mlsanityData to 14 Oct 2026

Get Experience from Practice: LLM Agents with Record & Replay

arXiv 2505.17716 ↗First posted 23 May 2025cs.LG cs.MAcited by 23

On the map: LLM agents and tool use → General LLM agent frameworks and benchmarks

Abstract (arXiv)

AI agents, empowered by Large Language Models (LLMs) and communication protocols such as MCP and A2A, have rapidly evolved from simple chatbots to autonomous entities capable of executing complex, multi-step tasks, demonstrating great potential. However, the LLMs' inherent uncertainty and heavy computational resource requirements pose four significant challenges to the development of safe and efficient agents: reliability, privacy, cost and performance. Existing approaches, like model alignment, workflow constraints and on-device model deployment, can partially alleviate some issues but often with limitations, failing to fundamentally resolve these challenges. This paper proposes a new paradigm called AgentRR (Agent Record & Replay), which introduces the classical record-and-replay mechanism into AI agent frameworks. The core idea is to: 1. Record an agent's interaction trace with its environment and internal decision process during task execution, 2. Summarize this trace into a structured "experience" encapsulating the workflow and constraints, and 3. Replay these experiences in subsequent similar tasks to guide the agent's behavior. We detail a multi-level experience abstraction method and a check function mechanism in AgentRR: the former balances experience specificity and generality, while the latter serves as a trust anchor to ensure completeness and safety during replay. In addition, we explore multiple application modes of AgentRR, including user-recorded task demonstration, large-small model collaboration and privacy-aware agent execution, and envision an experience repository for sharing and reusing knowledge to further reduce deployment cost.

Key facts. Cited by 23 papers in the map (data through 14 Oct 2026). Placed in General LLM agent frameworks and benchmarks, one of 16 directions in LLM agents and tool use. Builds on 58 of its 82 references that are in the map. First posted on arXiv 23 May 2025.

Builds on

58 of 82 references.

Closest ancestors · 5

Same direction, newest first

  1. 2025From Mind to Machine: The Rise of Manus AI as a Fully Autonomous Digital Agent4 May
  2. 2025Advances and Challenges in Foundation Agents: From Brain-Inspired Intelligence to Evolutionary, Collaborative, and Safe Systems31 Mar
  3. 2025SagaLLM: Context Management, Validation, and Transaction Guarantees for Multi-Agent LLM Planning15 Mar
  4. 2025Division-of-Thoughts: Harnessing Hybrid Language Model Synergy for Efficient On-Device Agents6 Feb
  5. 2023Agents: An Open-source Framework for Autonomous Language Agents14 Sep

Other references · 53

  1. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,380
  2. 2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,695
  3. 2023Safe RLHF: Safe Reinforcement Learning from Human Feedbackcited by 656
  4. 2025Stop Overthinking: A Survey on Efficient Reasoning for Large Language Modelscited by 464
  5. 2021Large Language Models Can Be Strong Differentially Private Learnerscited by 412
  6. 2025ReTool: Reinforcement Learning for Strategic Tool Use in LLMscited by 392
  7. 2025Absolute Zero: Reinforced Self-play Reasoning with Zero Datacited by 344
  8. 2024Hallucination is Inevitable: An Innate Limitation of Large Language Modelscited by 332
  9. 2025Reinforcement Learning for Reasoning in Large Language Models with One Training Examplecited by 295
  10. 2025DeepSeek-Prover-V2: Advancing Formal Mathematical Reasoning via Reinforcement Learning for Subgoal Decompositioncited by 287
  11. 2025Agentic Retrieval-Augmented Generation: A Survey on Agentic RAGcited by 249
  12. 2025ReSearch: Learning to Reason with Search for LLMs via Reinforcement Learningcited by 161
38 more
  1. 2023S-LoRA: Serving Thousands of Concurrent LoRA Adapterscited by 152
  2. 2025Mobile-Agent-E: Self-Evolving Mobile Assistant for Complex Taskscited by 131
  3. 2025SWEET-RL: Training Multi-Turn LLM Agents on Collaborative Reasoning Taskscited by 118
  4. 2025A Comprehensive Survey in LLM(-Agent) Full Stack Safety: Data, Training and Deploymentcited by 116
  5. 2023Cerebras-GPT: Open Compute-Optimal Language Models Trained on the Cerebras Wafer-Scale Clustercited by 109
  6. 2024AirGapAgent: Protecting Privacy-Conscious Conversational Agentscited by 94
  7. 2024LLMs Will Always Hallucinate, and We Need to Live With Thiscited by 78
  8. 2025Open Challenges in Multi-Agent Security: Towards Secure Systems of Interacting AI Agentscited by 71
  9. 2025Beyond Single-Turn: A Survey on Multi-Turn Interactions with Large Language Modelscited by 62
  10. 2025Efficient Inference for Large Reasoning Models: A Surveycited by 61
  11. 2025The Obvious Invisible Threat: LLM-Powered GUI Agents' Vulnerability to Fine-Print Injectionscited by 59
  12. 2023Explore, Select, Derive, and Recall: Augmenting LLM with Human-like Memory for Mobile Task Automationcited by 58
  13. 2025ACE: A Security Architecture for LLM-Integrated App Systemscited by 55
  14. 2025I Have Covered All the Bases Here: Interpreting Reasoning Features in Large Language Models via Sparse Autoencoderscited by 45
  15. 2025Synthetic Data Generation & Multi-Step RL for Reasoning & Tool Usecited by 45
  16. 2024R2-Guard: Robust Reasoning Enabled LLM Guardrail via Knowledge-Enhanced Logical Reasoningcited by 43
  17. 2025Sleep-time Compute: Beyond Inference Scaling at Test-timecited by 43
  18. 2024CLEAR: Towards Contextual LLM-Empowered Privacy Policy Analysis and Risk Generation for Large Language Model Applicationscited by 33
  19. 2025Hogwild! Inference: Parallel LLM Generation via Concurrent Attentioncited by 32
  20. 2025MxMoE: Mixed-precision Quantization for MoE with Accuracy and Performance Co-Designcited by 30
  21. 2024Fine-Tuning Large Language Models with User-Level Differential Privacycited by 29
  22. 2025Tina: Tiny Reasoning Models via LoRAcited by 25
  23. 2024When LLMs Go Online: The Emerging Threat of Web-Enabled LLMscited by 24
  24. 202570% Size, 100% Accuracy: Lossless LLM Compression for Efficient GPU Inference via Dynamic-Length Float (DFloat11)cited by 24
  25. 2025Enhancing Retrieval-Augmented Generation: A Study of Best Practicescited by 23
  26. 2025Equilibrate RLHF: Towards Balancing Helpfulness-Safety Trade-off in Large Language Modelscited by 22
  27. 2025An Empirical Study of Qwen3 Quantizationcited by 21
  28. 2025Retrieval Augmented Generation Evaluation in the Era of Large Language Models: A Comprehensive Surveycited by 20
  29. 2025FlowAgent: Achieving Compliance and Flexibility for Workflow Agentscited by 19
  30. 2024LoRA-Switch: Boosting the Efficiency of Dynamic LLM Adapters via System-Algorithm Co-designcited by 18
  31. 2025BitNet v2: Native 4-bit Activations with Hadamard Transformation for 1-bit LLMscited by 18
  32. 2025Instruction-Following Pruning for Large Language Modelscited by 17
  33. 2025TAMP: Token-Adaptive Layerwise Pruning in Multimodal Large Language Modelscited by 9
  34. 2025Bridging the Safety Gap: A Guardrail Pipeline for Trustworthy LLM Inferencescited by 8
  35. 2025Improving Large Language Model Planning with Action Sequence Similaritycited by 7
  36. 2025RT-Cache: Training-Free Retrieval for Real-Time Manipulationcited by 6
  37. 2025Cape: Context-Aware Prompt Perturbation Mechanism with Differential Privacycited by 5
  38. 2025Federated Intelligence: When Large AI Models Meet Federated Fine-Tuning and Collaborative Reasoning at the Network Edgecited by 4

50 most cited of 53 All 53 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 9; 2026: 14.

23 papers.

Sort:

By area:

  1. 2025Memory in the Age of AI Agentscited by 303
  2. 2025A Survey of Context Engineering for Large Language Modelscited by 126
  3. 2025Memp: Exploring Agent Procedural Memorycited by 103
  4. 2025SWE-Exp: Experience-Driven Software Issue Resolutioncited by 65
  5. 2026Atomix: Timely, Transactional Tool Use for Reliable Agentic Workflowscited by 30
  6. 2026Trajectory-Informed Memory Generation for Self-Improving Agent Systemscited by 17
  7. 2025AlphaOPT: Formulating Optimization Programs with Self-Improving LLM Experience Librarycited by 14
  8. 2025AI Meets Brain: Memory Systems from Cognitive Neuroscience to Autonomous Agentscited by 12
  9. 2026ACRFence: Preventing Semantic Rollback Attacks in Agent Checkpoint-Restorecited by 10
  10. 2025MobiAgent: A Systematic Framework for Customizable Mobile Agentscited by 9
  11. 2026PhGPO: Pheromone-Guided Policy Optimization for Long-Horizon Tool Planningcited by 9
  12. 2025What Is Your Agent's GPA? A Framework for Evaluating Agent Goal-Plan-Action Alignmentcited by 8
11 more
  1. 2026Beyond Experience Retrieval: Learning to Generate Utility-Optimized Structured Experience for Frozen LLMscited by 5
  2. 2026PreAct: Computer-Using Agents that Get Faster on Repeated Taskscited by 5
  3. 2025Ask WhAI:Probing Belief Formation in Role-Primed LLM Agentscited by 1
  4. 2026Is One Score Enough? Rethinking the Evaluation of Sequentially Evolving LLM Memorycited by 0
  5. 2026HippoSpark: An On-Demand Experience System for LLM Reasoningcited by 0
  6. 2026Translator vs. Challenger: Adversarial Agentic Learning for C-to-Rust Translationcited by 0
  7. 2026Not All AI Agents Are Equal: Characterizing Resource and Performance Dynamicscited by 0
  8. 2026Chronicle: Cut-Point Replay for Regression Testing of LLM Agentscited by 0
  9. 2026When Successful Memories Mislead Embodied Agents:Memory Adaption For Task-Conditioned Executioncited by 0
  10. 2026Do Tool Calls Execute as Intended? Measuring and Repairing Intent-Execution Correspondence in LLM Agentscited by 0
  11. 2026DeltaReplay: Task-Relative Memory Reuse for Mobile GUI Agentscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Agentic CLEAR: Automating Multi-Level Evaluation of LLM Agents
  2. 2026A Comprehensive Survey on Agent Skills: Taxonomy, Techniques, and Applications
  3. 2026A Large-Scale Empirical Study of Quality Assurance Practices and Gaps in AI Agents
  4. 2026Collaborative Agent Reasoning Engineering (CARE): A Three-Party Design Methodology for Systematically Engineering AI Agents with Subject Matter Experts, Developers, and Helper Agents
  5. 2025Evaluation and Benchmarking of LLM Agents: A Survey
  6. 2026AgentAbstain: Do LLM Agents Know When Not to Act?
  7. 2025A Survey of AI Agent Protocols
  8. 2025LLM-Based Human-Agent Collaboration and Interaction Systems: A Survey