mlsanityData to 7 Oct 2026

Meta Context Engineering via Agentic Skill Evolution

arXiv 2601.21557 ↗First posted 29 Jan 2026cs.AI cs.NEcited by 42

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

Abstract (arXiv)

The operational efficacy of large language models relies heavily on their inference-time context. This has established Context Engineering (CE) as a formal discipline for optimizing these inputs. Current CE methods rely on manually crafted harnesses, such as rigid generation-reflection workflows and predefined context schemas. They impose structural biases and restrict context optimization to a narrow, intuition-bound design space. To address this, we introduce Meta Context Engineering (MCE), a bi-level framework that supersedes static CE heuristics by co-evolving CE skills and context artifacts. In MCE iterations, a meta-level agent refines engineering skills via agentic crossover, a deliberative search over the history of skills, their executions, and evaluations. A base-level agent executes these skills, learns from training rollouts, and optimizes context as flexible files and code. We evaluate MCE across five disparate domains under offline and online settings. MCE demonstrates consistent performance gains, achieving 5.6--53.8% relative improvement over state-of-the-art agentic CE methods (mean of 16.9%), while maintaining superior context adaptability, transferability, and efficiency in both context usage and training.

Key facts. Cited by 42 papers in the map (data through 7 Oct 2026). Placed in General LLM agent frameworks and benchmarks, one of 16 directions in LLM agents and tool use. Builds on 48 of its 71 references that are in the map. First posted on arXiv 29 Jan 2026.

Builds on

48 of 71 references.

Closest ancestors · 4

Same direction, newest first

  1. 2026Computer Environments Elicit General Agentic Intelligence in LLMs22 Jan
  2. 2025FLEX: Continuous Agent Evolution via Forward Learning from Experience9 Nov
  3. 2025A Comprehensive Survey of Self-Evolving AI Agents: A New Paradigm Bridging Foundation Models and Lifelong Agentic Systems10 Aug
  4. 2025Darwin Godel Machine: Open-Ended Evolution of Self-Improving Agents29 May

Other references · 44

  1. 2023Reflexion: Language Agents with Verbal Reinforcement Learningcited by 4,064
  2. 2023AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversationcited by 1,906
  3. 2024SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineeringcited by 1,628
  4. 2025A-MEM: Agentic Memory for LLM Agentscited by 1,000
  5. 2023DSPy: Compiling Declarative Language Model Calls into Self-Improving Pipelinescited by 803
  6. 2025AlphaEvolve: A coding agent for scientific and algorithmic discoverycited by 763
  7. 2025Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memorycited by 722
  8. 2023Eureka: Human-Level Reward Design via Coding Large Language Modelscited by 542
  9. 2024Executable Code Actions Elicit Better LLM Agentscited by 537
  10. 2025GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learningcited by 449
  11. 2024AFlow: Automating Agentic Workflow Generationcited by 395
  12. 2025Agentic Context Engineering: Evolving Contexts for Self-Improving Language Modelscited by 349
32 more
  1. 2024Agent Workflow Memorycited by 342
  2. 2025Memory in the Age of AI Agentscited by 285
  3. 2024Optimizing Instructions and Demonstrations for Multi-Stage Language Model Programscited by 270
  4. 2024ReEvo: Large Language Models as Hyper-Heuristics with Reflective Evolutioncited by 266
  5. 2024Evolution of Heuristics: Towards Efficient Automatic Algorithm Design Using Large Language Modelcited by 258
  6. 2025Dynamic Cheatsheet: Test-Time Learning with Adaptive Memorycited by 146
  7. 2025ShinkaEvolve: Towards Open-Ended And Sample-Efficient Program Evolutioncited by 146
  8. 2025Scaling Long-Horizon LLM Agent via Context-Foldingcited by 136
  9. 2025MemOS: A Memory OS for AI Systemcited by 129
  10. 2025Evo-Memory: Benchmarking LLM Agent Test-time Learning with Self-Evolving Memorycited by 129
  11. 2025ACON: Optimizing Context Compression for Long-horizon LLM Agentscited by 126
  12. 2025Aegis2.0: A Diverse AI Safety Dataset and Risks Taxonomy for Alignment of LLM Guardrailscited by 122
  13. 2025A Survey of Context Engineering for Large Language Modelscited by 120
  14. 2025Memento: Fine-tuning LLM Agents without Fine-tuning LLMscited by 119
  15. 2023LawBench: Benchmarking Legal Knowledge of Large Language Modelscited by 111
  16. 2024HiAgent: Hierarchical Working Memory Management for Solving Long-Horizon Agent Tasks with Large Language Modelcited by 101
  17. 2025Recursive Language Modelscited by 94
  18. 2024Evolutionary Computation in the Era of Large Language Model: Survey and Roadmapcited by 89
  19. 2025Evolving Deeper LLM Thinkingcited by 58
  20. 2025MemoryBench: A Benchmark for Memory and Continual Learning in LLM Systemscited by 56
  21. 2025Training-Free Group Relative Policy Optimizationcited by 53
  22. 2025EvoFlow: Evolving Diverse Agentic Workflows On The Flycited by 49
  23. 2022FiNER: Financial Numeric Entity Recognition for XBRL Taggingcited by 47
  24. 2025Context Engineering 2.0: The Context of Context Engineeringcited by 17
  25. 2025Neural Architecture Discovery via Autonomous Evolutioncited by 15
  26. 2025From Code Foundation Models to Agents and Applications: A Comprehensive Survey and Practical Guide to Code Intelligencecited by 13
  27. 2025VRPAgent: LLM-Driven Discovery of Heuristic Operators for Vehicle Routing Problemscited by 6
  28. 2025AutoEP: LLMs-Driven Automation of Hyperparameter Evolution for Metaheuristic Algorithmscited by 4
  29. 2025From Understanding to Excelling: Template-Free Algorithm Design through Structural-Functional Co-Evolutioncited by 2
  30. 2025Beyond Algorithm Evolution: An LLM-Driven Framework for the Co-Evolution of Swarm Intelligence Optimization Algorithms and Promptscited by 2
  31. 2026JudgeFlow: Agentic Workflow Optimization via Block Judgecited by 2
  32. 2026LLM-as-RNN: A Recurrent Language Model for Memory Updates and Sequence Predictioncited by 1

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 42.

42 papers.

Sort:

By area:

  1. 2026Meta-Harness: End-to-End Optimization of Model Harnessescited by 248
  2. 2026Natural-Language Agent Harnessescited by 54
  3. 2026SkillNet: Create, Evaluate, and Connect AI Skillscited by 48
  4. 2026Building Effective AI Coding Agents for the Terminal: Scaffolding, Harness, Context Engineering, and Lessons Learnedcited by 18
  5. 2026SkillMOO: Multi-Objective Optimization of Agent Skills for Software Engineeringcited by 14
  6. 2026MetaSkill-Evolve: Recursive Self-Improvement of LLM Agents via Two-Timescale Meta-Skill Evolutioncited by 12
  7. 2026Harness Engineering for Agentic AI Coding Tools: An Exploratory Studycited by 10
  8. 2026Harness-Aware Self-Evolving: Co-Evolving Model Weights, Harness, and Task Solutionscited by 10
  9. 2026From Question Answering to Task Completion: A Survey on Agent System and Harness Designcited by 9
  10. 2026WikiSkill: Compiling Agent Experience into Persistent Knowledge for Skill Evolutioncited by 7
  11. 2026DemoEvolve: Overcoming Sparse Feedback in Agentic Harness Evolution with Demonstrationscited by 5
  12. 2026Co-Evolving Skill Generation and Policy Optimizationcited by 5
30 more
  1. 2026When Does Continual Learning Require Learningcited by 4
  2. 2026DREvo: Distilling Recalibrated Historical Experience for Harness Self-Evolutioncited by 4
  3. 2026Agent Skill Framework: Perspectives on the Potential of Small to Medium Language Models in Industrial Environmentscited by 3
  4. 2026A Self-Evolving Agentic Framework for Metasurface Inverse Designcited by 3
  5. 2026What Do Evolutionary Coding Agents Evolve?cited by 3
  6. 2026Organize then Retrieve: Hierarchical Memory Navigation for Efficient Agentscited by 3
  7. 2026Task-CoEvolve: Efficient Harness Optimization via Adaptive Validation Task Selectioncited by 3
  8. 2026Harness-Zero: Harness Distillation via Agent-as-Harnesscited by 3
  9. 2026Federation over Text: Insight Sharing for Multi-Agent Reasoningcited by 2
  10. 2026Agent-ValueBench: A Comprehensive Benchmark for Evaluating Agent Valuescited by 2
  11. 2026ConMem: Structured Memory-Guided Adaptation in Training-Free Multi-Agent Systemscited by 1
  12. 2026How Much Heavy Lifting Can an Agent Harness Do?: Measuring the LLM's Residual Role in a Planning Agentcited by 0
  13. 2026Safe Bilevel Delegation (SBD): A Formal Framework for Runtime Delegation Safety in Multi-Agent Systemscited by 0
  14. 2026KACE: Knowledge-Adaptive Context Engineering for Mathematical Reasoningcited by 0
  15. 2026Skill-3D: Evolving Scene-Aware Skills for Agentic 3D Spatial Reasoningcited by 0
  16. 2026SAGE: Stochastic Prompt Optimization via Agent-Guided Explorationcited by 0
  17. 2026Probe-and-Refine Tuning of Repository Guidance for Coding Agentscited by 0
  18. 2026Knowledge-Centric Self-Improvementcited by 0
  19. 2026A Vocabulary for Multi-Agent Automated Research Systemscited by 0
  20. 2026MemeMind: Reference-Guided Trace Construction for Offline Context Optimizationcited by 0
  21. 2026Towards a Systems Foundation for Agentic Skills: Architecture, Lifecycle, and Securitycited by 0
  22. 2026CoSkill: Joint Reinforcement Learning of Reasoning and Meta-Skill Agents for Hierarchical Skill Evolutioncited by 0
  23. 2026Failure-Guided Co-Evolution of Prompts and Training Datacited by 0
  24. 2026SelfOp: An Optimization Algorithm for Self-Improving Security Agentscited by 0
  25. 2026ExpVoyager: Direct Experience Navigation for Dynamic Agent Skill Synthesiscited by 0
  26. 2026Certified Long-Horizon Code Agent Evolution via Validation-Gated Skill Optimizationcited by 0
  27. 2026SkillFocus: Evolving Agent Skills via Capability Decompositioncited by 0
  28. 2026Context Language Modelscited by 0
  29. 2026Learning Meta-Skills for Agent Harness Design in Test-Time AI4AIcited by 0
  30. 2026Turbo Harness: Instance-Adaptive Harness Optimizationcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Agent Skills for Large Language Models: Architecture, Acquisition, Security, and the Path Forward
  2. 2026MetaEvo: A Meta-Optimization Framework for Experience-Driven Agent Evolution
  3. 2026Beyond Skill Evolution: Self-Evolving Context Management Policies for Long-Horizon Agent Harnesses
  4. 2026ContextEvolve: Multi-Agent Context Compression for Systems Code Optimization
  5. 2026Unified Context Evolution for LLM Agents
  6. 2026Mise en Place for Agentic Coding: Deliberate Preparation as Context Engineering Methodology
  7. 2026Skill-MAS: Evolving Meta-Skill for Automatic Multi-Agent Systems
  8. 2025Context as a Tool: Context Management for Long-Horizon SWE-Agents