Skip to content
Back to search

Awesome-Self-Evolving-Agents

A Survey of Self-Evolving Agents | A curated list of resources (surveys, papers, benchmarks, and opensource projects) on Self-Evolving Agents.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today

Check it deeper on your machine

It has changed since its scan on 11 Nov 2026.

As a Claude Code plugin, install it and then ask Claude Code:

/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/XMUDeepLIT/Awesome-Self-Evolving-Agents with scanskillsafety

Or from a terminal, clone it and run it:

git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/XMUDeepLIT/Awesome-Self-Evolving-Agents'

It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/xmudeeplit--awesome-self-evolving-agents/badge.svg)](/listings/xmudeeplit--awesome-self-evolving-agents)

HTML

<a href="/listings/xmudeeplit--awesome-self-evolving-agents"><img src="/api/listings/xmudeeplit--awesome-self-evolving-agents/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 665 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.

  1. # Awesome-Self-Evolving-Agents
  2. <div align="center">
  3. <a href="https://awesome.re"><img src="https://awesome.re/badge.svg"/></a>
  4. <a href="http://makeapullrequest.com"><img src="https://img.shields.io/badge/PRs-welcome-green.svg"/></a>
  5. <a href="https://doi.org/10.36227/techrxiv.177203250.05832634/v2" target="_blank"><img src="https://img.shields.io/badge/Paper-Arxiv-red?logo=arxiv&style=flat-square" alt="arXiv:2602.05665"></a>
  6. <a href="http://makeapullrequest.com"><img src="https://img.shields.io/github/stars/XMUDeepLIT/Awesome-Self-Evolving-Agents"/></a>
  7. </div>
  8. This repository provides a comprehensive collection of research papers, benchmarks, and open-source projects on **Self-Evolving Agents**. It includes contents from our survey paper 📖<em>"[**A Systematic Survey of Self-Evolving Agents: From Model-Centric to Environment-Driven Co-Evolution**](https://doi.org/10.36227/techrxiv.177203250.05832634/v2)"</em> and will be continuously updated.
  9. 🤗 **You're very welcome to contribute to this repository**. If you find any missing resources or come across interesting new research works, please don’t hesitate to launch an issue or submit a pull request!
  10. 📫 **Contact us via emails:** `{xiangzhishang,yangchengyi}@stu.xmu.edu.cn`, `qinggangzhang@jlu.edu.cn`
  11. **📃 Please [cite our paper](#-citation)** if you find our survey or repository helpful!
  12. ```
  13. @article{xiang2026systematic,
  14. title={A Systematic Survey of Self-Evolving Agents: From Model-Centric to Environment-Driven Co-Evolution},
  15. author={Xiang, Zhishang and Yang, Chengyi and Chen, Zerui and Wei, Zhimin and Tang, Yunbo and Teng, Zongpei and Peng, Zexi and Li, Zongxia and Huang, Chengsong and He, Yicheng and others},
  16. journal={Available at SSRN 6626878},
  17. year={2026}
  18. }
  19. ```
  20. ---
  21. # 🎉 News
  22. - **[2026-10]** [TTCS](https://arxiv.org/abs/2601.22628) has been accepted to [NeurIPS 2026](https://neurips.cc/virtual/2026/loc/sydney/poster/154412)!
  23. - **[2026-03]** We release the [TTCS](https://arxiv.org/abs/2601.22628) and accepted by [ICLR 2026 Lifelong Agent(LLA)](https://iclr.cc/virtual/2026/workshop/10000805) workshop!
  24. - **[2026-02]** We release the survey of [Self-Evolving Agents](https://github.com/XMUDeepLIT/Awesome-Agentic-Self-Evolution).
  25. ---
  26. <div>
  27. <h3 align="center">
  28. <p align="center"><img width="100%" src="figs/intro.png" /></p>
  29. <p align="center"><em>A Comprehensive Taxonomy of Self-Evolving Agents.</em></p>
  30. </div>
  31. Agentic Self-Evolving represents a paradigm shift in AI development, enabling systems to autonomously improve through three key dimensions:
  32. - **Model-Centric Self-Evolution**: Focuses on improving the model itself through inference-based evolution (parallel sampling, sequential self-correction, structured reasoning) and training-based evolution (synthesis-driven offline and exploration-driven online self-evolving).
  33. - **Environment-Centric Self-Evolution**: Enhances the agent's interaction with external knowledge and experience through static knowledge evolution, dynamic experience evolution, modular architecture evolution, and agentic topology evolution.
  34. - **Model-Environment Co-Evolution**: Enables simultaneous evolution of both the model and its environment through environment training and multi-agent policy co-evolution.
  35. ## 📈 Trends
  36. <div>
  37. <h3 align="center">
  38. <p align="center"><img width="100%" src="figs/trend.png" /></p>
  39. <p align="center"><em>The Development Trends of Self-Evolving Agents with Representative Works.</em></p>
  40. </div>
  41. ---
  42. ## Table of Content
  43. - [🔥 News](#-news)
  44. - [📚 Related Survey Papers](#-related-survey-papers)
  45. - [📜 Research Papers](#-research-papers)
  46. - [Model-Centric Self-Evolution](#model-centric-self-evolving)
  47. - [Inference-Based Evolution](#inference-based-evolution)
  48. - [Parallel Sampling](#parallel-sampling)
  49. - [Sequential Self-Correction](#sequential-self-correction)
  50. - [Structured Reasoning](#structured-reasoning)
  51. - [Training-Based Evolution 🔥](#training-based-evolution)
  52. - [Synthesis-Driven Offline Self-Evolving](#synthesis-driven-offline-self-evolving)
  53. - [Exploration-Driven Online Self-Evolving 🔥](#exploration-driven-online-self-evolving)
  54. - [Environment-Centric Self-Evolution](#environment-centric-self-evolving)
  55. - [Static Knowledge Evolution](#static-knowledge-evolution)
  56. - [Agentic Retrieval-Augmented Generation](#agentic-retrieval-augmented-generation)
  57. - [Reasoning-Driven Deep Research](#reasoning-driven-deep-research)
  58. - [Dynamic Experience Evolution 🔥](#dynamic-experience-evolution)
  59. - [Offline Experience Compilation](#offline-experience-compilation)
  60. - [Online Experience Adaptation](#online-experience-adaptation)
  61. - [Lifelong Experience Evolution 🔥](#lifelong-experience-evolution)
  62. - [Skill Augmented Evolution 🔥](#skill-augmented-evolution)
  63. - [Modular Architecture Evolution](#modular-architecture-evolution)
  64. - [Interaction Protocol Evolution](#interaction-protocol-evolution)
  65. - [Memory Architecture Evolution](#memory-architecture-evolution)
  66. - [Tool-Augmented Evolution](#tool-augmented-evolution)
  67. - [Agentic Topology Evolution](#agentic-topology-evolution)
  68. - [Offline Architecture Search](#offline-architecture-search)
  69. - [Runtime Dynamic Adaptation](#runtime-dynamic-adaptation)
  70. - [Structural Memory Evolution](#structural-memory-evolution)
  71. - [Model-Environment Co-Evolution 🔥](#model-environment-co-evolving)
  72. - [Multi-Agent Policy Co-Evolution](#multi-agent-policy-co-evolution)
  73. - [Environment Training 🔥](#environment-training)
  74. - [Adaptive Curriculum Evolution](#adaptive-curriculum-evolution)
  75. - [Scalable Environment Evolution](#scalable-environment-evolution)
  76. - [🏆 Benchmarks](#-benchmarks)
  77. - [Self-Evolution Evaluation](#self-evolution-evaluation)
  78. - [Intrinsic Capabilities](#intrinsic-capabilities)
  79. - [Agentic Reasoning Capabilities](#agentic-reasoning-capabilities)
  80. - [💻 Open Source Libraries](#-open-source-libraries)
  81. - [🚀 Applications](#-applications)
  82. - [Automated Scientific Discovery](#automated-scientific-discovery)
  83. - [Autonomous Software Engineering](#autonomous-software-engineering)
  84. - [Open-World Simulation](#open-world-simulation)
  85. - [🍀 Citation](#-citation)
  86. # 📚 Related Survey Papers
  87. - (arXiv 2024) A Survey on Self-Evolution of Large Language Models [[Paper]](https://arxiv.org/abs/2404.14387)
  88. - (TMLR 2026) A Survey of Self-Evolving Agents What, When, How, and Where to Evolve on the Path to Artificial Super Intelligence [[Paper]](https://arxiv.org/abs/2507.21046)
  89. - (arXiv 2025) A Comprehensive Survey of Self-Evolving AI Agents: A New Paradigm Bridging Foundation Models and Lifelong Agentic Systems [[Paper]](https://arxiv.org/abs/2508.07407)
  90. - (arXiv 2026) Safety in Embodied AI: A Survey of Risks, Attacks, and Defenses [[Paper]](https://arxiv.org/abs/2605.02900)
  91. - (SSRN 2026) The Agent Loop: A Survey of Control Strategies, Skills, and Harnesses for LLM Agents [[Paper]](https://ssrn.com/abstract=7186738)
  92. # 📜 Research Papers
  93. ## Model-Centric Self-Evolution
  94. ### Inference-Based Evolution
  95. #### Parallel Sampling
  96. - (ICLR'23) Self-consistency improves chain of thought reasoning in language models [[Paper]](https://arxiv.org/abs/2203.11171)
  97. - (NeurIPS'24) Are more llm calls all you need? towards scaling laws of compound inference systems [[Paper]](https://proceedings.neurips.cc/paper_files/paper/2024/hash/51173cf34c5faac9796a47dc2fdd3a71-Abstract-Conference.html)
  98. - (arxiv'24) Large language monkeys: Scaling inference compute with repeated sampling [[Paper]](https://arxiv.org/abs/2407.21787)
  99. - (arxiv'24) Scaling llm test-time compute optimally can be more effective than scaling model parameters [[Paper]](https://arxiv.org/abs/2408.03314)
  100. - (ACL'23) Llm-blender: Ensembling large language models with pairwise ranking and generative fusion [[Paper]](https://aclanthology.org/2023.acl-long.792/)
  101. - (EMNLP'23) Selfcheckgpt: Zero-resource black-box hallucination detection for generative large language models [[Paper]](https://aclanthology.org/2023.emnlp-main.557/)
  102. #### Sequential Self-Correction
  103. - (NeurIPS'23) Self-refine: Iterative refinement with self-feedback [[Paper]](https://proceedings.neurips.cc/paper_files/paper/2023/hash/91edff07232fb1b55a505a9e9f6c0ff3-Abstract-Conference.html)
  104. - (ICLR'24) Teaching large language models to self-debug [[Paper]](https://arxiv.org/abs/2304.05128)
  105. - (ACL'24 findings) Small language models need strong verifiers to self-correct reasoning [[Paper]](https://aclanthology.org/2024.findings-acl.924/)
  106. - (NeurIPS'23) Reflexion: Language agents with verbal reinforcement learning [[Paper]](https://proceedings.neurips.cc/paper_files/paper/2023/hash/1b44b878bb782e6954cd888628510e90-Abstract-Conference.html)
  107. - (ICLR'24) Critic: Large language models can self-correct with tool-interactive critiquing [[Paper]](https://arxiv.org/abs/2305.11738)
  108. - (arxiv'25) Evolving deeper llm thinking [[Paper]](https://arxiv.org/abs/2501.09891)
  109. - (arxiv'25) Towards system 2 reasoning in llms: Learning how to think with meta chain-of-thought [[Paper]](https://arxiv.org/abs/2501.04682)
  110. - (ICLR'25) Planning in natural language improves llm search for code generation [[Paper]](https://arxiv.org/abs/2409.03733)
  111. - (ICML'25) Reasoning-as-logic-units: Scaling test-time reasoning in large language models through logic unit alignment [[Paper]](https://arxiv.org/abs/2502.07803)
  112. - (arxiv'26) Steer, Don't Solve: Training Small Critic Models for Large Code Agents [[Paper]](https://arxiv.org/abs/2606.21811) [[Code]](https://github.com/shubhamrgandhi/critic-training)
  113. #### Structured Reasoning
  114. - (arxiv'24) Stream of search (sos): Learning to search in language [[Paper]](https://arxiv.org/abs/2404.03683)
  115. - (NeurIPS'24) Toward self-improvement of llms via imagination, searching, and criticizing [[Paper]](https://proceedings.neurips.cc/paper_files/paper/2024/hash/5e5853f35164e434015716a8c2a66543-Abstract-Conference.html)
  116. - (ICML'24) Alphazero-like tree-search can guide large language model decoding and training [[Paper]](https://arxiv.org/abs/2309.17179)
  117. - (arxiv'24) Litesearch: Efficacious tree search for llm [[Paper]](https://arxiv.org/abs/2407.00320)
  118. - (NeurIPS 2023) Tree of thoughts: Deliberate problem solving with large language models, [[Paper]](https://proceedings.neurips.cc/paper/2023/hash/271db9922b8d1f4dd7aaef84ed5ac703-Abstract.html)
  119. - (ICML'24) Language agent tree search unifies reasoning acting and planning in language models [[Paper]](https://arxiv.org/abs/2310.04406)
  120. - (ICLR'24) Think-on-graph: Deep and responsible reasoning of large language model on knowledge graph [[Paper]](https://arxiv.org/abs/2307.07697)
  121. - (ICLR'25) Think-on-graph 2.0: Deep and faithful large language model reasoning with knowledge-guided retrieval augmented generation [[Paper]](https://arxiv.org/abs/2407.10805)
  122. - (ICLR'24) Reasoning on graphs: Faithful and interpretable large language model reasoning [[Paper]](https://arxiv.org/abs/2310.01061)
  123. ### Training-Based Evolution 🔥
  124. <h3 align="center">
  125. <p align="center"><img width="100%" src="figs/model-centric.png" /></p>
  126. #### Synthesis-Driven Offline Self-Evolving
  127. - (TMLR'24) Beyond human data: Scaling self-training for problem-solving with language models [[Paper]](https://arxiv.org/abs/2312.06585)
  128. - (NeurIPS'25 workshop) Learning to solve and verify: A self-play framework for code and test generation [[Paper]](https://arxiv.org/abs/2502.14948)
  129. - (arxiv'25) Evolving language models without labels: Majority drives selection, novelty promotes variation [[Paper]](https://arxiv.org/abs/2509.15194)
  130. - (arxiv'25) Test-time offline reinforcement learning on goal-related experience [[Paper]](https://arxiv.org/abs/2507.18809)
  131. - (arxiv'25) Agent-R: Training Language Model Agents to Reflect via Iterative Self-Training [[Paper]](https://arxiv.org/abs/2501.11425)
  132. - (ACL'23) Self-instruct: Aligning language models with self-generated instructions [[Paper]](https://aclanthology.org/2023.acl-long.754.pdf)
  133. - (COLM'24) Self-guide: Better task-specific instruction following via self-synthetic finetuning [[Paper]](https://arxiv.org/abs/2407.12874)
  134. - (NeurIPS'25) Self-Adapting Language Models [[Paper]](https://arxiv.org/abs/2506.10943)
  135. - (arxiv'23) Self: Self-evolution with language feedback [[Paper]](https://arxiv.org/abs/2310.00533)
  136. - (ICML'24) Self-play fine-tuning converts weak language models to strong language models [[Paper]](https://arxiv.org/abs/2401.01335)
  137. - (ICLR'25) Self-play preference optimization for language model alignment [[Paper]](https://arxiv.org/abs/2405.00675)
  138. - (NeurIPS'22) Star: Bootstrapping reasoning with reasoning [[Paper]](https://proceedings.neurips.cc/paper_files/paper/2022/hash/639a9a172c044fbb64175b5fad42e9a5-Abstract-Conference.html)
  139. - (EMNLP'23) Large language models can self-improve [[Paper]](https://aclanthology.org/2023.emnlp-main.67/)
  140. - (NeurIPS'24) Rest-mcts*: Llm self-training via process reward guided tree search [[Paper]](https://proceedings.neurips.cc/paper_files/paper/2024/hash/76ec4dc30e9faaf0e4b6093eaa377218-Abstract-Conference.html)
  141. - (ICML'25) Self-improving language models for evolutionary program synthesis: A case study on ARC-AGI [[Paper]](https://icml.cc/virtual/2025/poster/43499)
  142. - (NeurIPS'25) Sirius: Self-improving multi-agent systems via bootstrapped reasoning [[Paper]](https://neurips.cc/virtual/2025/loc/san-diego/poster/118834)
  143. - (arxiv'25) Ragen: Understanding self-evolution in llm agents via multi-turn reinforcement learning [[Paper]](https://arxiv.org/abs/2504.20073)
  144. - (EMNLP'25) Samule: Self-learning agents enhanced by multi-level reflection [[Paper]](https://aclanthology.org/2025.emnlp-main.839/)
  145. - (arxiv'26) Harness-Zero: Harness Distillation via Agent-as-Harness [[Paper]](https://arxiv.org/abs/2609.24974) [[Code]](https://github.com/metaevo-ai/harness-zero)
  146. - (arxiv'26) ReSAIL: Mitigating Collapse in Iterative Agent Self-Distillation [[Paper]](https://arxiv.org/abs/2609.39306) [[Code]](https://github.com/ShengjieJin/ReSAIL)
  147. #### Exploration-Driven Online Self-Evolving 🔥
  148. - (arxiv'26) Learning from Interaction Experience via Localized Context Distillation [[Paper]](https://arxiv.org/abs/2610.21427) [[Code]](https://github.com/zz-haooo/LoCD)
  149. - (arxiv'26) Environmental Feedback Modeling Matters: Rethinking Feedback Treatment in Agentic Hindsight Self-Distillation [[Paper]](https://arxiv.org/abs/2610.11384)
  150. - (ECCV'2026) SyncLoop: A Multimodal Dual-Loop Framework for Self-Improving Mathematical Reasoning [[Paper]](https://arxiv.org/abs/2507.16518)
  151. - (NeurIPS'26) TTCS: Test-Time Curriculum Synthesis for Self-Evolving [[Paper]](https://arxiv.org/abs/2601.22628) [[Poster]](https://neurips.cc/virtual/2026/loc/sydney/poster/154412) [[Code]](https://github.com/XMUDeepLIT/TTCS)
  152. - (NeurIPS'25) Co-evolving llm coder and unit tester via reinforcement learning [[Paper]](neurips.cc/virtual/2025/loc/san-diego/poster/115329)
  153. - (arxiv'25) Towards agentic self-learning llms in search environment [[Paper]](https://arxiv.org/abs/2510.14253)
  154. - (arxiv'25) Multi-agent evolve: Llm self-improve through co-evolution [[Paper]](https://arxiv.org/abs/2510.23595)
  155. - (ICLR'26) Steporlm: A self-evolving framework with generative process supervision for operations research language models [[Paper]](https://arxiv.org/abs/2509.22558)
  156. - (NeurIPS'24) Coevolving with the other you: Fine-tuning llm with sequential cooperative multi-agent reinforcement learning [[Paper]](https://neurips.cc/virtual/2024/poster/95347)
  157. - (ICLR'26) R-zero: Self-evolving reasoning llm from zero data [[Paper]](https://arxiv.org/abs/2508.05004)
  158. - (COLM'26) Vocabulary Dropout for Curriculum Diversity in LLM Co-Evolution [[Paper]](https://arxiv.org/abs/2604.03472) [[Code]](https://github.com/ARC-ASU/vocab_dropout)
  159. - (arxiv'26) MM-Zero: Self-Evolving Multi-Model Vision Language Models From Zero Data
  160. [[Paper]](https://arxiv.org/abs/2603.09206)
  161. - (NeurIPS'25) Absolute zero: Reinforced self-play reasoning with zero data [[Paper]](https://neurips.cc/virtual/2025/loc/san-diego/poster/116121)
  162. - (arxiv'25) Language self-play for data-free training [[Paper]](https://arxiv.org/abs/2509.07414)
  163. - (arxiv'25) Self-questioning language models [[Paper]](https://arxiv.org/abs/2508.03682)
  164. - (arxiv'25) SPIRAL: Self-Play on Zero-Sum Games Incentivizes Reasoning via Multi-Agent Multi-Turn Reinforcement Learning [[Paper]](https://arxiv.org/abs/2506.24119)
  165. - (arxiv'25) Socratic-zero: Bootstrapping reasoning via data-free agent co-evolution [[Paper]](https://arxiv.org/abs/2509.24726)
  166. - (arxiv'25) Agent0: Unleashing self-evolving agents from zero data via tool-integrated reasoning [[Paper]](https://arxiv.org/abs/2511.16043)
  167. - (NeurIPS'25) SeRL: Self-Play Reinforcement Learning for Large Language Models with Limited Data [[Paper]](https://neurips.cc/virtual/2025/loc/san-diego/poster/117365)
  168. - (arxiv'26) Dr. Zero: Self-Evolving Search Agents without Training Data [[Paper]](https://arxiv.org/abs/2601.07055)
  169. - (ICLR'26) Theoretical Modeling of LLM Self-Improvement Training Dynamics Through Solver-Verifier Gap [[Paper]](https://arxiv.org/abs/2507.00075)
  170. - (ICLR'26) Search self-play: Pushing the frontier of agent capability without supervision [[Paper]](https://arxiv.org/abs/2510.18821)
  171. - (arxiv'25) ReVeal: Self-Evolving Code Agents via Iterative Generation-Verification [[Paper]](https://arxiv.org/abs/2506.11442)
  172. - (NeurIPS'25) Self-challenging language model agents [[Paper]](https://neurips.cc/virtual/2025/loc/san-diego/poster/119495)
  173. - (ICLR'26) Spice: Self-play in corpus environments improves reasoning [[Paper]](https://arxiv.org/abs/2510.24684)
  174. - (ICLR'26) Spell: Self-play reinforcement learning for evolving long-context language models [[Paper]](https://arxiv.org/abs/2509.23863)
  175. - (ICLR'25) Webrl: Training llm web agents via self-evolving online curriculum reinforcement learning [[Paper]](https://arxiv.org/abs/2411.02337)
  176. - (EMNLP'25) Enabling Self-Improving Agents to Learn at Test Time With Human-In-The-Loop Guidance [[Paper]](https://aclanthology.org/2025.emnlp-industry.115/)
  177. - (arxiv'25) Ladder: Self-improving llms through recursive problem decomposition [[Paper]](https://arxiv.org/abs/2503.00735)
  178. - (arxiv'25) Guided self-evolving llms with minimal human supervision [[Paper]](https://arxiv.org/abs/2512.02472)
  179. - (AAAI'25) Empowering Self-Learning of LLMs: Inner Knowledge Explicitation as a Catalyst [[Paper]](https://ojs.aaai.org/index.php/AAAI/article/view/34590)
  180. - (NeurIPS'25) Spc: Evolving self-play critic via adversarial games for llm reasoning [[Paper]](https://neurips.cc/virtual/2025/loc/san-diego/poster/118706)
  181. - (ICLR'26) Your agent may misevolve: Emergent risks in self-evolving llm agents [[Paper]](https://arxiv.org/abs/2509.26354)
  182. - (arxiv'26) DRACO: Fine-Grained Credit Assignment with Dynamic Rubrics for Long-Horizon Agent Training [[Paper]](https://arxiv.org/abs/2609.04094) [[Code]](https://github.com/IBM/draco)
  183. - (arxiv'26) Training Agents to Evolve with Their Harness: TaoLive Digital Avatar Agent Technical Report [[Paper]](https://arxiv.org/abs/2608.15763)
  184. ## Environment-Centric Self-Evolution
  185. <h3 align="center">
  186. <p align="center"><img width="100%" src="figs/exp-driven.png" /></p>
  187. ### Static Knowledge Evolution
  188. #### Agentic Retrieval-Augmented Generation
  189. - (ICLR'24) Self-rag: Learning to retrieve, generate, and critique through self-reflection [[Paper]](https://arxiv.org/abs/2310.11511)
  190. - (ACL'25) Rag-critic: Leveraging automated critic-guided agentic workflow for retrieval augmented generation [[Paper]](https://aclanthology.org/2025.acl-long.179/)
  191. - (EMNLP'25) Search-o1: Agentic search-enhanced large reasoning models [[Paper]](https://aclanthology.org/2025.emnlp-main.276.pdf)
  192. - (COLM 2025) Search-r1: Training llms to reason and leverage search engines with reinforcement learning [[Paper]](https://arxiv.org/abs/2503.09516)
  193. - (arxiv'25) Learning to reason with search for llms via reinforcement learning [[Paper]](https://arxiv.org/abs/2503.19470)
  194. - (arxiv'25) R1-searcher: Incentivizing the search capability in llms via reinforcement learning [[Paper]](https://arxiv.org/abs/2503.05592)
  195. - (ACL'23) Webcpm: Interactive web search for chinese long-form question answering [[Paper]](https://aclanthology.org/2023.acl-long.499/)
  196. - (NeurIPS'23 workshop)Dspy: Compiling declarative language model calls into self-improving pipelines [[Paper]](https://arxiv.org/abs/2310.03714)
  197. - (ICLR'25) Mindsearch: Mimicking human minds elicits deep ai searcher [[Paper]](https://arxiv.org/abs/2407.20183)
  198. - (arxiv'25) Open deep search: Democratizing search with open-source reasoning agents [[Paper]](https://arxiv.org/abs/2503.20201)
  199. #### Reasoning-Driven Deep Research
  200. - (arxiv'25) DeepResearch Bench: A Comprehensive Benchmark for Deep Research Agents [[Paper]](https://arxiv.org/abs/2506.11763)
  201. - (arxiv'25) Deep research agents: A systematic examination and roadmap [[Paper]](https://arxiv.org/abs/2506.18096)
  202. - (ACL'24) Webvoyager: Building an end-to-end web agent with large multimodal models [[Paper]](https://aclanthology.org/anthology-files/pdf/acl/2024.acl-long.371.pdf)
  203. - (arxiv'25) Deepdive: Advancing deep search agents with knowledge graphs and multi-turn rl [[Paper]](https://arxiv.org/abs/2509.10446)
  204. - (arxiv'25) DeepSearch: Overcome the Bottleneck of Reinforcement Learning with Verifiable Rewards via Monte Carlo Tree Search [[Paper]](https://arxiv.org/abs/2509.25454)
  205. - (arxiv'25) Hiersearch: A hierarchical enterprise deep search framework integrating local and web searches [[Paper]](https://arxiv.org/abs/2508.08088)
  206. - (EMNLP'25) Deepresearcher: Scaling deep research via reinforcement learning in real-world environments [[Paper]](https://aclanthology.org/2025.emnlp-main.22/)
  207. - (arxiv'25) Sfr-deepresearch: Towards effective reinforcement learning for autonomously reasoning single agents [[Paper]](https://arxiv.org/abs/2509.06283)
  208. - (arxiv'25) Tongyi deepresearch technical report [[Paper]](https://arxiv.org/abs/2510.24701)
  209. - (arxiv'25) Webweaver: Structuring web-scale evidence with dynamic outlines for open-ended deep research [[Paper]](https://arxiv.org/abs/2509.13312)
  210. - (arxiv'25) Webthinker: Empowering large reasoning models with deep research capability [[Paper]](https://arxiv.org/abs/2504.21776)
  211. - (arxiv'25) Surveyx: Academic survey automation via large language models [[Paper]](https://arxiv.org/abs/2502.14776)
  212. ### Dynamic Experience Evolution 🔥
  213. #### Offline Experience Compilation
  214. - (arxiv'26) TabAgent: Distilling LLM Agent Components into Cheap Classifiers [[Paper]](https://arxiv.org/abs/2602.16429) [[Code]](https://github.com/dolev31/TabAgent)
  215. - (arxiv'26) Who Remembers What? Multi-Granularity Memory Extraction for Multi-User, Multi-Agent Interaction [[Paper]](https://arxiv.org/abs/2611.10439)
  216. - (ICSE'27) Learning from Repair Experience: Self-Evolving Agents for Automated Vulnerability Repair [[Paper]](https://arxiv.org/abs/2605.30105) [[Code]](https://github.com/iSEngLab/Learn-from-Repair-Experience)
  217. - (arxiv'26) Recursive Harness Distillation across Agents for Robot Manipulation [[Paper]](https://arxiv.org/abs/2609.33378)
  218. - (arxiv'25) Remember me, refine me: A dynamic procedural memory framework for experience-driven agent evolution [[Paper]](https://arxiv.org/abs/2512.10696)
  219. - (arxiv'25) Get Experience from Practice: LLM Agents with Record & Replay [[Paper]](https://arxiv.org/abs/2505.17716)
  220. - (arxiv'25) Skillweaver: Web agents can self-improve by discovering and honing skills [[Paper]](https://arxiv.org/abs/2504.07079)
  221. - (ICML'25) Agent kb: Leveraging cross-domain experience for agentic problem solving [[Paper]](https://arxiv.org/abs/2507.06229)
  222. - (arxiv'25) From experience to strategy: Empowering llm agents with trainable graph memory [[Paper]](https://arxiv.org/abs/2511.07800)
  223. - (arxiv'24) Cops: Empowering llm agents with provable cross-task experience sharing [[Paper]](https://arxiv.org/abs/2410.16670)
  224. - (arxiv'25) Arcmemo: Abstract reasoning composition with lifelong llm memory [[Paper]](https://arxiv.org/abs/2509.04439)
  225. #### Online Experience Adaptation
  226. - (arxiv'26) ExpSeek: Self-Triggered Experience Seeking for Web Agents [[Paper]](https://arxiv.org/abs/2601.08605)
  227. - (arxiv'25) Dynamic cheatsheet: Test-time learning with adaptive memory [[Paper]](https://arxiv.org/abs/2504.07952)
  228. - (arxiv'25) Memento: Fine-tuning llm agents without fine-tuning llms [[Paper]](https://arxiv.org/abs/2508.16153)
  229. - (ICLR'26) Gepa: Reflective prompt evolution can outperform reinforcement learning [[Paper]](https://arxiv.org/abs/2507.19457)
  230. - (ICLR'26) Agentic context engineering: Evolving contexts for self-improving language models [[Paper]](https://arxiv.org/abs/2510.04618)
  231. - (ECCV'26) AnchorGUI: Asymmetric Memory for Dual-Scale Learning in GUI Navigation [[Paper]](https://arxiv.org/abs/2609.15457)
  232. #### Lifelong Experience Evolution 🔥
  233. - (arxiv'26) From a Prompt to Repertoires: Evolving Functional REpertoires Enable LLM Continual Learning [[Paper]](https://arxiv.org/abs/2610.11373)
  234. - (arxiv'25) Xolver: Multi-Agent Reasoning with Holistic Experience Learning Just Like an Olympiad Team [[Paper]](https://arxiv.org/abs/2506.14234)
  235. - (arxiv'25) Agentevolver: Towards efficient self-evolving agent system [[Paper]](https://arxiv.org/abs/2511.10395)
  236. - (arxiv'26) Evolving Programmatic Skill Networks [[Paper]](https://arxiv.org/abs/2601.03509)
  237. - (arxiv'26) Meta Context Engineering via Agentic Skill Evolution [[Paper]](https://arxiv.org/abs/2601.21557)
  238. - (arxiv'26) Self-Consolidation for Self-Evolving Agents [[Paper]](https://arxiv.org/abs/2602.01966)
  239. - (ICLR'26) Reasoningbank: Scaling agent self-evolving with reasoning memory [[Paper]](https://arxiv.org/abs/2509.25140)
  240. - (arxiv'25) Evolver: Self-evolving llm agents through an experience-driven lifecycle [[Paper]](https://arxiv.org/abs/2510.16079)
  241. - (arxiv'25) Training-free group relative policy optimization [[Paper]](https://arxiv.org/abs/2510.08191)
  242. - (arxiv'25) Flex: Continuous agent evolution via forward learning from experience [[Paper]](https://arxiv.org/abs/2511.06449)
  243. - (arxiv'25) Agent learning via early experience [[Paper]](https://arxiv.org/abs/2510.08558)
  244. - (COLM'25)Inducing programmatic skills for agentic tasks [[Paper]](https://arxiv.org/abs/2504.06821)
  245. - (arxiv'25) Reinforcement learning for self-improving agent with skill library [[Paper]](https://arxiv.org/abs/2512.17102)
  246. - (arxiv'25) AccelOpt: A Self-Improving LLM Agentic System for AI Accelerator Kernel Optimization [[Paper]](https://arxiv.org/abs/2511.15915)
  247. - (ICLR'26) Memgen: Weaving generative latent memory for self-evolving agents [[Paper]](https://arxiv.org/abs/2509.24704)
  248. - (arxiv'25) Latentevolve: Self-evolving test-time scaling in latent space [[Paper]](https://arxiv.org/abs/2509.24771)
  249. - (arxiv'26) RetroAgent: From Solving to Evolving via Retrospective Dual Intrinsic Feedback [[Paper]](https://arxiv.org/abs/2603.08561)
  250. #### Skill Augmented Evolution 🔥
  251. - (arxiv'26) RewardHarness: Self-Evolving Agentic Post-Training [[Paper]](https://arxiv.org/abs/2605.08703) [[Code]](https://github.com/TIGER-AI-Lab/RewardHarness)
  252. - (arxiv'26) AutoSkill: Experience-Driven Lifelong Learning via Skill Self-Evolution
  253. [[Paper]](https://arxiv.org/abs/2603.01145)
  254. - (arxiv'26) SkillRL: Evolving Agents via Recursive Skill-Augmented Reinforcement Learning [[Paper]](https://arxiv.org/abs/2602.08234)
  255. - (arxiv'26) SkillNet: Create, Evaluate, and Connect AI Skills [[Paper]](https://arxiv.org/abs/2603.04448)
  256. - (arxiv'26) SkillsBench: Benchmarking How Well Agent Skills Work Across Diverse Tasks [[Paper]](https://arxiv.org/abs/2602.12670)
  257. - (arxiv'26) ARISE: Agent Reasoning with Intrinsic Skill Evolution in Hierarchical Reinforcement Learning [[Paper]](https://arxiv.org/abs/2603.16060)
  258. - (arxiv'26) Evolving Medical Imaging Agents via Experience-driven Self-skill Discovery [[Paper]](https://arxiv.org/abs/2603.05860)
  259. - (arxiv'26) OpenSkill: Open-World Self-Evolution for LLM Agents [[Paper]](https://arxiv.org/abs/2606.06741) [[Code]](https://github.com/OpenLAIR/OpenSkill)
  260. - (arxiv'26) SKILL-KD: Contrastive Skill Distillation for LLM Agents [[Paper]](https://arxiv.org/abs/2607.28048) [[Code]](https://github.com/KhanCold/SKILL-KD)
  261. - (ICML'26) Meta Context Engineering via Agentic Skill Evolution [[Paper]](https://arxiv.org/abs/2601.21557) [[Code]](https://github.com/metaevo-ai/meta-context-engineering)
  262. ### Modular Architecture Evolution
  263. #### Interaction Protocol Evolution
  264. - (arxiv'23) Think-in-memory: Recalling and post-thinking enable llms with long-term memory [[Paper]](https://arxiv.org/abs/2311.08719)
  265. - (EMNLP'23) Mot: Memory-of-thought enables chatgpt to self-improve [[Paper]](https://aclanthology.org/2023.emnlp-main.392/)
  266. - (ICML'24) A human-inspired reading agent with gist memory of very long contexts [[Paper]](https://arxiv.org/abs/2402.09727)
  267. - (ICLR'26) Lightmem: Lightweight and efficient memory-augmented generation [[Paper]](https://arxiv.org/abs/2510.18866)
  268. - (arxiv'25) General agentic memory via deep research [[Paper]](https://arxiv.org/abs/2511.18423)
  269. - (arxiv'23) MemGPT: Towards LLMs as Operating Systems[[Paper]](https://arxiv.org/abs/2310.08560)
  270. - (AAAI'24) Memorybank: Enhancing large language models with long-term memory [[Paper]](https://ojs.aaai.org/index.php/AAAI/article/view/29946)
  271. - (ICLR'26) AgentFold: Long-Horizon Web Agents with Proactive Context Management [[Paper]](https://arxiv.org/abs/2510.24699)
  272. #### Memory Architecture Evolution
  273. - (ICML'25) M+: Extending MemoryLLM with Scalable Long-Term Memory [[Paper]](https://icml.cc/virtual/2025/poster/45431)
  274. - (arxiv'26) MemSkill: Learning and Evolving Memory Skills for Self-Evolving Agents [[Paper]](https://arxiv.org/abs/2602.02474)
  275. - (arxiv'26) MemRL: Self-Evolving Agents via Runtime Reinforcement Learning on Episodic Memory [[Paper]](https://arxiv.org/abs/2601.03192)
  276. - (NeurIPS'25) A-mem: Agentic memory for llm agents [[Paper]](https://arxiv.org/abs/2502.12110)
  277. - (arxiv'25) Mem0: Building production-ready ai agents with scalable long-term memory [[Paper]](https://arxiv.org/abs/2504.19413)
  278. - (NeurIPS'25) Cam: A constructivist view of agentic memory for llm-based reading comprehension [[Paper]](https://arxiv.org/abs/2510.05520)
  279. - (arxiv'25) Memory as action: Autonomous context curation for long-horizon agentic tasks [[Paper]](https://arxiv.org/abs/2510.12635)
  280. - (arxiv'25) Mem-α: Learning Memory Construction via Reinforcement Learning [[Paper]](https://arxiv.org/abs/2509.25911)
  281. - (arxiv'25) Memevolve: Meta-evolution of agent memory systems [[Paper]](https://arxiv.org/abs/2512.18746)
  282. - (arxiv'26) EvoRoute: Experience-Driven Self-Routing LLM Agent Systems [[Paper]](https://arxiv.org/abs/2601.02695)
  283. #### Tool-Augmented Evolution
  284. - (ICLR'26) Toward Effective Tool-Integrated Reasoning via Self-Evolved Preference Learning [[Paper]](https://arxiv.org/abs/2509.23285)
  285. - (ICLR'23) React: Synergizing reasoning and acting in language models [[Paper]](https://arxiv.org/abs/2210.03629)
  286. - (arxiv'21) Webgpt: Browser-assisted question-answering with human feedback [[Paper]](https://arxiv.org/abs/2112.09332)
  287. - (ICML'23) Pal: Program-aided language models [[Paper]](https://arxiv.org/abs/2211.10435)
  288. - (TMLR'24) Voyager: An open-ended embodied agent with large language models [[Paper]](https://arxiv.org/abs/2305.16291)
  289. - (ICLR'24) Large language models as tool makers [[Paper]](https://arxiv.org/abs/2305.17126)
  290. - (EMNLP'23 findings) Creator: Tool creation for disentangling abstract and concrete reasoning of large language models [[Paper]](https://aclanthology.org/anthology-files/anthology-files/pdf/findings/2023.findings-emnlp.462.pdf)
  291. - (ICLR'24) Craft: Customizing llms by creating and retrieving from specialized toolsets [[Paper]](https://arxiv.org/abs/2309.17428)
  292. - (ACL'25) Llm agents making agent tools [[Paper]](https://aclanthology.org/2025.acl-long.1266/)
  293. - (ACL'25) Gödel Agent: A Self-Referential Agent Framework for Recursively Self-Improvement [[Paper]](https://aclanthology.org/2025.acl-long.1354/)
  294. - (arxiv'25) Alita: Generalist agent enabling scalable agentic reasoning with minimal predefinition and maximal self-evolution [[Paper]](https://arxiv.org/abs/2505.20286)
  295. - (arxiv'25) TOUCAN: Synthesizing 1.5M Tool-Agentic Data from Real-World MCP Environments [[Paper]](https://arxiv.org/abs/2510.01179)
  296. - (arxiv'25) Live-SWE-agent: Can Software Engineering Agents Self-Evolve on the Fly? [[Paper]](https://arxiv.org/abs/2511.13646)
  297. - (ICLR'26) EvoTest: Evolutionary Test-Time Learning for Self-Improving Agentic Systems [[Paper]](https://arxiv.org/abs/2510.13220)
  298. - (arxiv'26) Beyond Static Tools: Test-Time Tool Evolution for Scientific Reasoning [[Paper]](https://arxiv.org/abs/2601.07641)
  299. - (arxiv'26) Teaching LLMs to Learn Tool Trialing and Execution through Environment Interaction [[Paper]](https://arxiv.org/abs/2601.12762) [[Code]](https://github.com/NEUIR/ToolMaster)
  300. ### Agentic Topology Evolution
  301. #### Offline Architecture Search
  302. - (ICML'24) Language agents as optimizable graphs [[Paper]](https://arxiv.org/abs/2402.16823)
  303. - (ICLR'25) Scaling large language model-based multi-agent collaboration [[Paper]](https://arxiv.org/abs/2406.07155)
  304. - (arxiv'24) Autoflow: Automated workflow generation for large language model agents [[Paper]](https://arxiv.org/abs/2407.12821)
  305. - (ICLR'25) Aflow: Automating agentic workflow generation [[Paper]](https://arxiv.org/abs/2410.10762)
  306. - (ICLR'25) Automated design of agentic systems [[Paper]](https://arxiv.org/abs/2408.08435)
  307. - (ICML'25) MAS-GPT: Training LLMs to build LLM-based multi-agent systems [[Paper]](https://arxiv.org/abs/2503.03686)
  308. - (arXiv'26) AutoRef: Harness Optimization for Agentic Multi-Reference Image Generation [[Paper]](https://github.com/KuOnoda/AutoRef)
  309. - (arXiv'26) PluginRSI: Recursive Improvement of Agent Harnesses with Reusable Plugins [[Paper]](https://arxiv.org/abs/2609.32423) [[Code]](https://github.com/syr-cn/PluginRSI)
  310. #### Runtime Dynamic Adaptation
  311. - (COLM'25) Malt: Improving reasoning with multi-agent llm training [[Paper]](https://arxiv.org/abs/2412.01928)
  312. - (arxiv'25) Adaptive Graph Pruning for Multi-Agent Communication [[Paper]](https://arxiv.org/abs/2506.02951)
  313. - (IJCAI'24) Autoagents: A framework for automatic agent generation [[Paper]](https://arxiv.org/abs/2309.17288)
  314. - (NAACL'25) Evoagent: Towards automatic multi-agent generation via evolutionary algorithms [[Paper]](https://aclanthology.org/anthology-files/pdf/naacl/2025.naacl-long.315.pdf)
  315. - (ICML'25) G-designer: Architecting multi-agent communication topologies via graph neural networks [[Paper]](https://arxiv.org/abs/2410.11782)
  316. - (ICML'25) Multi-agent architecture search via agentic supernet [[Paper]](https://arxiv.org/abs/2502.04180)
  317. - (NeurIPS'25) Rema: Learning to meta-think for llms with multi-agent reinforcement learning [[Paper]](https://neurips.cc/virtual/2025/loc/san-diego/poster/115462)
  318. - (ICLR'26) Multi-agent design: Optimizing agents with better prompts and topologies [[Paper]](https://arxiv.org/abs/2502.02533)
  319. #### Structural State Evolution
  320. - (NeurIPS'25 workshop) Sedm: Scalable self-evolving distributed memory for agents [[Paper]](https://arxiv.org/abs/2509.09498)
  321. - (arxiv'26) EvoConfig: Self-Evolving Multi-Agent Systems for Efficient Autonomous Environment Configuration [[Paper]](https://arxiv.org/abs/2601.16489)
  322. - (NeurIPS'25) G-Memory: Tracing Hierarchical Memory for Multi-Agent Systems [[Paper]](https://arxiv.org/abs/2506.07398)
  323. - (arxiv'25) Collaborative Memory: Multi-User Memory Sharing in LLM Agents with Dynamic Access Control [[Paper]](https://neurips.cc/virtual/2025/loc/san-diego/poster/116187)
  324. - (arxiv'25) Latent collaboration in multi-agent systems [[Paper]](https://arxiv.org/abs/2511.20639)
  325. - (arxiv'26) Evolve as a Team: Collaborative Self-Evolution for LLM-based Multi-Agent Systems [[Paper]](https://arxiv.org/abs/2605.29790) [[Code]](https://github.com/zz-haooo/Meta-Team)
  326. ## Model-Environment Co-Evolution 🔥
  327. <h3 align="center">
  328. <p align="center"><img width="100%" src="figs/co-evolving.png" /></p>
  329. ### Multi-Agent Policy Co-Evolution
  330. - (STAI-X'26) NeuroMAS: Multi-Agent Systems as Neural Networks with Joint Reinforcement Learning [[Paper]](https://arxiv.org/abs/2605.16757) [[Code]](https://github.com/haoranlustat/NeuroMAS)
  331. - (ICLR'26) Stronger-MAS: Multi-Agent Reinforcement Learning for Collaborative LLMs [[Paper]](https://arxiv.org/abs/2510.11062)
  332. - (AAAI'26) LLM Collaboration with Multi-Agent Reinforcement Learning [[Paper]](https://arxiv.org/abs/2508.04652)
  333. - (arxiv'26) Decentralized LLM Collaboration with Multi-Agent Actor Critic [[Paper]](https://arxiv.org/pdf/2601.21972)
  334. - (arxiv'26) Group-Evolving Agents: Open-Ended Self-Improvement via Experience Sharing [[Paper]](https://arxiv.org/abs/2602.04837)
  335. - (arxiv'26) CORAL: Towards Autonomous Multi-Agent Evolution for Open-Ended Discovery [[Paper]](https://arxiv.org/abs/2604.01658) [[Code]](https://github.com/Human-Agent-Society/CORAL)
  336. - (ACL'25 findings) Optima: Optimizing effectiveness and efficiency for llm-based multi-agent system [[Paper]](https://aclanthology.org/2025.findings-acl.601.pdf)
  337. - (ACL'25) Maporl: Multi-agent post-co-training for collaborative large language models with reinforcement learning [[Paper]](https://aclanthology.org/2025.acl-long.1459/)
  338. - (arxiv'25) Marft: Multi-agent reinforcement fine-tuning [[Paper]](https://arxiv.org/abs/2504.16129)
  339. - (ICLR'26) CoMAS: Co-Evolving Multi-Agent Systems via Interaction Rewards [[Paper]](https://arxiv.org/abs/2510.08529)
  340. ### Environment Training 🔥
  341. #### Adaptive Curriculum Evolution
  342. - (arxiv'25) GenEnv: Difficulty-Aligned Co-Evolution Between LLM Agents and Environment Simulators [[Paper]](https://arxiv.org/abs/2512.19682)
  343. - (ICLR'26) Don't Just Fine-tune the Agent, Tune the Environment [[Paper]](https://arxiv.org/abs/2510.10197)
  344. - (arxiv'25) Rlve: Scaling up reinforcement learning for language models with adaptive verifiable environments [[Paper]](https://arxiv.org/abs/2511.07317)
  345. #### Scalable Environment Evolution
  346. - (ICLR'26) Scaling agent learning via experience synthesis [[Paper]](https://arxiv.org/abs/2511.03773)
  347. - (arxiv'25) AutoEnv: Automated Environments for Measuring Cross-Environment Agent Learning [[Paper]](https://arxiv.org/abs/2511.19304)
  348. - (arxiv'26) Endless Terminals: Scaling RL Environments for Terminal Agents [[Paper]](https://arxiv.org/abs/2601.16443)
  349. - (NeurIPS'25) REASONING GYM: Reasoning Environments for Reinforcement Learning with Verifiable Rewards [[Paper]](https://neurips.cc/virtual/2025/loc/san-diego/poster/121745)
  350. - (ICLR'26) Gem: A gym for agentic llms [[Paper]](https://arxiv.org/abs/2510.01051)
  351. - (arxiv'24) AgentGym: Evolving Large Language Model-based Agents across Diverse Environments [[Paper]](https://arxiv.org/abs/2406.04151)
  352. - (arxiv'26) ScaleEnv: Scaling Environment Synthesis from Scratch for Generalist Interactive Tool-Use Agent Training [[Paper]](https://arxiv.org/abs/2602.06820)
  353. - (arxiv'25) Towards General Agentic Intelligence via Environment Scaling [[Paper]](https://arxiv.org/abs/2509.13311)
  354. - (arxiv'25) Simulating Environments with Reasoning Models for Agent Training [[Paper]](https://arxiv.org/abs/2511.01824)
  355. - (arxiv'25) Feedback-Driven Tool-Use Improvements in Large Language Models via Automated Build Environments [[Paper]](https://arxiv.org/abs/2508.08791)
  356. - (arxiv'25) AutoForge: Automated Environment Synthesis for Agentic Reinforcement Learning [[Paper]](https://arxiv.org/abs/2512.22857)
  357. - (arxiv'26) EnvScaler: Scaling Tool-Interactive Environments for LLM Agent via Programmatic Synthesis [[Paper]](https://arxiv.org/abs/2601.05808)
  358. - (arxiv'26) Agent-World: Scaling Real-World Environment Synthesis for Evolving General Agent Intelligence [[Paper]](https://arxiv.org/abs/2604.18292)
  359. # 🏆 Benchmarks
  360. ## Self-Evolution Evaluation
  361. Benchmarks that evaluate whether agents can turn capability goals, interaction experience, or harness revisions into improvements on held-out tasks.
  362. | Name | Domain | Modality | Feature | Link | Paper |
  363. | --- | --- | --- | --- | --- | --- |
  364. | Aspire | Vague-Goal Self-Evolution | Text | Hidden Downstream Tasks, Model-Weight and Agent-Harness Updates | [🌐 Project](https://self-developing-agents.github.io/) | [[Paper]](https://arxiv.org/abs/2608.31111) |
  365. | S3Gym | Experience-Based Self-Improvement | Text | Seven Games, Self-Testing and Self-Judging, Held-Out Evaluation | [🌐 Project](https://self-developing-agents.github.io/) | [[Paper]](https://arxiv.org/abs/2608.31100) |
  366. | HarnessDev | Agent Harness Creation and Evolution | Text / Code | Runnable Harness Artifacts, Downstream Feedback, Held-Out Evaluation | [🌐 Project](https://self-developing-agents.github.io/) | [[Paper]](https://arxiv.org/abs/2609.01437) |
  367. ## Intrinsic Capabilities
  368. ### General Knowledge
  369. | Name | Domain | Modality | Feature | Link | Paper |
  370. | --- | --- | --- | --- | --- | --- |
  371. | MMLU-Pro | General Knowledge | Text | Robust Reasoning, 10-Choice | [🤗 HF](https://huggingface.co/datasets/TIGER-Lab/MMLU-Pro) | [[Paper]](https://arxiv.org/abs/2406.01574) |
  372. | HotpotQA | General Knowledge | Text | Multi-hop Reasoning, Wiki-based | [🤗 HF](https://huggingface.co/datasets/hotpotqa/hotpot_qa) | [[Paper]](https://arxiv.org/abs/1809.09600) |
  373. | MMLU | General Knowledge | Text | Massive Multitask, 57 Disciplines | [🤗 HF](https://huggingface.co/datasets/cais/mmlu) | [[Paper]](https://arxiv.org/abs/2009.03300) |
  374. | MuSiQue | General Knowledge | Text | Connected Multi-hop, Robustness | [💻 GitHub](https://github.com/StonyBrookNLP/musique) | [[Paper]](https://aclanthology.org/2022.tacl-1.31/) |
  375. | NQ | General Knowledge | Text | Real User Queries, Open-Domain | [💻 GitHub](https://github.com/google-research-datasets/natural-questions) | [[Paper]](https://aclanthology.org/Q19-1026/) |
  376. | TriviaQA | General Knowledge | Text | Reading Comprehension, Triples | [🤗 HF](https://huggingface.co/datasets/mandarjoshi/trivia_qa) | [[Paper]](https://arxiv.org/abs/1705.03551) |
  377. | PopQA | General Knowledge | Text | Long-Tail Knowledge, RAG Focus | [🤗 HF](https://huggingface.co/datasets/akariasai/PopQA) | [[Paper]](https://aclanthology.org/2023.acl-long.546/) |
  378. | 2WikiMultiHopQA | General Knowledge | Text | Structured Multi-hop, Explanations | [🤗 HF](https://huggingface.co/datasets/xanhho/2WikiMultihopQA) | [[Paper]](https://arxiv.org/abs/2011.01060) |
  379. | BBH | General Knowledge | Text | Challenging Tasks, CoT Focus | [💻 GitHub](https://github.com/suzgunmirac/BIG-Bench-Hard) | [[Paper]](https://aclanthology.org/2023.findings-acl.824/) |
  380. | AGIEval | General Knowledge | Text | Human-Centric Exams, General | [💻 GitHub](https://github.com/ruixiangcui/AGIEval) | [[Paper]](https://aclanthology.org/2024.findings-naacl.149/) |
  381. | ARC | General Knowledge | Visual | Abstraction, Few-Shot Reasoning | [💻 GitHub](https://github.com/fchollet/ARC-AGI) | [[Paper]](https://arxiv.org/abs/1911.01547) |
  382. | NarrativeQA | General Knowledge | Text | Long Context, Story Understanding | [💻 GitHub](https://github.com/google-deepmind/narrativeqa) | [[Paper]](https://aclanthology.org/Q18-1023/) |
  383. | LongBench | General Knowledge | Text | Long Context, Multi-Task Eval | [💻 GitHub](https://github.com/THUDM/LongBench) | [[Paper]](https://aclanthology.org/2024.acl-long.170/) |
  384. | HLE | General Knowledge | Multimodal | Frontier Knowledge, Un-googleable | [🤗 HF](https://huggingface.co/datasets/cais/hle) | [[Paper]](https://arxiv.org/abs/2501.14249) |
  385. ### Scientific Reasoning
  386. | Name | Domain | Modality | Feature | Link | Paper |
  387. | --- | --- | --- | --- | --- | --- |
  388. | GPQA | Scientific Reasoning | Text | Google-Proof, PhD-Level Experts | [💻 GitHub](https://github.com/idavidrein/gpqa) | [[Paper]](https://arxiv.org/abs/2311.12022) |
  389. | SuperGPQA | Scientific Reasoning | Text | 285 Disciplines, Light Industry/Agri | [🤗 HF](https://huggingface.co/datasets/m-a-p/SuperGPQA) | [[Paper]](https://arxiv.org/abs/2502.14739) |
  390. | SciBench | Scientific Reasoning | Text | College Science, Calculation | [💻 GitHub](https://github.com/mandyyyyii/scibench) | [[Paper]](https://arxiv.org/abs/2307.10635) |
  391. | ChemBench | Scientific Reasoning | Text | Chemistry, Autonomous Labs | [💻 GitHub](https://github.com/lamalab-org/chembench) | [[Paper]](https://arxiv.org/abs/2404.01475) |
  392. | SciQA | Scientific Reasoning | Text | Scientific QA, Knowledge Graph | [🤗 HF](https://huggingface.co/datasets/orkg/SciQA) | [[Paper]](https://www.nature.com/articles/s41598-023-33607-z) |
  393. ### Mathematical Reasoning
  394. | Name | Domain | Modality | Feature | Link | Paper |
  395. | --- | --- | --- | --- | --- | --- |
  396. | AIME | Mathematical Reasoning | Text | Competition Math, Hard Difficulty | [🤗 HF](https://huggingface.co/datasets/math-ai/aime25) | [[Paper]](https://huggingface.co/datasets/math-ai/aime25) |
  397. | OlympiadBench | Mathematical Reasoning | Multimodal | Visual Reasoning, Olympiad-Level | [💻 GitHub](https://github.com/OpenBMB/OlympiadBench) | [[Paper]](https://aclanthology.org/2024.acl-long.210/) |
  398. | GSM8K | Mathematical Reasoning | Text | Grade School Math, Chain-of-Thought | [🤗 HF](https://huggingface.co/datasets/openai/gsm8k) | [[Paper]](https://arxiv.org/abs/2110.14168) |
  399. | MATH | Mathematical Reasoning | Text | Challenging Math, Diverse Topics | [🤗 HF](https://huggingface.co/datasets/nlile/hendrycks-MATH-benchmark) | [[Paper]](https://arxiv.org/abs/2103.03874) |
  400. | AMC | Mathematical Reasoning | Text | Pre-Olympiad, Competition Math | [🤗 HF](https://huggingface.co/datasets/zwhe99/amc23) | [[Paper]](https://huggingface.co/datasets/zwhe99/amc23) |
  401. ### Code Generation
  402. | Name | Domain | Modality | Feature | Link | Paper |
  403. | --- | --- | --- | --- | --- | --- |
  404. | LiveCodeBench | Code Generation | Text | Contamination-Free, Dynamic | [💻 GitHub](https://github.com/LiveCodeBench/LiveCodeBench) | [[Paper]](https://openreview.net/forum?id=chfJJYC3iL) |
  405. | BigCodeBench | Code Generation | Text | Complex Libraries, Instruction | [💻 GitHub](https://github.com/bigcode-project/bigcodebench) | [[Paper]](https://openreview.net/forum?id=YrycTjllL0) |
  406. | HumanEval | Code Generation | Text | Functional Correctness, Synthesis | [💻 GitHub](https://github.com/openai/human-eval) | [[Paper]](https://arxiv.org/abs/2107.03374) |
  407. | MBPP | Code Generation | Text | Basic Programming, Semantic | [💻 GitHub](https://github.com/google-research/google-research/blob/master/mbpp/README.md) | [[Paper]](https://arxiv.org/abs/2108.07732) |
  408. | EvalPlus | Code Generation | Text | Rigorous Eval, 80x Test Cases | [💻 GitHub](https://github.com/evalplus/evalplus) | [[Paper]](https://proceedings.neurips.cc/paper_files/paper/2023/hash/b10e724de50e2362cfd04bfeb7d5668f-Abstract-Datasets_and_Benchmarks.html) |
  409. | MultiPL-E | Code Generation | Text | 18+ Languages, Parallel Corpus | [💻 GitHub](https://github.com/nuprl/MultiPL-E) | [[Paper]](https://arxiv.org/abs/2208.08227) |
  410. | CRUXEval | Code Generation | Text | Execution Simulation, CoT Focus | [💻 GitHub](https://github.com/facebookresearch/cruxeval) | [[Paper]](https://arxiv.org/abs/2401.03065) |
  411. ## Agentic Reasoning Capabilities
  412. ### Web Navigation
  413. | Name | Domain | Modality | Feature | Link | Paper |
  414. | --- | --- | --- | --- | --- | --- |
  415. | ClawBench | Web Navigation | Multimodal | Live Websites, Everyday Tasks, Auditable Traces | [💻 GitHub](https://github.com/TIGER-AI-Lab/ClawBench) | [[Paper]](https://arxiv.org/abs/2604.08523) · [[Project]](https://claw-bench.com/) |
  416. | WebArena | Web Navigation | Text | Realistic Tasks, Long-Horizon | [💻 GitHub](https://github.com/web-arena-x/webarena) | [[Paper]](https://openreview.net/forum?id=oKn9c6ytLx) |
  417. | WebShop | Web Navigation | Text | E-commerce, Decision Making | [💻 GitHub](https://github.com/princeton-nlp/WebShop) | [[Paper]](https://proceedings.neurips.cc/paper_files/paper/2022/hash/82ad13ec01f9fe44c01cb91814fd7b8c-Abstract-Conference.html) |
  418. | MT-Mind2Web | Web Navigation | Text | Multi-Turn, Generalization | [🤗 HF](https://huggingface.co/datasets/magicgh/MT-Mind2Web) | [[Paper]](https://aclanthology.org/2024.acl-long.475/) |
  419. | Mind2Web | Web Navigation | Text | Generalist Agent, Real Websites | [💻 GitHub](https://github.com/OSU-NLP-Group/Mind2Web) | [[Paper]](https://proceedings.neurips.cc/paper_files/paper/2023/hash/5950bf290a1570ea401bf98882128160-Abstract-Datasets_and_Benchmarks.html) |
  420. | WebVoyager | Web Navigation | Multimodal | End-to-End, Visual Navigation | [💻 GitHub](https://github.com/MinorJerry/WebVoyager) | [[Paper]](https://arxiv.org/abs/2401.13919) |
  421. | VisualWebArena | Web Navigation | Multimodal | Visual/HTML, Interactive | [💻 GitHub](https://github.com/web-arena-x/visualwebarena) | [[Paper]](https://aclanthology.org/2024.acl-long.49/) |
  422. ### Tool Usage
  423. | Name | Domain | Modality | Feature | Link | Paper |
  424. | --- | --- | --- | --- | --- | --- |
  425. | ToolLLM | Tool Usage | Text | Large-Scale APIs, Instruction Tuning | [💻 GitHub](https://github.com/OpenBMB/ToolBench) | [[Paper]](https://openreview.net/forum?id=dHng2O0Jjr) |
  426. ### Unified Frameworks
  427. | Name | Domain | Modality | Feature | Link | Paper |
  428. | --- | --- | --- | --- | --- | --- |
  429. | AgentGym | Unified Frameworks | Multimodal | Interactive Learning, Diversity | [💻 GitHub](https://github.com/WooooDyy/AgentGym) | [[Paper]](https://arxiv.org/abs/2406.04151) |
  430. | AgentBoard | Unified Frameworks | Multimodal | Analytic Dashboard, Unified | [💻 GitHub](https://github.com/hkust-nlp/AgentBoard) | [[Paper]](https://proceedings.neurips.cc/paper_files/paper/2024/hash/51ac6936ab73c83e65fb25ed28ccb2b4-Abstract-Datasets_and_Benchmarks.html) |
  431. | Reasoning Gym | Unified Frameworks | Text | Algorithmic, Dynamic Tasks | [💻 GitHub](https://github.com/open-thought/reasoning-gym) | [[Paper]](https://arxiv.org/abs/2505.24760) |
  432. | ALFWorld | Unified Frameworks | Text | Text-World, Household Tasks | [💻 GitHub](https://github.com/alfworld/alfworld) | [[Paper]](https://arxiv.org/abs/2010.03768) |
  433. | RoboQuest | Unified Frameworks | Multimodal | Embodied Exploration, Hidden Information | [💻 GitHub](https://github.com/declare-lab/RoboQuest) | [[Paper]](https://arxiv.org/abs/2610.10388) |
  434. | AgentBench | Unified Frameworks | Text | Comprehensive, Multi-Environment | [💻 GitHub](https://github.com/THUDM/AgentBench) | [[Paper]](https://arxiv.org/abs/2308.03688) |
  435. | GAIA | Unified Frameworks | Multimodal | General Assistant, Hard Tasks | [🤗 HF](https://huggingface.co/gaia-benchmark) | [[Paper]](https://openreview.net/forum?id=oOte_397Q4) |
  436. | DeepResearch Bench | Unified Frameworks | Text | Long-form Research, Citation Eval | [💻 GitHub](https://github.com/Ayanami0730/deep_research_bench) | [[Paper]](https://arxiv.org/abs/2506.11763) |
  437. ### Software Engineering & OS Operations
  438. | Name | Domain | Modality | Feature | Link | Paper |
  439. | --- | --- | --- | --- | --- | --- |
  440. | SWE-bench | Software Engineering | Text | Real GitHub Issues, Patch Gen | [💻 GitHub](https://github.com/SWE-bench/SWE-bench) | [[Paper]](https://openreview.net/forum?id=VTF8yNQM66) |
  441. | Terminal-Bench | OS Operations | Text | Linux Command Line, Security | [💻 GitHub](https://github.com/laude-institute/terminal-bench) | [[Paper]](https://github.com/laude-institute/terminal-bench) |
  442. | OSWorld | OS Operations | Multimodal | GUI/Desktop, Cross-App | [💻 GitHub](https://github.com/xlang-ai/OSWorld) | [[Paper]](https://proceedings.neurips.cc/paper_files/paper/2024/hash/82df44e3d94a6a3ff7e3e6edd8f4caa7-Abstract-Datasets_and_Benchmarks.html) |
  443. # 💻 Open Source Libraries
  444. ## Foundational Agent Orchestration
  445. | Library | Key Features | Link | Paper |
  446. | --- | --- | --- | --- |
  447. | **LangGraph** | Enables multi-actor applications with cyclic graphs for complex looping logic | [💻 GitHub](https://github.com/langchain-ai/langgraph) | [[Paper]](https://github.com/langchain-ai/langchain) |
  448. | **LlamaIndex** | Integrates private data with LLMs via robust connectors and query engines | [💻 GitHub](https://github.com/run-llama/llama_index) | [[Paper]](https://github.com/jerryjliu/llama_index) |
  449. | **AutoGen** | Automates tasks via customizable agents using conversation and tool integration | [💻 GitHub](https://github.com/microsoft/autogen) | [[Paper]](https://openreview.net/forum?id=uAjxFFing2) |
  450. | **MetaGPT** | Encodes SOPs into LLMs for role-based software development | [💻 GitHub](https://github.com/FoundationAgents/MetaGPT) | [[Paper]](https://openreview.net/forum?id=VtmBAGCN7o) |
  451. | **OpenProgram** | Lets agents create and refine executable workflows while the runtime manages models, tools, memory, context, permissions, and multi-agent coordination | [💻 GitHub](https://github.com/Fzkuji/OpenProgram) | [[Paper]](https://arxiv.org/abs/2606.15874) |
  452. | **Aeon** | Runs autonomous coding agents inside GitHub Actions for continuous, self-directed task execution without approval loops | [💻 GitHub](https://github.com/aaronjmars/aeon) | [[Website]](https://www.aeon.fun/) |
  453. | **AgentDescent** | Optimizes a shared library of skills, prompts, and harness modules by running N workers that propose diffs in parallel, with a barrier-free asynchronous aggregator that resolves conflicts and accepts merges on a Beta posterior over held-out reward | [💻 GitHub](https://github.com/Birfy/agentdescent) | [[Docs]](https://birfy.github.io/agentdescent/) |
  454. | **YYLO** | Command-line orchestrator for coding agents with typed task, validation, merge, and release-readiness boundaries; each task runs in a dedicated branch/worktree and lands via an immutable task source composed in a private detached candidate | [💻 GitHub](https://github.com/yylo-dev/yylo) | [[Website]](https://www.npmjs.com/package/@yylo/cli) |
  455. | **Raven** | Multi-agent harness built for recursive self-improvement that orchestrates built-in and third-party agents as task DAGs; an experimental Curator rewrites each agent's Memory, Planning, Capability, and Action strategy modules and installs only verified changes | [💻 GitHub](https://github.com/EverMind-AI/Raven) | [[Paper]](https://github.com/EverMind-AI/Raven/releases/download/tech-report-v1/technical-report.pdf) |
  456. | **Reef** | Serving and delivery layer for a continually self-improving agent: an OpenAI- and Anthropic-compatible endpoint (`/v1/chat/completions`, `/v1/responses`, `/v1/messages`) records every interaction and returns a record receipt in a response header, a later report attaches a score and/or textual or structured feedback to those receipts, and eligible records are batched into a candidate update to one of two surfaces chosen by the deployment's recipe — model weights (trained with Slime, served with SGLang) or a harness tree of rules, skills, prompts, configuration and extensions; the configured selection policy evaluates the candidate on the deployment's tasks before publication, and accepted updates enter a per-scenario version history that later traffic reads without restarting the server | [💻 GitHub](https://github.com/Human-Agent-Society/reef) | [[Docs]](https://reefinfra.ai/docs/) |
  457. ## Distributed Training
  458. | Library | Key Features | Link | Paper |
  459. | --- | --- | --- | --- |
  460. | **Megatron-LM** | Facilitates high-performance training utilizing multi-dimensional parallelism | [💻 GitHub](https://github.com/NVIDIA/Megatron-LM) | [[Paper]](https://arxiv.org/abs/1909.08053) |
  461. | **DeepSpeed** | Optimizes memory efficiency featuring ZeRO technology | [💻 GitHub](https://github.com/deepspeedai/DeepSpeed) | [[Paper]](https://dl.acm.org/doi/10.1145/3394486.3406703) |
  462. ## Post-training & Alignment
  463. | Library | Key Features | Link | Paper |
  464. | --- | --- | --- | --- |
  465. | **slime** | High-Performance Training and Flexible Data Generation | [💻 GitHub](https://github.com/THUDM/slime)
  466. | **VeRL** | Provides a HybridFlow-based RL library with 3D-HybridEngine | [💻 GitHub](https://github.com/verl-project/verl) | [[Paper]](https://arxiv.org/abs/2409.19256) |
  467. | **OpenRLHF** | Supports distributed RLHF based on Ray and vLLM frameworks | [💻 GitHub](https://github.com/OpenRLHF/OpenRLHF) | [[Paper]](https://arxiv.org/abs/2405.11143) |
  468. | **TRL** | Offers a full-stack library for SFT, Reward Modeling, and RL alignment | [💻 GitHub](https://github.com/huggingface/trl) | [[Paper]](https://github.com/huggingface/trl) |
  469. ## Efficient Fine-tuning
  470. | Library | Key Features | Link | Paper |
  471. | --- | --- | --- | --- |
  472. | **LLaMA Factory** | Provides a unified "code-free" WebUI supporting 100+ models | [💻 GitHub](https://github.com/hiyouga/LlamaFactory) | [[Paper]](https://aclanthology.org/2024.acl-demos.41/) |
  473. | **Unsloth** | Accelerates training via manually derived backpropagation and Triton kernels | [💻 GitHub](https://github.com/unslothai/unsloth) | [[Paper]](http://github.com/unslothai/unsloth) |
  474. ## Inference & Serving
  475. | Library | Key Features | Link | Paper |
  476. | --- | --- | --- | --- |
  477. | **vLLM** | Serves models with high throughput utilizing PagedAttention | [💻 GitHub](https://github.com/vllm-project/vllm) | [[Paper]](https://dl.acm.org/doi/10.1145/3600006.3613165) |
  478. | **SGLang** | Manages structured generation using RadixAttention for aggressive cache reuse | [💻 GitHub](https://github.com/sgl-project/sglang) | [[Paper]](https://proceedings.neurips.cc/paper_files/paper/2024/hash/c7dc7d1dd443c422b8a4063853de4e23-Abstract-Conference.html) |
  479. # 🚀 Applications
  480. ## Automated Scientific Discovery
  481. | Application | Domain | Environment | Evolution Mechanism | Core Technology | Breakthrough Results | Link | Paper |
  482. | --- | --- | --- | --- | --- | --- | --- | --- |
  483. | **Agon** | Omnidisciplinary Research | Prompt-driven research workspace | Generate/Critique loops | Prompt Economy | 30-day unattended run | [💻 GitHub](https://github.com/AutoResearch-Factory/Agon) | [[Paper]](https://arxiv.org/abs/2606.24177) |
  484. | **The AI Scientist** | Academic Research | Simulated review system | Gen-Review cycle | Auto peer-review | Paper auto-generation | [💻 GitHub](https://github.com/SakanaAI/AI-Scientist) | [[Paper]](https://arxiv.org/abs/2408.06292) |
  485. | **AlphaProof** | Logic & Math | Lean verifier | Search-Verify loop | Prover net | IMO 2024 silver-level | [🌐 Web](https://deepmind.google/blog/ai-solves-imo-problems-at-silver-medal-level/) | [[Paper]](https://www.nature.com/articles/s41586-025-09421-4) |
  486. | **ChemCrow** | Chemistry | Lab tools | Plan-Safety-Execute loop | Robotics control | Generalized lab automation | [💻 GitHub](https://github.com/ur-whitelab/chemcrow-public) | [[Paper]](https://arxiv.org/abs/2304.05376) |
  487. | **Coscientist** | Automated Science | Lab env; hardware APIs | Hypothesis–Debate loop | Multi-agent debate | Zero-shot hardware control | [💻 GitHub](https://github.com/gomesgroup/coscientist) | [[Paper]](https://arxiv.org/abs/2502.18864) |
  488. | **GNoME** | Materials Science | DFT simulation space | Active learning loop | GNN predictor | 2.2M stable crystals | [💻 GitHub](https://github.com/google-deepmind/materials_discovery) | [[Paper]](https://www.nature.com/articles/s41586-023-06735-9) |
  489. | **A-Lab** | Materials Science | Robotic lab | Active-learning synthesis | ML-guided planning | 71% synthesis success | [🌐 Web](https://www.nature.com/articles/s41586-023-06734-w) | [[Paper]](https://www.nature.com/articles/s41586-023-06734-w) |
  490. | **CRESt** | Catalysis Discovery | Multimodal robotic lab | Multimodal BO loop | KABO; VLM | 9.3× cost-performance gain | [🌐 Web](https://www.nature.com/articles/s41586-025-09640-5) | [[Paper]](https://www.nature.com/articles/s41586-025-09640-5) |
  491. | **FARS** | Academic Research | Open research workspace | Hypothesis loop | Multi-agent automation | Auto paper generation | [🌐 Web](https://analemma.ai/blog/introducing-fars/) | [[Paper]](https://analemma.ai/blog/introducing-fars/) |
  492. | **AutoSR** | Symbolic Regression | Research-state workspace | Propose–Review MCTS loop | Research State search | Exact laws on 3 cp3-bench tasks no prior SR system solved | [💻 GitHub](https://github.com/AutoResearch-Factory/AutoSR) | [[Paper]](https://arxiv.org/abs/2608.16876) |
  493. | **AgonAlpha** | Quantitative Finance | WorldQuant BRAIN | Propose–Review MCTS loop | Adversarial reviewer with veto | 17 SPECTACULAR alphas from 60 submissions | [💻 GitHub](https://github.com/AutoResearch-Factory/AgonAlpha) | [[Paper]](https://arxiv.org/abs/2608.11250) |
  494. | **SwarmEvolve** | Math & Scientific Computing | Program evaluators | Evolutionary search loop | Latent agent swarms with KV-cache sharing | Improves on 3 AlphaEvolve results with Qwen3 8B | [💻 GitHub](https://github.com/Qile-J/SwarmEvolve) | [[Paper]](https://arxiv.org/abs/2611.08483) |
  495. ## Autonomous Software Engineering
  496. | Application | Domain | Environment | Evolution Mechanism | Core Technology | Breakthrough Results | Link | Paper |
  497. | --- | --- | --- | --- | --- | --- | --- | --- |
  498. | **agent-qa** | Software QA | Browser; mobile apps; CI | Experience-backed run memory | Playwright; Appium; MCP/skills | Self-healing regression detection | [💻 GitHub](https://github.com/vostride/agent-qa) | [[Docs]](https://vostride.com/docs/agent-qa) |
  499. | **SWE-agent** | Software Engineering | Terminal; codebase; CI | Error-feedback correction | ACI interface | High bug-fix success rate | [💻 GitHub](https://github.com/SWE-agent/SWE-agent) | [[Paper]](https://proceedings.neurips.cc/paper_files/paper/2024/hash/6060b4e76adbfbcd170d1ffc19e26fc1-Abstract-Conference.html) |
  500. | **Claude Code** | Long-term Engineering | Project history | Skill accumulation | Skill memory | Senior-level coding | [🌐 Web](https://claude.com/product/claude-code) | [[Paper]](https://claude.com/product/claude-code) |
  501. | **Manus** | Software Engineering | Cloud VM sandbox | Plan-Act-Verify loop | CodeAct | Human-like env interaction | [🌐 Web](https://manus.im/) | [[Paper]](https://manus.im/) |
  502. | **OpenClaw** | Local Agents | Local FS | Community skill | Skill hub | Long-term local adaptation | [🌐 Web](https://openclaw.ai/) | [[Paper]](https://openclaw.ai/) |
  503. | **Ouroboros** | Software Engineering | Local desktop; CLI; project workspaces | Reviewed self-modification | Durable identity and memory | Continuity across tasks and restarts | [💻 GitHub](https://github.com/razzant/ouroboros) | [[Paper]](https://arxiv.org/abs/2608.08311) |
  504. | **Devin** | Software Engineering | Browser; terminal; IDE | Web-based correction | Tool autonomy | Fully autonomous SWE | [🌐 Web](https://devin.ai/) | [[Paper]](https://devin.ai/) |
  505. | **Cursor** | Human-AI Coding | Repo index; shadow env | Human-AI co-evolution | Shadow workspace | Productivity co-adaptation | [🌐 Web](https://cursor.com/) | [[Paper]](https://cursor.com/) |
  506. | **hmharness** | HarmonyOS Development (vertical) | Terminal; HarmonyOS devices (hdc); browser (CDP + real-browser extension); desktop; ARC-AGI-3 (official REST) | Gated skill evolution + world-model belief learning with calibration-targeted exploration | Cognitive OS runtime (world model / five-layer memory / UCB exploration / SPRT evolution gates / learning-target registry) | Cross-environment calibration transfer +0.4166 (pure harness contribution, LLM held fixed); real-data learning cycles: world-model calibration improvement 86.3% relative, split-robust across 5/5 seeds (sign-test p=0.031) | [💻 GitHub](https://github.com/swsgbl/hmharness) | [[Evidence]](https://swsgbl.github.io/hmharness/evidence/) |
  507. ## Open-World Simulation
  508. | Application | Domain | Environment | Evolution Mechanism | Core Technology | Breakthrough Results | Link | Paper |
  509. | --- | --- | --- | --- | --- | --- | --- | --- |
  510. | **Voyager** | Gaming (Minecraft) | Minecraft open world | Explore–Code–Store | Auto curriculum | 15.3× faster progression | [💻 GitHub](https://github.com/MineDojo/Voyager) | [[Paper]](https://arxiv.org/abs/2305.16291) |
  511. | **GITM** | Gaming (Minecraft) | Minecraft open world | Decompose-Plan-Act | Text memory | +47.5% success (Diamond) | [💻 GitHub](https://github.com/OpenGVLab/GITM) | [[Paper]](https://arxiv.org/abs/2305.17144) |
  512. | **Cradle** | General Computer Control | GUI interface | Observe-Plan-Act loop | MLLM; skill curation | API-free computer control | [💻 GitHub](https://github.com/BAAI-Agents/Cradle) | [[Paper]](https://arxiv.org/abs/2403.03186) |
  513. | **Project Sid** | Digital Civilization | Multi-agent society | Social norm co-evolution | PIANO | Emergent economy & laws | [💻 GitHub](https://github.com/altera-al/project-sid) | [[Paper]](https://arxiv.org/abs/2411.00114) |
  514. | **Generative Agents** | Social Simulation | Virtual town sandbox | Observe-Reflect-Plan | Reflection | Emergent group activities | [💻 GitHub](https://github.com/joonspk-research/generative_agents) | [[Paper]](https://dl.acm.org/doi/10.1145/3586183.3606763) |
  515. | **Contemplative Agent** | Agent Social Network | Moltbook (an agent-only social network); local LLM via Ollama | Human-gated harness revision: the agent proposes changes to its constitution, identity and skills, never its weights | Episode log distilled into patterns, then staged proposals | Owner adopted 82 of 629 logged skill proposals since late March 2026 (as of 2026-10-08) | [💻 GitHub](https://github.com/shimo4228/contemplative-agent) | [[DOI]](https://doi.org/10.5281/zenodo.19212118) |
  516. | **SIMA** | Embodied AI | Generative 3D worlds | GenEnv feedback loop | World model | Embodied data reduction | [🌐 Web](https://deepmind.google/blog/sima-2/) | [[Paper]](https://arxiv.org/abs/2512.04797) |
  517. | **Genie** | World Modeling | Text-to-3D worlds | Interactive world loop | Generative world model | Persistent 3D worlds | [🌐 Web](https://deepmind.google/models/genie/) | [[Paper]](https://arxiv.org/abs/2402.15391) |
  518. # 🍀 Citation
  519. If you find this survey helpful, please cite our paper:
  520. ```
  521. @article{xiang2026systematic,
  522. title={A Systematic Survey of Self-Evolving Agents: From Model-Centric to Environment-Driven Co-Evolution},
  523. author={Xiang, Zhishang and Yang, Chengyi and Chen, Zerui and Wei, Zhimin and Tang, Yunbo and Teng, Zongpei and Peng, Zexi and Li, Zongxia and Huang, Chengsong and He, Yicheng and others},
  524. journal={Available at SSRN 6626878},
  525. year={2026}
  526. }
  527. ```

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.