Long-horizon LLM agents accumulate tool results, files, plans, and user constraints that are too structured to be treated as a disposable text suffix. Current systems mostly rely on in-run heuristics such as chronological pruning and tool-output masking, or on final self-summary near a context limit. Heuristics are cheap but blind to future dependencies; summaries preserve narrative state but often hide exact evidence, locators, and editable artifacts. We present Self-GC, where GC denotes self-governing context while deliberately echoing garbage collection: the system does not merely reclaim unused tokens, but governs the lifecycle of agent context objects. Self-GC turns user turns, tool spans, and skill state into indexed objects; asks a side-channel planner to propose fold, mask, and prune actions; and lets the harness enforce recoverable sidecars, safe commit boundaries, and cache-aware commit. On a 33-session Hard Set, Self-GC prunes 43.95% of prefix tokens while leaving 84.85% of future continuations unaffected, compared with no-impact rates of 54.55% to 69.70% for heuristic baselines. On a 332-session production-derived suite, three planner backbones reach no-impact rates of 91.27% to 94.58%, while baselines remain at 77.71% to 87.46%. In production, an online account-level split reduces daytime average input tokens by 10% to 15%, with peak reductions near 20%. These results point to context management as runtime lifecycle control over indexed, recoverable objects rather than post hoc text cleanup.
Fudan University(复旦大学);Shanghai Key Laboratory of Data Science(上海市数据科学重点实验室);College of Computer Science and Artificial Intelligence, Fudan University(复旦大学计算机科学与人工智能学院)
Can LLM Agents Be CFOs? Benchmarking Long-Horizon Resource Allocation in an Uncertain Enterprise Environment
Georgia Institute of Technology(佐治亚理工学院);The Fin AI;Stevens Institute of Technology(史蒂文斯理工学院);Columbia University(哥伦比亚大学);George Mason University(乔治·马歇尔大学);McGill University(麦吉尔大学);Mohamed bin Zayed University of Artificial Intelligence(莫扎大学人工智能学院);California State University, Monterey Bay(加州州立大学蒙特雷湾分校);University of Manchester(曼彻斯特大学);Mila – Quebec Artificial Intelligence Institute(魁北克人工智能研究所);Université de Montréal(蒙特利尔大学)
Co-Evolving LLM Decision and Skill Bank Agents for Long-Horizon Tasks
University of Maryland(马里兰大学);University of Southern California(南加州大学);Good Start Labs(Good Start实验室);Independent Researcher(独立研究者);Mohamed bin Zayed University of Artificial Intelligence(穆罕默德·本·扎耶德人工智能大学)
Learning on the Job: An Experience-Driven Self-Evolving Agent for Long-Horizon Tasks
Central South University(中南大学);Shanghai Artificial Intelligence Laboratory(上海人工智能实验室);Fudan University(复旦大学);Shanghai Innovation Institute(上海创新研究院);Zhejiang University(浙江大学)
Travel Back To Move Forward: Retrospective Exploration and Evolving for Long Horizon Agents
University of Science and Technology of China(中国科学技术大学);Shanghai Innovation Institute(上海创新研究院);Facebook(脸书);Zhejiang University(浙江大学);Tsinghua University(清华大学);Meituan(美团);Fudan University(复旦大学)
Hebrew University of Jerusalem(耶路撒冷希伯来大学);University of Toronto(多伦多大学);Schwartz Reisman Institute for Technology and Society(施瓦茨·里斯曼技术与社会研究所);Institute for Law & AI(法律与人工智能研究所)
Governing Execution Risk in Agentic AI Systems: A Trajectory-Guided Framework for Red Teaming
Department of Information Systems, Business Statistics and Operations Management (ISOM), Hong Kong University of Science and Technology(香港科技大学信息系统、商务统计与运营管理系(ISOM))
Towards Governing Agent's Efficacy: Action-Conditional $β$-VAE for Deep Transparent Reinforcement Learning
Seoul National University(首尔国立大学);Samsung Electronics(三星电子)
我们在强化学习(RL)的设置中解决了深度神经网络的黑盒问题,其中神经代理以不可控的方式学习最大化奖励收益。当交互环境包括广阔的状态空间时,这种学习方法是有风险的,因为几乎不可能预见所有不想要的结果并预先用负奖励惩罚它们。与以前的作品中学习的神经功能的反向分析不同,我们提出的方法\nj{tackles the blackbox issue by encouraging}是一个RL策略网络,通过实现一个解纠缠的表示学习方法来学习可解释的潜在特征。为此,我们的方法允许RL代理通过将其影响与不可控的环境因素区分开来来理解自我效能,这与人类理解场景的方式非常相似。我们的实验结果表明,学习的潜在因素不仅是可解释的,而且还可以建模整个访问状态空间的分布与特定的动作条件。我们已经实验表明,拟议结构的这一特征可以导致对RL代理的期望行为进行事后治理。
Long-Horizon-Terminal-Bench: Testing the Limits of Agents on Long-Horizon Terminal Tasks with Dense Reward-Based Grading
Tencent(腾讯);University of Maryland, College Park(马里兰大学帕克分校);University of Georgia(佐治亚大学);University of Minnesota, Twin Cities(明尼苏达大学双城分校);Indiana University(印第安纳大学);Lehigh University(里海大学);National University of Singapore(新加坡国立大学);The Hong Kong Polytechnic University(香港理工大学)
Learning Agent-Compatible Context Management for Long-Horizon Tasks
Renmin University of China(中国人民大学);Tongyi Lab, Alibaba Group(阿里云实验室);Beijing University of Posts and Telecommunications(北京邮电大学);Université de Montréal(蒙特利尔大学)