mlsanityData to 14 Oct 2026

Memory as Action: Autonomous Context Curation for Long-Horizon Agentic Tasks

arXiv 2510.12635 ↗First posted 14 Oct 2025cs.AIcited by 66

On the map: LLM agents and tool use → Agent memory systems

Abstract (arXiv)

Long-context Large Language Models, despite their expanded capacity, require careful working memory management to mitigate attention dilution during long-horizon tasks. Yet existing approaches rely on external mechanisms that lack awareness of the agent's reasoning state, leading to suboptimal decisions. We propose Memory-as-Action (MemAct), a framework that treats working memory management as learnable policy actions. By formulating context management as in-place editing operations (deletion, insertion), MemAct enables joint optimization of information retention and task performance through end-to-end reinforcement learning. To address the computational challenges of dynamic context updates, we introduce Dynamic Context Policy Optimization, which restores training efficiency without compromising reasoning integrity. Experiments show that MemAct-RL-14B matches the accuracy of models 16× larger while reducing average context length by 51%, with learned strategies that adapt to model capabilities and generalize across task complexities.

Key facts. Cited by 66 papers in the map (data through 14 Oct 2026). Placed in Agent memory systems, one of 16 directions in LLM agents and tool use. Builds on 24 of its 31 references that are in the map. First posted on arXiv 14 Oct 2025.

Builds on

24 of 31 references.

Closest ancestors · 3

Same direction, newest first

  1. 2025ReSum: Unlocking Long-Horizon Search Intelligence via Context Summarization16 Sep
  2. 2025Memory-R1: Enhancing Large Language Model Agents to Manage and Utilize Memories via Reinforcement Learning27 Aug
  3. 2025MEM1: Learning to Synergize Memory and Reasoning for Efficient Long-Horizon Agents18 Jun

Other references · 21

  1. 2022Training language models to follow instructions with human feedbackcited by 15,971
  2. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 9,379
  3. 2018HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answeringcited by 3,754
  4. 2023Lost in the Middle: How Language Models Use Long Contextscited by 3,455
  5. 2023SWE-bench: Can Language Models Resolve Real-World GitHub Issues?cited by 3,373
  6. 2025Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learningcited by 1,444
  7. 2023MemGPT: Towards LLMs as Operating Systemscited by 1,389
  8. 2025gpt-oss-120b & gpt-oss-20b Model Cardcited by 1,386
  9. 2020Constructing A Multi-hop QA Dataset for Comprehensive Evaluation of Reasoning Stepscited by 1,282
  10. 2023SGLang: Efficient Execution of Structured Language Model Programscited by 1,141
  11. 2022Measuring and Narrowing the Compositionality Gap in Language Modelscited by 1,075
  12. 2021MuSiQue: Multihop Questions via Single-hop Question Compositioncited by 916
9 more
  1. 2025BrowseComp: A Simple Yet Challenging Benchmark for Browsing Agentscited by 647
  2. 2023YaRN: Efficient Context Window Extension of Large Language Modelscited by 601
  3. 2025MemAgent: Reshaping Long-Context LLM with Multi-Conv RL-based Memory Agentcited by 246
  4. 2025BrowseComp-Plus: A More Fair and Transparent Evaluation Benchmark of Deep-Research Agentcited by 200
  5. 2024Fact, Fetch, and Reason: A Unified Evaluation of Retrieval-Augmented Generationcited by 162
  6. 2025Beyond Ten Turns: Unlocking Long-Horizon Agentic Search with Large-Scale Asynchronous RLcited by 150
  7. 2025A Survey of Context Engineering for Large Language Modelscited by 126
  8. 2025Scaling LLM Multi-turn RL with End-to-end Summarization-based Context Managementcited by 51
  9. 2025Hierarchical Document Refinement for Long-context Retrieval-augmented Generationcited by 16

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 3; 2026: 63.

66 papers.

Sort:

By area:

  1. 2025Memory in the Age of AI Agentscited by 303
  2. 2026Agentic Memory: Learning Unified Long-Term and Short-Term Memory Management for Large Language Model Agentscited by 78
  3. 2026A Survey of Agentic Reasoning for Large Language Models: Towards Recursively Self-Improving and Collective Agentscited by 50
  4. 2026SkillOS: Learning Skill Curation for Self-Evolving Agentscited by 49
  5. 2026A Survey of Agent Memory in the Second Half: Towards Self-Evolving and Long-Horizon Agentscited by 44
  6. 2026Anatomy of Agentic Memory: Taxonomy and Empirical Analysis of Evaluation and System Limitationscited by 21
  7. 2026From Storage to Experience: A Survey on the Evolution of LLM Agent Memory Mechanismscited by 21
  8. 2026Governing Evolving Memory in LLM Agents: Risks, Mechanisms, and the Stability and Safety Governed Memory (SSGM) Frameworkcited by 18
  9. 2026MemMark: State-Evolution Attribution Watermarking for Agent Long-Term Memory Systemscited by 16
  10. 2026JIT-Agent: Scaling Harness Intelligence via Just-in-Time Harness Evolutioncited by 16
  11. 2026AtomMem : Learnable Dynamic Agentic Memory with Atomic Memory Operationcited by 15
  12. 2026Learning Agent-Compatible Context Management for Long-Horizon Taskscited by 15
38 more
  1. 2026PERMA: Benchmarking Personalized Memory Agents via Event-Driven Preference and Realistic Task Environmentscited by 14
  2. 2026ContextBudget: Budget-Aware Context Management for Long-Horizon Search Agentscited by 14
  3. 2025AI Meets Brain: Memory Systems from Cognitive Neuroscience to Autonomous Agentscited by 12
  4. 2026Remember When It Matters: Proactive Memory Agent for Long-Horizon Agentscited by 11
  5. 2026Agent-BRACE: Decoupling Beliefs from Actions in Long-Horizon Tasks via Verbalized State Uncertaintycited by 10
  6. 2026Autonomous Research Agents: A Survey of AI Scientists and the Verification Gapcited by 9
  7. 2025Beyond Pipelines: A Survey of the Paradigm Shift toward Model-Native Agentic AIcited by 8
  8. 2026Remember the Decision, Not the Description: A Rate-Distortion Framework for Agent Memorycited by 7
  9. 2026Memory as a Controlled Process: Learned Adaptive Memory Management for LLM Agentscited by 7
  10. 2026An Empirical Study of Harness Design for Coding Agentscited by 6
  11. 2026Aligning Agentic World Models via Knowledgeable Experience Learningcited by 5
  12. 2026Large Language Model Agents Are Not Always Faithful Self-Evolverscited by 5
  13. 2026MemQ: Integrating Q-Learning into Self-Evolving Memory Agents over Provenance DAGscited by 5
  14. 2026AutoMem: Automated Learning of Memory as a Cognitive Skillcited by 5
  15. 2026Cortex: A Bidirectionally Aligned Embodied Agent Framework for Long-horizon Manipulationcited by 5
  16. 2026AREX: Towards a Recursively Self-Improving Agent for Deep Researchcited by 5
  17. 2026See, Remember, Explore: A Benchmark and Baselines for Streaming Spatial Reasoningcited by 4
  18. 2026MemFactory: Unified Inference & Training Framework for Agent Memorycited by 4
  19. 2026ClawVM: Harness-Managed Virtual Memory for Stateful Tool-Using LLM Agentscited by 4
  20. 2026Learning What to Remember: Observability-Safe Memory Retention via Constrained Optimization for Long-Horizon Language Agentscited by 4
  21. 2026What to Keep, What to Forget: A Rate--Distortion View of Memory Compaction in LLMs and Agentscited by 4
  22. 2026OneDayAgent: Towards a Long-Horizon Harness for Autonomous Agentscited by 4
  23. 2026Yunque DeepResearch Technical Reportcited by 3
  24. 2026Organize then Retrieve: Hierarchical Memory Navigation for Efficient Agentscited by 3
  25. 2026Diagnosing and Mitigating Context Rot in Long-horizon Searchcited by 3
  26. 2026LLM Agents Are Latent Context Managers: Eliciting Self-Managed Context via State Proprioceptioncited by 3
  27. 2026ECHO: Prune To Act, Trace To Learn With Selective Turn Memory In Agentic RLcited by 3
  28. 2026MemArbiter: Decision-Time Memory Arbitration for Long-Horizon LLM Agentscited by 3
  29. 2026Evaluate-as-Action: Self-Evaluated Process Rewards for Retrieval-Augmented Agentscited by 2
  30. 2026Beyond Fluency: Toward Reliable Trajectories in Agentic IRcited by 2
  31. 2026Learning How and What to Memorize: Cognition-Inspired Two-Stage Optimization for Evolving Memorycited by 2
  32. 2026ACE: Pluggable Adaptive Context Elasticizer across Agentscited by 2
  33. 2026CrystalMem: Elastic Memory for Self-Evolving LLM Agents via Knowledge Crystallizationcited by 2
  34. 2026ContextPilot: Teaching Agents for Proactive Context Management via Fine-grained RLcited by 2
  35. 2026Cyclical Entropy Eruption: Entropy Dynamics in Agent Reinforcement Learningcited by 1
  36. 2026ActiveMem: Distributed Active Memory for Long-Horizon LLM Reasoningcited by 1
  37. 2026Dual-Grained Agent Memory and Shapley Context Attribution for Multimodal Agentic Learnercited by 1
  38. 2026Self-Programmed Execution for Language-Model Agentscited by 0

50 most cited of 66 All 66 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2026Verifiable Memory: Learning Unified Memory Management with Local and Global Verifiers for Large Language Model Agents
  2. 2026SAM: State-Adaptive Memory for Long-Horizon Reasoning Agent
  3. 2026Mem-π: Adaptive Memory through Learning When and What to Generate
  4. 2026PRO-LONG: Programmatic Memory Enables Long-Horizon Reasoning
  5. 2025MemGen: Weaving Generative Latent Memory for Self-Evolving Agents
  6. 2026Memory Control Signals Emerge Before Action in Long Horizon Agents
  7. 2026MemCo: Memory-Centric Collaboration for Generalizing LLM Agents to Unseen Environments
  8. 2026Mem2ActBench: A Benchmark for Evaluating Long-Term Memory Utilization in Task-Oriented Autonomous Agents