The papers

Most central to this idea first, not most recent.

• Top Universitycs.AI

  Agents Can Drop Old Reasoning After Externalizing Task State

  TierFlow Team, Renmin University of China · Sept 2026

• Top Universitycs.AIcode

  Dropping Rather Than Rewriting Context Cuts Long-Horizon Agent Costs

  Carnegie Mellon University, Bosch Center for AI · Sept 2026

• Chinese Techcs.CL

  Selecting evidence before summarizing improves long-context reasoning efficiency

  Peking University, Baidu Inc. · Sept 2026

• Top Universitycs.AI

  Compressing old tool outputs cuts coding-agent context with modest accuracy loss

  Peking University · Sept 2026

• Industrycs.AI

  Coding-agent harness choices should match model skill and context budget

  UMass Amherst, Zoom Video Communications · Sept 2026

• Industrycs.AI

  Typed trace folding helps agents and observers manage long runs

  Salesforce AI Research · Sept 2026

• Top Universitycs.LG

  RL with context compaction trains long-horizon agents

  Tsinghua University · July 2026

• Big Techcs.CLcode

  Letting language models decide when to compact their context windows

  Johns Hopkins University, Apple · June 2026

• Independentcs.AI

  Pruning context to recent tool calls improves agent reliability and efficiency

  June 2026

• Chinese Techcs.CL

  Separating planning from synthesis improves long-horizon search agents

  Zhejiang University, Tencent · Sept 2026

• Industrycs.AI

  Research agent improves itself through seven successive code rewrites

  Weco AI · Sept 2026

• Chinese Techcs.AI

  Restructuring agent environments improves performance on noisy, evolving tasks

  Shanghai Jiao Tong University, Theseus Labs · Sept 2026

• Chinese Techcs.AIcode

  C3M preserves multimodal evidence across sessions within fixed memory budgets

  Huzhou Normal University, Alibaba Group · Sept 2026

• Big Techcs.AI

  Recursive harness search cuts coding-agent token use nearly in half

  NVIDIA, NTU · Sept 2026

• Independentcs.AI

  Smarter merging and packing improve LLM memory under tight budgets

  Sept 2026

• Top Universitycs.IRcode

  VikingRAG Cuts Token Use for Retrieval Over Structured Documents

  Renmin University of China, Independent Researcher · Sept 2026

• Top Universitycs.AI

  Markdown interfaces cut agent context costs without reducing task success

  Seoul National University, H1R.AI · Sept 2026

• Chinese Techcs.CV

  Streaming video memory works better when internalized as evolving latent tokens

  Nanjing University of Science and Technology, Ant Group · Sept 2026

• Top Universitycs.CR

  Retained Tool Outputs Can Repeatedly Inflate LLM Agent Costs

  Institute of Information Engineering, Chinese Academy of Sciences, University of Chinese Academy of Sciences · Sept 2026

• Top Universitycs.AI

  Trajectory shortcut trees improve agents without outcome labels or annotations

  Fudan University, Meituan Longcat Team · Sept 2026

• Top Universitycs.RO

  Saliency-driven workspace tokens give robots lightweight task memory

  Massachusetts Institute of Technology, Carnegie Mellon University · Sept 2026

• Independentcs.AI

  Structured agent memories withstand model upgrades better than compressed notes

  Sept 2026

• Research Labcs.AI

  Hierarchical memory trees improve long-context reasoning without model training

  Institute of Automation, Chinese Academy of Sciences · Sept 2026

• Chinese Techcs.CL

  Similarity-aware context windows improve model routing across multi-turn conversations

  Shanghai Jiao Tong University, Baidu Inc. · Sept 2026

• Independentcs.CL

  LLM agents do not reliably improve from their own experience

  Sept 2026

• Chinese Techcs.AI

  Full-context drafters help compressed agent models retain accuracy

  Huawei Technologies Co., Ltd., University of Science and Technology of China · Aug 2026

• Chinese Techcs.CL

  Agents learn to manage long contexts through fine-grained reinforcement learning

  Tsinghua University, Tencent Youtu Lab · Sept 2026

• Top Universitycs.AI

  Coding Agents Need Semantic Measures of Working Memory Performance

  Argonne National Laboratory, Columbia University · Sept 2026

• Big Techcs.LG

  A lightweight PyTorch-native framework matches Megatron-based agentic RL training performance.

  NVIDIA · July 2026

• Big Techcs.AI

  Memory as state management instead of semantic retrieval improves long-horizon agents

  University of Science and Technology of China, Microsoft · June 2026

• Industrycs.AI

  Distilled repository skills improve agents conducting machine-learning research

  Beijing Academy of Artificial Intelligence, University of Science and Technology of China · Sept 2026