Skill Misevolution in Self-Improving LLM Agents
Aug 13, 2026Xutao Mao, Liangjie Zhao, Xiang Zheng +1LLM Agent Skill Learning AI Agent Safety Benchmarks
City University of Hong Kong · Adelaide University
AQuA: Recursively Self-Improving Quantitative Trading Research Agents
Aug 13, 2026Jiacheng Guo, Suozhi Huang, Yunlong Gao +5Quantitative Finance Automated Algorithm Discovery
Princeton University · Ant Group · Stanford University
SBCO: Self-Supervised, Verifier-Grounded Harness Optimization For Planning Agents
Aug 10, 2026Vivek Kulkarni, Sudipta Paul, Aounon Kumar +2Agent Harness Optimization Self-Improving Agents
Samsung Research America
Continuous Improvement and Parallel Autonomous Exploration: An LLM-Agent Framework for Searching Large Solution Spaces
Aug 5, 2026Dulmini Hettiarachchi, Andre Rusli, Julio Christian Young +1Multi-Agent LLM Systems Self-Improving Agents
Mercari, Inc. Japan
PAST-Bench: Benchmarking the Foundations of Recursive Self-Improvement in Personal Agents
Aug 4, 2026Shuhan Xue, Zixin Ding, Yichen Shen +6LLM Agent Memory Long-Horizon Agent Evaluation
ADIAS: Automated Design of Interactive Agentic Systems
Aug 3, 2026Lekang Jiang, Bohan Tang, Stephan Goetz +1Agent Harness Optimization Self-Improving Agents
University of Cambridge · LIGHTSPEED · Independent Researcher
KC-Agent: A Dual-Process Cognitive Architecture for Efficient ML Model Improvement
Aug 3, 2026Gusseppe Bravo-Rocca, Jordi Guitart, Ajay Dholakia +2Cognitive Modeling Concept Drift
Barcelona Supercomputing Center, Barcelona, Spain · Cotiviti, South Jordan, UT, United States · Universitat Polit`ecnica de Catalunya, Barcelona, Spain +1
Weights or Skills? A Survey of Robot-Learning Techniques: from Action-Predicting Weights to Robots that Write their Own Skills
Aug 3, 2026Gaytri Jena, Kapil Wanaskar, Vinija Jain +3Robot Skill Learning Agent Skill Learning
UC Berkeley, Berkeley, CA, USA · San Jose State University, USA · Meta, USA +3
Frontis-MA1: Training an AI4AI Model towards Recursive Self-Improvement in Machine Learning Engineering
Jul 30, 2026Junlin Yang, Che Jiang, Yu Fu +21Evolutionary Optimization Self-Improving Agents
Horizon Research, Frontis.AI
Self-Authored Verification Is Unreliable in Heuristic Self-Improving Agents
Jul 27, 2026Diandian Guo, Cong Cao, Fangfang Yuan +3AI Agent Evaluation LLM Agent Self-Improvement
Institute of Information Engineering, Chinese Academy of Sciences, Beijing, China · School of Cyber Security, University of Chinese Academy of Sciences, Beijing, China
AREX: Towards a Recursively Self-Improving Agent for Deep Research
Jul 23, 2026Shuqi Lu, Chaofan Li, Kun Luo +21Agentic RL Deep Research Agents
Beijing Academy of Artificial Intelligence (BAAI)
Knowledge-Centric Self-Improvement
Jul 21, 2026Xuefei Julie Wang, Lauren Hyoseo Yoon, Chengrui Qu +4Self-Improving Agents Memory-Augmented Agents
Caltech
SR-Agent: An Experience-Driven Agentic Framework for Post-Ranking Strategy Refinement in E-Commerce Recommendation
Jul 20, 2026Hanchen Yang, Kaiwen Yang, Junpeng Zhuang +7Recommender Systems Self-Improving Agents
Kuaishou Technology, Beijing, China
AIGB-R1: Self-Evolving Generative Auto-Bidding via Hierarchical Planner-Executor Optimization
Jul 19, 2026Yuejia Dou, Hesong Wang, Xinyu Zhang +6Automated Bidding Large Language Model-Guided Optimization
Gaoling School of Artificial Intelligence, Renmin University of China Beijing, China · Alibaba Group Beijing, China
Fantastic Adaptive Taxonomies and How to Use Them
Jul 17, 2026Mert Cemri, Andrei Cojocaru, Melissa Pan +9Agent Failure Analysis LLM Agent Evaluation
University of California, Berkeley · Apple · Bespoke Labs
Self-Improvements in Modern Agentic Systems: A Survey
Jul 14, 2026Zhe Ren, Yimeng Chen, Dandan Guo +9LLM Agent Self-Improvement Self-Improving Agents
School of Artificial Intelligence, Jilin University · King Abdullah University of Science and Technology (KAUST) · Independent Researcher +2
TTHE: Test-Time Harness Evolution
Jul 9, 2026Jun Nie, Yonggang Zhang, Jun Song +5Test-Time Adaptation Agent Harness Optimization
Hong Kong Baptist University · University of Science and Technology of China · Hong Kong Generative AI Research & Development Center, The Hong Kong University of Science and Technology +1
Recursive Self-Improvement in AI: From Bounded Self-Refinement to Autonomous Research Loops
Jul 8, 2026Mingguang Chen, Licheng Wang, Bo QuLLM Self-Refinement Human-in-the-Loop Evaluation
University of California, Riverside (UCR) · AlphaAvatar · Illinois Institute of Technology (IIT)
SelfMem: Self-Optimizing Memory for AI Agents
Jul 4, 2026Shu Yang, Junchao Wu, Derek F. Wong +1LLM Agent Memory Long-Term Conversational Memory
1PRADA Lab, King Abdullah University of Science and Technology · NLP2CT Lab, University of Macau
Learning from Failure: Inference-Time Self-Improvement for Computer-Use Agents
Jun 30, 2026Xueqiao Sun, Xiaohan Wang, Ludwig Schmidt +2Computer-Use Agents Self-Improving Agents
1Stanford University, Stanford, CA 94305, USA
Experience Graphs: The Data Foundation for Self-Improving Agents
Jun 29, 2026Gang Liao, Yujia He, Abdullah Ozturk +22Self-Improving Agents Long-Horizon Agent Tasks
Meta Platforms · University of Maryland, College Park
Hierarchical Experimentalist Agents
Jun 28, 2026Abhranil Chandra, Utsav Dhanuka, Varun Gandhi +2LLM Agent Skill Learning AI Agent Benchmarks
University of Massachusetts Amherst
Algorithmic Unverifiability of Safety for Fixed and Recursively Self-Improving Systems
Jun 26, 2026Jose Pascual Gumbau MezquitaFormal Verification Self-Improving Agents
Universitat Jaume I, Castelló de la Plana, Spain
AgentX: Towards Agent-Driven Self-Iteration of Industrial Recommender Systems
Jun 25, 2026Changxin Lao, Fei Pan, Guozhuang Ma +59Software Engineering Agents Multi-Agent Collaboration
The Red Queen Gödel Machine: Co-Evolving Agents and Their Evaluators
Jun 24, 2026Alex Iacob, Andrej Jovanović, William F. Shen +12Evolutionary Optimization AI Agent Evaluation
University of Cambridge · Flower Labs · NVIDIA +2
Self-Evolution for Multi-Turn Tool-Calling Agents via Divergence-Point Preference Learning
Jun 22, 2026Jiaqiang TangTool-Using Agents Self-Improving Agents
The Hong Kong University of Science and Technology (Guangzhou)
ENPIRE: Agentic Robot Policy Self-Improvement in the Real World
Jun 18, 2026Wenli Xiao, Jia Xie, Tonghe Zhang +14Robot Policy Learning Robot Skill Learning
1NVIDIA · 2CMU · 3UC Berkeley
Skill-Guided Continuation Distillation for GUI Agents
Jun 17, 2026Zhimin Fan, Hongwei Yu, Yeqing Shen +9Imitation Learning Agent Skill Learning
1StepFun · University of Science and Technology Beijing · 4Nanyang Technological University +1
Closing the Reflection Gap: A Free Calibration Bonus for Agentic RL
Jun 12, 2026Yinglun ZhuLLM Agent Evaluation Language Model Calibration
University of California, Riverside
From AGI to ASI
Jun 10, 2026Tim Genewein, Matija Franklin, Alexander Lerchner +11Self-Improving Agents Recursive Self-Improvement
Google DeepMind · University of Waterloo (work conducted while at Google DeepMind) · Australian National University +1

Aug 13, 2026J/K move · Enter open · S save

Practice Makes Unsafe: Skill Misevolution in Self-Improving LLM Agents

Xutao Mao, Liangjie Zhao, Xiang Zheng, Cong Wang

City University of Hong Kong · Adelaide University

LLM Agent Skill Learning AI Agent Safety Benchmarks Self-Improving Agents LLM Agent Safety

Open paper AskPDF arXiv

LLM agents increasingly save experience from past tasks as reusable skills through skill evolution. When an agent completes an unsafe task, skill evolution can keep the unsafe procedure inside an otherwise useful skill, and a later benign task can repeat the harm after the unsafe instruction is gone. We call this failure skill misevolution. It requires that the agent write the unsafe procedure into a skill and that a later task retrieve and follow that skill. We introduce SkillMisevo-Gym, which clears every task session but keeps the skill library, so each of these steps can be checked separately. We also build SkillMisevo-Bench with 525 tasks in 25 episodes, where each episode alternates malicious and related benign tasks before three benign tasks in a new session. Across 21 combinations of agents and evolution methods, all of them write unsafe skills and 15 carry the harm into the new session, so the risk comes from skill evolution itself. Evolution raises benign utility in 15 combinations but also raises malicious-task harm in 17, so ranking methods by utility alone would hide this risk. We propose SafeEvolve, which checks each skill when it is written and when a later task retrieves it. It cuts unsafe retrieval in the new session from 35.3% to 8.7% and harm from 21.3% to 4.0% while keeping benign utility during learning within 0.4 points. SafeEvolveUp cuts new-session harm by 16.0 points and keeps its utility within 1.3 points of evolution without this mitigation. As agents learn from their own experience, safety must cover the skills that carry past behavior into future tasks. Code is available at https://github.com/henrymao2004/misevolve.