@inproceedings{agent_eval_survey2025,
  title = {{A Survey on Evaluation of LLM-based Agents}},
  author = {Yehudai, Asaf and Eden, Lilach and Li, Alan and Uziel, Guy and Zhao, Yilun and Bar-Haim, Roy and Cohan, Arman and Shmueli-Scheuer, Michal},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  month = {jul},
  address = {San Diego, California, United States},
  publisher = {Association for Computational Linguistics},
  pages = {26690--26714},
  doi = {10.18653/v1/2026.findings-acl.1330},
  url = {https://aclanthology.org/2026.findings-acl.1330/},
  urldate = {2026-10-02},
  keywords = {category-surveys-definitions-and-conceptual-frameworks, section-1, section-4},
}

@misc{agent_selfimprovement_survey2026,
  title = {{Self-Improvements in Modern Agentic Systems: A Survey}},
  author = {Ren, Zhe and Chen, Yimeng and Guo, Dandan and Rong, Guowei and Li, Tonghui and Xiong, R. B. and Lan, Qingfeng and Wang, Wenyi and Nanbo, Li and Yang, Yibo and Zhuge, Mingchen and Schmidhuber, Jürgen},
  year = {2026},
  eprint = {2607.13104},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2607.13104},
  url = {https://arxiv.org/abs/2607.13104v1},
  urldate = {2026-10-02},
  keywords = {category-surveys-definitions-and-conceptual-frameworks, section-1},
}

@misc{bounded_rsi_survey2026,
  title = {{Recursive Self-Improvement in AI: From Bounded Self-Refinement to Autonomous Research Loops}},
  author = {Chen, Mingguang and Wang, Licheng and Qu, Bo},
  year = {2026},
  eprint = {2607.07663},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2607.07663},
  url = {https://arxiv.org/abs/2607.07663v2},
  urldate = {2026-10-02},
  keywords = {category-surveys-definitions-and-conceptual-frameworks, reading-corpus},
}

@misc{last_ai2026,
  title = {{The Last AI Built by Humans: Toward Genuine Recursive Self-Improvement}},
  author = {Duan, Yi and Liu, Ying and Tang, Zirui and Chen, Haodong and Zhou, Jun and Liu, Yumou and Xu, Bangrui and Wu, Yukai and Chen, Sidi and Zhou, Yuhan and Wang, Haoyu and Yu, Xiaoyou and Han, Shaokun and Zhu, Xuzhou and Zhou, Le and Lu, Bolin and Zhou, Wei and Liu, Jiachen and Fang, Nuozhou and Tian, Jiaxin and Chen, Ruoyu and Li, Yuxuan and Zuo, Kai and Zhang, Kaiyan and Yang, Qianyu and Wang, Zijie and Qiu, Jiantao and He, Conghui and Li, Guoliang and Zhou, Bowen and Liu, Zhiyuan and Wen, Zhoufutu and Kang, Jihua and Zhou, Xuanhe and Wu, Fan},
  year = {2026},
  eprint = {2609.11873},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.11873},
  url = {https://arxiv.org/abs/2609.11873v3},
  urldate = {2026-10-02},
  keywords = {category-surveys-definitions-and-conceptual-frameworks, reading-corpus},
}

@misc{code_harness2026,
  title = {{Code as Agent Harness}},
  author = {Ning, Xuying and Tieu, Katherine and Fu, Dongqi and Wei, Tianxin and Li, Zihao and Bei, Yuanchen and Zou, Jiaru and Ai, Mengting and Liu, Zhining and Li, Ting-Wei and Chen, Lingjie and Zhao, Yanjun and Yang, Ke and Li, Bingxuan and Qian, Cheng and Li, Gaotang and Lin, Xiao and Zeng, Zhichen and Qiu, Ruizhong and Chen, Sirui and Sun, Yifan and Yang, Xiyuan and Wang, Ruida and Pan, Rui and Yang, Chenyuan and Zhang, Dylan and Fang, Liri and Cui, Zikun and Cao, Yang and Chen, Pan and Sun, Dorothy and Chen, Ren and Srinivasan, Mahesh and Mathur, Nipun and Xia, Yinglong and Li, Hong and Yan, Hong and Lu, Pan and Zhang, Lingming and Zhang, Tong and Tong, Hanghang and He, Jingrui},
  year = {2026},
  eprint = {2605.18747},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2605.18747},
  url = {https://arxiv.org/abs/2605.18747v1},
  urldate = {2026-10-02},
  keywords = {category-surveys-definitions-and-conceptual-frameworks, section-2},
}

@article{selfevolving_survey2025,
  title = {{A Survey of Self-Evolving Agents: What, When, How, and Where to Evolve on the Path to Artificial Super Intelligence}},
  author = {Gao, Huan-ang and Geng, Jiayi and Hua, Wenyue and Hu, Mengkang and Juan, Xinzhe and Liu, Hongzhang and Liu, Shilong and Qiu, Jiahao and Qi, Xuan and Ren, Qihan and Wu, Yiran and Wang, Hongru and Xiao, Han and Zhou, Yuhang and Zhang, Shaokun and Zhang, Jiayi and Xiang, Jinyu and Fang, Yixiong and Zhao, Qiwen and Liu, Dongrui and Qian, Cheng and Wang, Zhenhailong and Hu, Minda and Wang, Huazheng and Wu, Qingyun and Ji, Heng and Wang, Mengdi},
  year = {2026},
  url = {https://openreview.net/forum?id=CTr3bovS5F},
  urldate = {2026-10-10},
  keywords = {category-surveys-definitions-and-conceptual-frameworks, reading-corpus},
  journal = {Transactions on Machine Learning Research},
  issn = {2835-8856},
  note = {Survey source: arXiv:2507.21046v4},
}

@inproceedings{metaharness2026,
  title = {{Meta-Harness: End-to-End Optimization of Model Harnesses}},
  author = {Lee, Yoonho and Nair, Roshen Sanjay and Zhang, Qizheng and Lee, Kangwook and Khattab, Omar and Finn, Chelsea},
  year = {2026},
  url = {https://openreview.net/forum?id=tmbOUyFx3R},
  urldate = {2026-10-10},
  keywords = {category-harness-optimization-and-search-workflows, section-2, section-3, section-4},
  booktitle = {Third Conference on Language Modeling},
  note = {Survey source: arXiv:2603.28052v1},
}

@misc{ahe2026,
  title = {{Agentic Harness Engineering: Observability-Driven Automatic Evolution of Coding-Agent Harnesses}},
  author = {Lin, Jiahang and Liu, Shichun and Pan, Chengjun and Lin, Lizhi and Dou, Shihan and Xi, Zhiheng and Huang, Xuanjing and Yan, Hang and Han, Zhenhua and Gui, Tao and Jiang, Yu-Gang},
  year = {2026},
  eprint = {2604.25850},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2604.25850},
  url = {https://arxiv.org/abs/2604.25850v4},
  urldate = {2026-10-02},
  keywords = {category-harness-optimization-and-search-workflows, section-2, section-3, section-4},
}

@misc{rrsi2026,
  title = {{RRSI: Regularized Recursive Self-Improvement of Agent Harnesses}},
  author = {Xia, Peng and Han, Rujun and Wang, Zifeng and Chen, Yanfei and Zhuang, Yufan and Lee, Yoonho and Huang, Chengsong and Yu, Han and CuiZhu, Zhongying and Ming, Yifei and Yao, Huaxiu and Gokturk, Burak and Pfister, Tomas and Lee, Chen-Yu},
  year = {2026},
  eprint = {2609.24972},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.24972},
  url = {https://arxiv.org/abs/2609.24972v3},
  urldate = {2026-10-10},
  keywords = {category-harness-optimization-and-search-workflows, section-2, section-3, section-4, section-5},
  note = {Survey source: arXiv:2609.24972v2},
}

@misc{modularrsi2026,
  title = {{ModularRSI: Modular and Generalizable Recursive Harness Self-Improvement}},
  author = {Wu, Siwei and Ren, Jincheng and Li, Yizhi and Li, Haau-Sing and Yang, Chengran and Zhang, Yuxuan and Gu, Weicheng and Yang, Jian and Batista-Navarro, Riza and Zhang, Chuanyi and Liu, Xianglong and Zhou, Ming and Dai, Bryan and Lin, Chenghua},
  year = {2026},
  eprint = {2609.14857},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.14857},
  url = {https://arxiv.org/abs/2609.14857v1},
  urldate = {2026-10-02},
  keywords = {category-harness-optimization-and-search-workflows, section-5},
}

@misc{sia2026,
  title = {{SIA: Self Improving AI with Harness \& Weight Updates}},
  author = {Hebbar, Prannay and Manawat, Yogendra and Verboomen, Samuel and Ivanova, Alesia and Palanimalai, Selvam and Bhatia, Kunal and Baskaran, Vignesh},
  year = {2026},
  eprint = {2605.27276},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2605.27276},
  url = {https://arxiv.org/abs/2605.27276v2},
  urldate = {2026-10-02},
  keywords = {category-model-and-policy-training-harness-co-evolution-and-distillation, reading-corpus},
}

@misc{selfharness2026,
  title = {{Self-Harness: Harnesses That Improve Themselves}},
  author = {Zhang, Hangfan and Zhang, Shao and Li, Kangcong and Zhang, Chen and Chen, Yang and Zhang, Yiqun and Bai, Lei and Hu, Shuyue},
  year = {2026},
  eprint = {2606.09498},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2606.09498},
  url = {https://arxiv.org/abs/2606.09498v3},
  urldate = {2026-10-02},
  keywords = {category-harness-optimization-and-search-workflows, reading-corpus},
}

@inproceedings{dgm2025,
  author = {Zhang, Jenny and Hu, Shengran and Lu, Cong and Lange, Robert and Clune, Jeff},
  booktitle = {International Conference on Learning Representations},
  editor = {C. Vondrick and B. Hariharan and C. Raffel and L. Pinto and D. Yang and A. Faust},
  pages = {104223--104294},
  title = {{Darwin G\"{o}del Machine: Open-Ended Evolution of Self-Improving Agents}},
  url = {https://proceedings.iclr.cc/paper_files/paper/2026/file/aa5f5e6eb6f613ec412f1d948dfa21a5-Paper-Conference.pdf},
  volume = {2026},
  year = {2026},
  keywords = {category-self-modification-and-meta-level-improvement, section-2, section-3, section-4},
  urldate = {2026-10-10},
  note = {Case-analysis source version: \href{https://arxiv.org/abs/2505.22954v3}{arXiv:2505.22954v3}},
}

@misc{hyperagents2026,
  title = {{Hyperagents}},
  author = {Zhang, Jenny and Zhao, Bingchen and Yang, Wannan and Foerster, Jakob and Clune, Jeff and Jiang, Minqi and Devlin, Sam and Shavrina, Tatiana},
  year = {2026},
  eprint = {2603.19461},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2603.19461},
  url = {https://arxiv.org/abs/2603.19461v1},
  urldate = {2026-10-02},
  keywords = {category-self-modification-and-meta-level-improvement, section-3, section-5},
}

@misc{harnessbank2026,
  title = {{HarnessBank: Semantic Gene-Bank Search with Gated Verification for Agent-Harness Self-Evolution}},
  author = {Luo, Xiaotian and Xue, Dizhan and Wang, Fengxingyu and Hu, Chuanrui and Deng, Yafeng},
  year = {2026},
  eprint = {2607.13683},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2607.13683},
  url = {https://arxiv.org/abs/2607.13683v2},
  urldate = {2026-10-02},
  keywords = {category-harness-optimization-and-search-workflows, section-2, section-5},
}

@misc{rhi2026,
  title = {{Recursive Harness Self-Improvement}},
  author = {Lee, Hyunin and Xu, Jinglue and Seely, Jeffrey and Lee, Donghyun and Zaharia, Matei and Tang, Yujin},
  year = {2026},
  eprint = {2607.15524},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2607.15524},
  url = {https://arxiv.org/abs/2607.15524v1},
  urldate = {2026-10-02},
  keywords = {category-harness-optimization-and-search-workflows, reading-corpus},
}

@misc{hsi2026,
  title = {{Hierarchical Self-Improvement: A Framework for Task-Specific Evolvable Agent Harnesses}},
  author = {Zhou, Tailin},
  year = {2026},
  eprint = {2608.08466},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.08466},
  url = {https://arxiv.org/abs/2608.08466v1},
  urldate = {2026-10-02},
  keywords = {category-self-modification-and-meta-level-improvement, section-3},
}

@misc{harnessr1_2026,
  title = {{Harness-R1: Learning to Edit Executable Runtime Harnesses from Agent Failure Trajectories}},
  author = {Shao, Shuai and Zhang, Kangning and Li, Qingyao and Wang, Shijian and Wang, Hao and Jiao, Wenxiang and Lu, Yuan and Guo, Yi and Liu, Weiwen and Zhang, Weinan},
  year = {2026},
  eprint = {2608.02276},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.02276},
  url = {https://arxiv.org/abs/2608.02276v1},
  urldate = {2026-10-02},
  keywords = {category-harness-optimization-and-search-workflows, section-2, section-3},
}

@misc{harnesslens2026,
  title = {{Verify Smarter, Evolve Further: Efficient Harness Evolution through Behavior-Aware Verification}},
  author = {Xu, Jinghan and Zhang, Yikai and Chen, Aili and Li, Weiyuan and Liang, Jiaqing and Yang, Deqing},
  year = {2026},
  eprint = {2608.27311},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.27311},
  url = {https://arxiv.org/abs/2608.27311v1},
  urldate = {2026-10-02},
  keywords = {category-harness-optimization-and-search-workflows, section-2, section-5},
}

@misc{rsiagent2026,
  title = {{RSIAgent: Autonomous Exploration for Recursive Self-improvement in New Environments}},
  author = {Zhu, Sibo and Fan, Shicheng and Wang, Xinyue and Wu, Wenyi and Zhou, Kun and Huang, Biwei},
  year = {2026},
  eprint = {2609.15364},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.15364},
  url = {https://arxiv.org/abs/2609.15364v2},
  urldate = {2026-10-02},
  keywords = {category-memory-context-skills-and-experience-reuse, reading-corpus},
}

@misc{sciencebuddy2026,
  title = {{ScienceBuddy: Recursive-in-Recursive Self-Improvement for Interactive Scientific Agents}},
  author = {Xue, Shuhan and Zhong, Jianyuan and Nan, Ziyuan and Li, Wenbin and Yu, Zhaochen and Ding, Jinchao and Gao, Qiang and Zhan, Pengyu and Zhang, Yuntong and Cheng, Tian and Yin, Zhenfei and Wu, Yingcheng and Yang, Ling},
  year = {2026},
  eprint = {2609.17523},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.17523},
  url = {https://arxiv.org/abs/2609.17523v1},
  urldate = {2026-10-02},
  keywords = {category-model-and-policy-training-harness-co-evolution-and-distillation, reading-corpus},
}

@misc{harnessdev2026,
  title = {{HarnessDev: Can LLMs Create and Evolve Their Own Agent Harness?}},
  author = {Wu, Yuhao and Zhang, Jingyuan and Shi, Jiajun and Lei, Xinping and Gu, Qingshui and Zhang, Yuxuan and Wang, Zexuan and He, Chen and Huang, Chen and Song, Maojia and Zeng, Zhiyuan and Wang, Shaowen and Liu, Jinkai and Shi, Yunfeng and Liu, Jiaheng and Yan, Shen and Huang, Wenhao and Zhang, Ge and Zhang, Wenxuan},
  year = {2026},
  eprint = {2609.01437},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.01437},
  url = {https://arxiv.org/abs/2609.01437v1},
  urldate = {2026-10-02},
  keywords = {category-benchmarks-for-harness-optimization-and-adaptation, section-4},
}

@misc{harness_tampering2026,
  title = {{Auditing Harness Tampering in Self-Improving Agents}},
  author = {Wang, Xing and Zhang, Xiaoyi and Shao, Jie},
  year = {2026},
  eprint = {2609.00069},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.00069},
  url = {https://arxiv.org/abs/2609.00069v1},
  urldate = {2026-10-02},
  keywords = {category-generalization-evaluation-methodology-and-safety, section-5},
}

@misc{generalized_iteration2026,
  title = {{Generalized Agent Iteration: One Formal Framework for Iterative Policy Improvement and Recursive Self-Improvement}},
  author = {Tang, Hongyao and Ma, Yi and Li, Pengyi and Yuan, Yifu},
  year = {2026},
  eprint = {2609.13406},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.13406},
  url = {https://arxiv.org/abs/2609.13406v1},
  urldate = {2026-10-02},
  keywords = {category-surveys-definitions-and-conceptual-frameworks, section-3},
}

@misc{updating_benefit2026,
  title = {{Harness Updating Is Not Harness Benefit: Disentangling Evolution Capabilities in Self-Evolving LLM Agents}},
  author = {Lin, Minhua and Wu, Juncheng and Wang, Zijun and Shi, Zhan and Sang, Yisi and He, Bing and Liu, Zewen and Wei, Tianxin and Wu, Zongyu and Zhang, Zhiwei and Wang, Dakuo and Zhang, Xiang and Dumoulin, Benoit and Xie, Cihang and Zhou, Yuyin and Wang, Suhang and Lu, Hanqing},
  year = {2026},
  eprint = {2605.30621},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2605.30621},
  url = {https://arxiv.org/abs/2605.30621v1},
  urldate = {2026-10-02},
  keywords = {category-generalization-evaluation-methodology-and-safety, section-4},
}

@inproceedings{rethinking_harness_eval2026,
  title = {{Rethinking the Evaluation of Harness Evolution for Agents}},
  author = {Wang, Yike and Zhu, Huaisheng and Hu, Zhengyu and Yuan, Yige and Chen, Zhengyu and Senthil, Shakti and Hajishirzi, Hannaneh and Tsvetkov, Yulia and Dasigi, Pradeep and Xiao, Teng},
  year = {2026},
  url = {https://openreview.net/forum?id=WVAeeSlVim},
  urldate = {2026-10-10},
  keywords = {category-generalization-evaluation-methodology-and-safety, section-4},
  note = {Version 3 (October 1, 2026); Table 1 used for matched-budget comparisons; Survey source: arXiv:2607.12227v3},
  booktitle = {COLM 2026 The 2nd Workshop on Lifelong Agents: Learning, Aligning, and Evolving},
}

@inproceedings{one_recipe2026,
  title = {{One Recipe, Many Harnesses: What Self-Evolution Encodes Across Languages and Models}},
  author = {Yang, Siqi and Yang, Qianlan and Wang, Yu-Xiong and Pujar, Saurabh and Hirzel, Martin},
  year = {2026},
  url = {https://openreview.net/forum?id=jabJQtnOtH},
  urldate = {2026-10-10},
  keywords = {category-generalization-evaluation-methodology-and-safety, section-5},
  booktitle = {NeurIPS 2026 Workshop on Towards Test-Time Continual Learning Agents},
  note = {Survey source: arXiv:2608.10178v1; Accepted workshop paper; forthcoming},
}

@misc{evolutionary_safety2026,
  title = {{Evolutionary Safety of Recursive Self-Improving AI: Taxonomy, Risk Discovery, and Evaluation}},
  author = {Gong, Chang and Bi, Jingping and Yao, Di and Liang, Xinjian and Xiang, Chao and Guo, Ruijie},
  year = {2026},
  eprint = {2609.31186},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.31186},
  url = {https://arxiv.org/abs/2609.31186v1},
  urldate = {2026-10-02},
  keywords = {category-generalization-evaluation-methodology-and-safety, section-5},
}

@misc{robotics_rsi_limits2026,
  title = {{What Stops Recursive Self-Improvement in Robotics? Lessons from 123 Rounds of Agentic Skill Discovery}},
  author = {Wang, Jiaming},
  year = {2026},
  eprint = {2609.31760},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.31760},
  url = {https://arxiv.org/abs/2609.31760v1},
  urldate = {2026-10-02},
  keywords = {category-generalization-evaluation-methodology-and-safety, section-5},
}

@inproceedings{mce2026,
  title = {{Meta Context Engineering via Agentic Skill Evolution}},
  author = {Ye, Haoran and He, Xuning and Arak, Vincent and Dong, Haonan and Song, Guojie},
  year = {2026},
  url = {https://openreview.net/forum?id=P1jHroBS5E},
  urldate = {2026-10-10},
  keywords = {category-self-modification-and-meta-level-improvement, section-3},
  booktitle = {Forty-third International Conference on Machine Learning},
  note = {Survey source: arXiv:2601.21557v2},
}

@misc{metan2026,
  title = {{Meta$^n$: Recursive Self-Improvement through Emergent Depth}},
  author = {Kim, Zae Myung and Lee, Young-Jun and Jwa, Seungyeon and Kang, Dongyeop},
  year = {2026},
  eprint = {2608.24735},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.24735},
  url = {https://arxiv.org/abs/2608.24735v1},
  urldate = {2026-10-02},
  keywords = {category-self-modification-and-meta-level-improvement, reading-corpus},
}

@misc{semaclaw2026,
  title = {{SemaClaw: A Step Towards General-Purpose Personal AI Agents through Harness Engineering}},
  author = {Zhu, Ningyan and Wang, Huacan and Zhou, Jie and Chen, Feiyu and Zhang, Shuo and Chen, Ge and Liu, Chen and Wu, Jiarou and Chen, Wangyi and Mou, Xiaofeng and Xu, Yi},
  year = {2026},
  eprint = {2604.11548},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2604.11548},
  url = {https://arxiv.org/abs/2604.11548v1},
  urldate = {2026-10-02},
  keywords = {category-agent-architectures-and-harness-infrastructure, reading-corpus},
}

@misc{apex_principles2026,
  title = {{APEX: Adaptive Principle EXtraction A Three-Layer Self-Evolution Framework for Production AI Agents}},
  author = {Chen, Ya-Chuan and Lai, Tien-Jen and Hu, Hsiang-Wei},
  year = {2026},
  eprint = {2606.15363},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2606.15363},
  url = {https://arxiv.org/abs/2606.15363v1},
  urldate = {2026-10-02},
  keywords = {category-harness-optimization-and-search-workflows, reading-corpus},
}

@inproceedings{sica2025,
  title = {{A Self-Improving Coding Agent}},
  author = {Robeyns, Maxime and Szummer, Martin and Aitchison, Laurence},
  year = {2025},
  url = {https://openreview.net/forum?id=rShJCyLsOr},
  urldate = {2026-10-10},
  keywords = {category-self-modification-and-meta-level-improvement, section-3, section-4},
  note = {Survey source: arXiv:2504.15228v2},
  booktitle = {ICLR 2025 Workshop on Scaling Self-Improving Foundation Models without Human Supervision},
}

@inproceedings{adas2024,
  author = {Hu, Shengran and Lu, Cong and Clune, Jeff},
  booktitle = {International Conference on Learning Representations},
  editor = {Y. Yue and A. Garg and N. Peng and F. Sha and R. Yu},
  pages = {21344--21377},
  title = {{Automated Design of Agentic Systems}},
  url = {https://proceedings.iclr.cc/paper_files/paper/2025/file/36b7acf6f6010652b3f2a433774a66fe-Paper-Conference.pdf},
  volume = {2025},
  year = {2025},
  keywords = {category-prompt-workflow-and-program-optimization, reading-corpus},
  urldate = {2026-10-10},
  note = {Case-analysis source version: \href{https://arxiv.org/abs/2408.08435v2}{arXiv:2408.08435v2}},
}

@inproceedings{aflow2024,
  author = {Zhang, Jiayi and Xiang, Jinyu and Yu, Zhaoyang and Teng, Fengwei and Chen, XiongHui and Chen, Jiaqi and Zhuge, Mingchen and Cheng, Xin and Hong, Sirui and Wang, Jinlin and Zheng, Bingnan and Liu, Bang and Luo, Yuyu and Wu, Chenglin},
  booktitle = {International Conference on Learning Representations},
  editor = {Y. Yue and A. Garg and N. Peng and F. Sha and R. Yu},
  pages = {34040--34077},
  title = {{AFlow: Automating Agentic Workflow Generation}},
  url = {https://proceedings.iclr.cc/paper_files/paper/2025/file/5492ecbce4439401798dcd2c90be94cd-Paper-Conference.pdf},
  volume = {2025},
  year = {2025},
  keywords = {category-prompt-workflow-and-program-optimization, section-2, section-3, section-4},
  urldate = {2026-10-10},
  note = {Case-analysis source version: \href{https://arxiv.org/abs/2410.10762v4}{arXiv:2410.10762v4}},
}

@inproceedings{dspy2023,
  title = {{{DSP}y: Compiling Declarative Language Model Calls into State-of-the-Art Pipelines}},
  author = {Khattab, Omar and Singhvi, Arnav and Maheshwari, Paridhi and Zhang, Zhiyuan and Santhanam, Keshav and A, Sri Vardhamanan and Haq, Saiful and Sharma, Ashutosh and Joshi, Thomas T. and Moazam, Hanna and Miller, Heather and Zaharia, Matei and Potts, Christopher},
  year = {2024},
  url = {https://openreview.net/forum?id=sY5N0zY5Od},
  urldate = {2026-10-10},
  keywords = {category-prompt-workflow-and-program-optimization, reading-corpus},
  booktitle = {The Twelfth International Conference on Learning Representations},
  note = {Survey source: arXiv:2310.03714v1},
}

@article{textgrad2024,
  title = {{Optimizing generative AI by backpropagating language model feedback}},
  author = {Yuksekgonul, Mert and Bianchi, Federico and Boen, Joseph and Liu, Sheng and Lu, Pan and Huang, Zhi and Guestrin, Carlos and Zou, James},
  year = {2025},
  url = {https://www.nature.com/articles/s41586-025-08661-4},
  urldate = {2026-10-10},
  keywords = {category-prompt-workflow-and-program-optimization, reading-corpus},
  doi = {10.1038/s41586-025-08661-4},
  pages = {609--616},
  journal = {Nature},
  volume = {639},
  number = {8055},
  note = {Survey source: arXiv:2406.07496v1},
}

@inproceedings{gepa2025,
  author = {Agrawal, Lakshya A and Tan, Shangyin and Soylu, Dilara and Ziems, Noah and Khare, Rishi and Opsahl-Ong, Krista and Singhvi, Arnav and Shandilya, Herumb and Ryan, Michael J and Jiang, Meng and Potts, Christopher and Sen, Koushik and Dimakis, Alex and Stoica, Ion and Klein, Dan and Zaharia, Matei and Khattab, Omar},
  booktitle = {International Conference on Learning Representations},
  editor = {C. Vondrick and B. Hariharan and C. Raffel and L. Pinto and D. Yang and A. Faust},
  pages = {8479--8565},
  title = {{GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning}},
  url = {https://proceedings.iclr.cc/paper_files/paper/2026/file/0e9e708b6f48e14fd0ac29e167413f76-Paper-Conference.pdf},
  volume = {2026},
  year = {2026},
  keywords = {category-prompt-workflow-and-program-optimization, reading-corpus},
  urldate = {2026-10-10},
  note = {Case-analysis source version: \href{https://arxiv.org/abs/2507.19457v2}{arXiv:2507.19457v2}},
}

@inproceedings{mass2025,
  title = {{Multi-Agent Design: Optimizing Agents with Better Prompts and Topologies}},
  author = {Zhou, Han and Wan, Xingchen and Sun, Ruoxi and Palangi, Hamid and Iqbal, Shariq and Vulić, Ivan and Korhonen, Anna and Arik, Sercan O},
  year = {2026},
  url = {https://openreview.net/forum?id=I05H9RUzHB},
  urldate = {2026-10-10},
  keywords = {category-prompt-workflow-and-program-optimization, reading-corpus},
  booktitle = {The Fourteenth International Conference on Learning Representations},
  note = {Survey source: arXiv:2502.02533v2},
}

@inproceedings{reflexion2023,
  title = {{Reflexion: language agents with verbal reinforcement learning}},
  author = {Shinn, Noah and Cassano, Federico and Gopinath, Ashwin and Narasimhan, Karthik R and Yao, Shunyu},
  year = {2023},
  url = {https://openreview.net/forum?id=vAElhFcKW6},
  urldate = {2026-10-10},
  keywords = {category-memory-context-skills-and-experience-reuse, reading-corpus},
  booktitle = {Thirty-seventh Conference on Neural Information Processing Systems},
  note = {Survey source: arXiv:2303.11366v4},
}

@inproceedings{selfrefine2023,
  title = {{Self-Refine: Iterative Refinement with Self-Feedback}},
  author = {Madaan, Aman and Tandon, Niket and Gupta, Prakhar and Hallinan, Skyler and Gao, Luyu and Wiegreffe, Sarah and Alon, Uri and Dziri, Nouha and Prabhumoye, Shrimai and Yang, Yiming and Gupta, Shashank and Majumder, Bodhisattwa Prasad and Hermann, Katherine and Welleck, Sean and Yazdanbakhsh, Amir and Clark, Peter},
  year = {2023},
  url = {https://openreview.net/forum?id=S37hOerQLB},
  urldate = {2026-10-10},
  keywords = {category-task-time-search-and-output-refinement-adjacent-work, reading-corpus},
  booktitle = {Thirty-seventh Conference on Neural Information Processing Systems},
  note = {Survey source: arXiv:2303.17651v2},
}

@article{voyager2023,
  title = {{Voyager: An Open-Ended Embodied Agent with Large Language Models}},
  author = {Wang, Guanzhi and Xie, Yuqi and Jiang, Yunfan and Mandlekar, Ajay and Xiao, Chaowei and Zhu, Yuke and Fan, Linxi and Anandkumar, Anima},
  year = {2024},
  url = {https://openreview.net/forum?id=ehfRiF0R3a},
  urldate = {2026-10-10},
  keywords = {category-memory-context-skills-and-experience-reuse, reading-corpus},
  journal = {Transactions on Machine Learning Research},
  issn = {2835-8856},
  note = {Survey source: arXiv:2305.16291v2},
}

@inproceedings{toolmakers2023,
  title = {{Large Language Models as Tool Makers}},
  author = {Cai, Tianle and Wang, Xuezhi and Ma, Tengyu and Chen, Xinyun and Zhou, Denny},
  year = {2024},
  url = {https://openreview.net/forum?id=qV83K9d5WB},
  urldate = {2026-10-10},
  keywords = {category-memory-context-skills-and-experience-reuse, reading-corpus},
  booktitle = {The Twelfth International Conference on Learning Representations},
  note = {Survey source: arXiv:2305.17126v2},
}

@inproceedings{react2022,
  title = {{ReAct: Synergizing Reasoning and Acting in Language Models}},
  author = {Yao, Shunyu and Zhao, Jeffrey and Yu, Dian and Du, Nan and Shafran, Izhak and Narasimhan, Karthik and Cao, Yuan},
  year = {2023},
  booktitle = {International Conference on Learning Representations},
  url = {https://openreview.net/forum?id=WE_vluYUL-X},
  urldate = {2026-10-10},
  keywords = {category-agent-architectures-and-harness-infrastructure, section-3},
  note = {Source version: \href{https://arxiv.org/abs/2210.03629v3}{arXiv:2210.03629v3}},
}

@inproceedings{sweagent2024,
  author = {Yang, John and Jimenez, Carlos and Wettig, Alexander and Lieret, Kilian and Yao, Shunyu and Narasimhan, Karthik and Press, Ofir},
  booktitle = {Advances in Neural Information Processing Systems},
  doi = {10.52202/079017-1601},
  editor = {A. Globerson and L. Mackey and D. Belgrave and A. Fan and U. Paquet and J. Tomczak and C. Zhang},
  pages = {50528--50652},
  publisher = {Curran Associates, Inc.},
  title = {{SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering}},
  url = {https://proceedings.neurips.cc/paper_files/paper/2024/file/5a7c947568c1b1328ccc5230172e1e7c-Paper-Conference.pdf},
  volume = {37},
  year = {2024},
  keywords = {category-agent-architectures-and-harness-infrastructure, section-3},
  urldate = {2026-10-10},
  note = {Case-analysis source version: \href{https://arxiv.org/abs/2405.15793v3}{arXiv:2405.15793v3}},
}

@inproceedings{ape2022,
  title = {{Large Language Models are Human-Level Prompt Engineers}},
  author = {Zhou, Yongchao and Muresanu, Andrei Ioan and Han, Ziwen and Paster, Keiran and Pitis, Silviu and Chan, Harris and Ba, Jimmy},
  year = {2023},
  url = {https://openreview.net/forum?id=92gvk82DE-},
  urldate = {2026-10-10},
  keywords = {category-prompt-workflow-and-program-optimization, reading-corpus},
  booktitle = {The Eleventh International Conference on Learning Representations},
  note = {Survey source: arXiv:2211.01910v2},
}

@inproceedings{opro2023,
  title = {{Large Language Models as Optimizers}},
  author = {Yang, Chengrun and Wang, Xuezhi and Lu, Yifeng and Liu, Hanxiao and Le, Quoc V. and Zhou, Denny and Chen, Xinyun},
  year = {2024},
  url = {https://openreview.net/forum?id=Bb4VGOWELI},
  urldate = {2026-10-10},
  keywords = {category-prompt-workflow-and-program-optimization, reading-corpus},
  booktitle = {The Twelfth International Conference on Learning Representations},
  note = {Survey source: arXiv:2309.03409v3},
}

@misc{aiscientist2024,
  title = {{The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery}},
  author = {Lu, Chris and Lu, Cong and Lange, Robert Tjarko and Foerster, Jakob and Clune, Jeff and Ha, David},
  year = {2024},
  eprint = {2408.06292},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2408.06292},
  url = {https://arxiv.org/abs/2408.06292v3},
  urldate = {2026-10-02},
  keywords = {category-agent-architectures-and-harness-infrastructure, reading-corpus},
}

@misc{aiscientistv2_2025,
  title = {{The AI Scientist-v2: Workshop-Level Automated Scientific Discovery via Agentic Tree Search}},
  author = {Yamada, Yutaro and Lange, Robert Tjarko and Lu, Cong and Hu, Shengran and Lu, Chris and Foerster, Jakob and Clune, Jeff and Ha, David},
  year = {2025},
  eprint = {2504.08066},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2504.08066},
  url = {https://arxiv.org/abs/2504.08066v1},
  urldate = {2026-10-02},
  keywords = {category-agent-architectures-and-harness-infrastructure, reading-corpus},
}

@misc{alphaevolve2025,
  title = {{AlphaEvolve: A coding agent for scientific and algorithmic discovery}},
  author = {Novikov, Alexander and Vũ, Ngân and Eisenberger, Marvin and Dupont, Emilien and Huang, Po-Sen and Wagner, Adam Zsolt and Shirobokov, Sergey and Kozlovskii, Borislav and Ruiz, Francisco J. R. and Mehrabian, Abbas and Kumar, M. Pawan and See, Abigail and Chaudhuri, Swarat and Holland, George and Davies, Alex and Nowozin, Sebastian and Kohli, Pushmeet and Balog, Matej},
  year = {2025},
  eprint = {2506.13131},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2506.13131},
  url = {https://arxiv.org/abs/2506.13131v1},
  urldate = {2026-10-02},
  keywords = {category-prompt-workflow-and-program-optimization, reading-corpus},
}

@inproceedings{swebench2023,
  title = {{{SWE}-bench: Can Language Models Resolve Real-world Github Issues?}},
  author = {Jimenez, Carlos E. and Yang, John and Wettig, Alexander and Yao, Shunyu and Pei, Kexin and Press, Ofir and Narasimhan, Karthik R},
  year = {2024},
  url = {https://openreview.net/forum?id=VTF8yNQM66},
  urldate = {2026-10-10},
  keywords = {category-downstream-task-benchmarks, section-4},
  booktitle = {The Twelfth International Conference on Learning Representations},
  note = {Survey source: arXiv:2310.06770v3},
}

@inproceedings{gaia2023,
  title = {{{GAIA}: a benchmark for General {AI} Assistants}},
  author = {Mialon, Grégoire and Fourrier, Clémentine and Wolf, Thomas and LeCun, Yann and Scialom, Thomas},
  year = {2024},
  url = {https://openreview.net/forum?id=fibxvahvs3},
  urldate = {2026-10-10},
  keywords = {category-downstream-task-benchmarks, reading-corpus},
  booktitle = {The Twelfth International Conference on Learning Representations},
  note = {Survey source: arXiv:2311.12983v1},
}

@inproceedings{mlagentbench2023,
  title = {{{MLA}gentBench: Evaluating Language Agents on Machine Learning Experimentation}},
  author = {Huang, Qian and Vora, Jian and Liang, Percy and Leskovec, Jure},
  year = {2024},
  url = {https://openreview.net/forum?id=1Fs1LvjYQW},
  urldate = {2026-10-10},
  keywords = {category-downstream-task-benchmarks, reading-corpus},
  booktitle = {Forty-first International Conference on Machine Learning},
  note = {Survey source: arXiv:2310.03302v2},
}

@inproceedings{webarena2023,
  title = {{WebArena: A Realistic Web Environment for Building Autonomous Agents}},
  author = {Zhou, Shuyan and Xu, Frank F. and Zhu, Hao and Zhou, Xuhui and Lo, Robert and Sridhar, Abishek and Cheng, Xianyi and Ou, Tianyue and Bisk, Yonatan and Fried, Daniel and Alon, Uri and Neubig, Graham},
  year = {2024},
  url = {https://openreview.net/forum?id=oKn9c6ytLx},
  urldate = {2026-10-10},
  keywords = {category-downstream-task-benchmarks, reading-corpus},
  booktitle = {The Twelfth International Conference on Learning Representations},
  note = {Survey source: arXiv:2307.13854v4},
}

@inproceedings{workarena2024,
  title = {{WorkArena: How Capable are Web Agents at Solving Common Knowledge Work Tasks?}},
  author = {Drouin, Alexandre and Gasse, Maxime and Caccia, Massimo and Laradji, Issam H. and Del Verme, Manuel and Marty, Tom and Vazquez, David and Chapados, Nicolas and Lacoste, Alexandre},
  year = {2024},
  url = {https://openreview.net/forum?id=BRfqYrikdo},
  urldate = {2026-10-10},
  keywords = {category-downstream-task-benchmarks, reading-corpus},
  booktitle = {Forty-first International Conference on Machine Learning},
  note = {Survey source: arXiv:2403.07718v5},
}

@inproceedings{appworld2024,
  title = {{AppWorld: A Controllable World of Apps and People for Benchmarking Interactive Coding Agents}},
  author = {Trivedi, Harsh and Khot, Tushar and Hartmann, Mareike and Manku, Ruskin and Dong, Vinty and Li, Edward and Gupta, Shashank and Sabharwal, Ashish and Balasubramanian, Niranjan},
  year = {2024},
  url = {https://aclanthology.org/2024.acl-long.850/},
  urldate = {2026-10-10},
  keywords = {category-downstream-task-benchmarks, reading-corpus},
  doi = {10.18653/v1/2024.acl-long.850},
  pages = {16022--16076},
  booktitle = {Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  publisher = {Association for Computational Linguistics},
  note = {Survey source: arXiv:2407.18901v1},
}

@inproceedings{taubench2024,
  title = {{$\tau$-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains}},
  author = {Yao, Shunyu and Shinn, Noah and Razavi, Pedram and Narasimhan, Karthik R},
  year = {2025},
  url = {https://openreview.net/forum?id=roNSXZpUDN},
  urldate = {2026-10-10},
  keywords = {category-downstream-task-benchmarks, reading-corpus},
  booktitle = {The Thirteenth International Conference on Learning Representations},
  note = {Survey source: arXiv:2406.12045v1},
}

@inproceedings{tau2bench2025,
  title = {{\${\textbackslash}tau{\textasciicircum}2\$-Bench: Evaluating Conversational Agents in a Dual-Control Environment}},
  author = {Barres, Victor and Dong, Honghua and Ray, Soham and Si, Xujie and Narasimhan, Karthik R},
  year = {2026},
  url = {https://openreview.net/forum?id=OC2z7iSQKa},
  urldate = {2026-10-10},
  keywords = {category-downstream-task-benchmarks, reading-corpus},
  booktitle = {Forty-third International Conference on Machine Learning},
  note = {Survey source: arXiv:2506.07982v1},
}

@inproceedings{mlebench2024,
  title = {{{MLE}-bench: Evaluating Machine Learning Agents on Machine Learning Engineering}},
  author = {Chan, Jun Shern and Chowdhury, Neil and Jaffe, Oliver and Aung, James and Sherburn, Dane and Mays, Evan and Starace, Giulio and Liu, Kevin and Maksin, Leon and Patwardhan, Tejal and Mądry, Aleksander and Weng, Lilian},
  year = {2025},
  url = {https://openreview.net/forum?id=6s5uXNWGIh},
  urldate = {2026-10-10},
  keywords = {category-downstream-task-benchmarks, reading-corpus},
  booktitle = {The Thirteenth International Conference on Learning Representations},
  note = {Survey source: arXiv:2410.07095v6},
}

@inproceedings{agentbench2023,
  title = {{AgentBench: Evaluating {LLM}s as Agents}},
  author = {Liu, Xiao and Yu, Hao and Zhang, Hanchen and Xu, Yifan and Lei, Xuanyu and Lai, Hanyu and Gu, Yu and Ding, Hangliang and Men, Kaiwen and Yang, Kejuan and Zhang, Shudan and Deng, Xiang and Zeng, Aohan and Du, Zhengxiao and Zhang, Chenhui and Shen, Sheng and Zhang, Tianjun and Su, Yu and Sun, Huan and Huang, Minlie and Dong, Yuxiao and Tang, Jie},
  year = {2024},
  url = {https://openreview.net/forum?id=zAdUB0aCTQ},
  urldate = {2026-10-10},
  keywords = {category-downstream-task-benchmarks, reading-corpus},
  booktitle = {The Twelfth International Conference on Learning Representations},
  note = {Survey source: arXiv:2308.03688v3},
}

@inproceedings{autogen2023,
  title = {{AutoGen: Enabling Next-Gen {LLM} Applications via Multi-Agent Conversations}},
  author = {Wu, Qingyun and Bansal, Gagan and Zhang, Jieyu and Wu, Yiran and Li, Beibin and Zhu, Erkang and Jiang, Li and Zhang, Xiaoyun and Zhang, Shaokun and Liu, Jiale and Awadallah, Ahmed Hassan and White, Ryen W and Burger, Doug and Wang, Chi},
  year = {2024},
  url = {https://openreview.net/forum?id=BAakY1hNKS},
  urldate = {2026-10-10},
  keywords = {category-agent-architectures-and-harness-infrastructure, reading-corpus},
  booktitle = {First Conference on Language Modeling},
  note = {Survey source: arXiv:2308.08155v2},
}

@inproceedings{absolutezero2025,
  title = {{Absolute Zero: Reinforced Self-play Reasoning with Zero Data}},
  author = {Zhao, Andrew and Wu, Yiran and Yue, Yang and Wu, Tong and Xu, Quentin and Yue, Yang and Lin, Matthieu and Wang, Shenzhi and Wu, Qingyun and Zheng, Zilong and Huang, Gao},
  year = {2025},
  url = {https://openreview.net/forum?id=neZSGqhxDa},
  urldate = {2026-10-10},
  keywords = {category-model-and-policy-training-harness-co-evolution-and-distillation, reading-corpus},
  booktitle = {The Thirty-ninth Annual Conference on Neural Information Processing Systems},
  note = {Survey source: arXiv:2505.03335v3},
}

@inproceedings{autoharness2026,
  title = {{AutoHarness: improving LLM agents by automatically synthesizing a code harness}},
  author = {Lou, Xinghua and Lázaro-Gredilla, Miguel and Dedieu, Antoine and Wendelken, Carter and Lehrach, Wolfgang and Murphy, Kevin},
  year = {2026},
  url = {https://openreview.net/forum?id=g9rEYVNn5T},
  urldate = {2026-10-10},
  keywords = {category-harness-optimization-and-search-workflows, reading-corpus},
  booktitle = {ICLR 2026 Workshop on Recursive Self-Improvement},
  note = {Survey source: arXiv:2603.03329v1},
}

@misc{safeevolve2026,
  title = {{SafeEvolve: Harness-Policy Co-Evolution from Agent Experience for Safety Alignment}},
  author = {Mao, Qinghua and Qu, Wanying and Guo, Dadi and Yuan, Leitao and Liu, Qingyu and Li, Yu and Chen, Guanxu and Fu, Yanwei and Lin, Xi and Hu, Xia and Liu, Dongrui},
  year = {2026},
  eprint = {2609.02786},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.02786},
  url = {https://arxiv.org/abs/2609.02786v1},
  urldate = {2026-10-02},
  keywords = {category-generalization-evaluation-methodology-and-safety, section-5},
}

@misc{she2026,
  title = {{SHE: Trajectory-driven Safety Harness Evolution for LLM Agents}},
  author = {Qu, Wanying and Mao, Qinghua and Li, Yu and Liu, Jiyao and Zhang, Xin and Guo, Dadi and Zhu, Yanxu and Liu, Qingyu and Yuan, Leitao and Lin, Xi and Zhu, Shanfeng and Fu, Yanwei and Shao, Jing and Hu, Xia and Liu, Dongrui},
  year = {2026},
  eprint = {2608.09885},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.09885},
  url = {https://arxiv.org/abs/2608.09885v1},
  urldate = {2026-10-02},
  keywords = {category-generalization-evaluation-methodology-and-safety, section-5},
}

@misc{compositional_safety2026,
  title = {{Compositional Safety Failures in Harness Evolution: Identification and Runtime Monitoring}},
  author = {Zhang, Zhixiang and Liu, Zesen and Lai, Wai Ip and chen, Hongxu and She, Dongdong},
  year = {2026},
  eprint = {2609.33123},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.33123},
  url = {https://arxiv.org/abs/2609.33123v2},
  urldate = {2026-10-02},
  keywords = {category-generalization-evaluation-methodology-and-safety, section-5},
}

@misc{harness_anatomy2026,
  title = {{Harness Engineering: Anatomy, Architecture, and Evolution of Coding Agents -- A Source-Code Study of Eleven Systems}},
  author = {Barbaste, Paul and Darrigol, Tristan and Vu, Germain and Wiltberger, Tom},
  year = {2026},
  eprint = {2609.00006},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.00006},
  url = {https://arxiv.org/abs/2609.00006v1},
  urldate = {2026-10-02},
  keywords = {category-agent-architectures-and-harness-infrastructure, section-2},
}

@misc{neohorse2026,
  title = {{NeoHorse-1: Towards Recursive Self-Improvement via Agentic Post-Training with Routing Harness}},
  author = {NeoHorse Team and Cao, Guoliang and Dai, Guohao and Guo, Tianyu and Han, Kai and Hu, Hailin and Jiang, Zihan and Kuang, Xiang and Li, Boxun and Li, Yulong and Pei, Zehua and Tian, Yuchuan and Wang, Jiamin and Wang, Yu and Wang, Yunhe and Wu, Yihong and Xu, Haiyang and Zhang, Shuo and Zhou, Hang and Cheng, Siyang and Fan, Jiayu and He, Wei and Jiao, Qingrui and Li, Hongguang and Li, Zhiyuan and Liu, Runke and Liu, Xi and Liu, Xinchen and Pan, Sinno Jialin and Ren, Yi and Song, Liuyang and Wang, Chenyu and Yu, Bei and Zhang, Quanlu and Zhang, Xiangyu and Zheng, Mengyu and Zong, Yingjie},
  year = {2026},
  eprint = {2609.08183},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.08183},
  url = {https://arxiv.org/abs/2609.08183v1},
  urldate = {2026-10-02},
  keywords = {category-model-and-policy-training-harness-co-evolution-and-distillation, reading-corpus},
}

@misc{regenharness2026,
  title = {{RegenHarness: A Robot Agent Harness with Evidence-Gated Recursive Self-Improvement}},
  author = {Wang, Kailin and Jie, Haoxiang and Yan, Yaoyuan and Heng, Zhiyou and Li, Zhaosong},
  year = {2026},
  eprint = {2609.27612},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.27612},
  url = {https://arxiv.org/abs/2609.27612v1},
  urldate = {2026-10-02},
  keywords = {category-agent-architectures-and-harness-infrastructure, reading-corpus},
}

@misc{mixture_branches2026,
  title = {{Mixture of Self-Improving Branches For Agent Harness Optimization}},
  author = {Dong, Haoyu and Zhou, Yuhang and Lin, Zihao and Wu, Yifan and Peng, Bo and Wang, Mingyi and Fan, Xiangjun and Zhang, Lizhu and Zhao, Zhuokai},
  year = {2026},
  eprint = {2609.37834},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.37834},
  url = {https://arxiv.org/abs/2609.37834v1},
  urldate = {2026-10-02},
  keywords = {category-harness-optimization-and-search-workflows, section-2},
}

@misc{helix2026,
  title = {{HELIX: Model-Harness Co-evolution for Recursive Self-Improvement}},
  author = {Fan, Tianyu and Huang, Chao},
  year = {2026},
  eprint = {2608.13951},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.13951},
  url = {https://arxiv.org/abs/2608.13951v1},
  urldate = {2026-10-02},
  keywords = {category-agent-architectures-and-harness-infrastructure, reading-corpus},
}

@misc{arbor2026,
  title = {{Arbor: Tree Search as a Cognition Layer for Autonomous Agents}},
  author = {Prakriya, Neha and Hou, Chaojun and Gong, Zheng and Zhao, Huasha and Zhao, Xi and Li, Mou and Gu, Zhenyu and Barsoum, Emad},
  year = {2026},
  eprint = {2606.12563},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2606.12563},
  url = {https://arxiv.org/abs/2606.12563v1},
  urldate = {2026-10-02},
  keywords = {category-prompt-workflow-and-program-optimization, reading-corpus},
}

@inproceedings{code_to_harness2026,
  title = {{Code-to-Harness: Distilling Black-Box Optimizers from Self-Play}},
  author = {Wu, Yi and Ren, Zheng and Hu, Zhiyu and Wang, Haochen and Chang, Daryl and Wei, Li and Wang, Ting and Li, Zhen and Gupta, Pooja and Jindal, Nitin and Heldt, Lukasz},
  year = {2026},
  url = {https://openreview.net/forum?id=LynZ1vTZ8A},
  urldate = {2026-10-10},
  keywords = {category-prompt-workflow-and-program-optimization, reading-corpus},
  booktitle = {OPT 2026: Optimization for Machine Learning (NeurIPS 2026 Workshop)},
  note = {Survey source: arXiv:2609.09468v2; Accepted workshop paper; forthcoming},
}

@inproceedings{momha2026,
  title = {{MoMHa: Multi-Objective Optimization of LLM Harnesses over Accuracy, Safety, and Tokens}},
  author = {Mukherjee, Subhojyoti and Tanjim, Md Mehrab},
  year = {2026},
  eprint = {2609.30967},
  archiveprefix = {arXiv},
  url = {https://arxiv.org/abs/2609.30967v1},
  urldate = {2026-10-10},
  keywords = {category-harness-optimization-and-search-workflows, section-2},
  booktitle = {Advances in Neural Information Processing Systems},
  note = {Accepted for publication; forthcoming; Survey source: arXiv:2609.30967v1},
}

@misc{whale2026,
  title = {{WHALE: A Simple Recipe for Joint Harness-Weight Optimization}},
  author = {Kim, Haechan and Lee, Yoonho and Lee, Gisang and Finn, Chelsea and Lee, Kangwook},
  year = {2026},
  eprint = {2609.00196},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.00196},
  url = {https://arxiv.org/abs/2609.00196v1},
  urldate = {2026-10-02},
  keywords = {category-model-and-policy-training-harness-co-evolution-and-distillation, reading-corpus},
}

@misc{metabench_harness2026,
  title = {{MetaBench-Harness: Unlocking End-to-End Optimization of Benchmark Harnesses}},
  author = {Chen, Xuanjun and Chen, Hua-Hsuan and Lu, Wei-Chung and Ma, Yinghao and Jang, Jyh-Shing Roger and Lee, Hung-yi},
  year = {2026},
  eprint = {2609.33411},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.33411},
  url = {https://arxiv.org/abs/2609.33411v1},
  urldate = {2026-10-02},
  keywords = {category-generalization-evaluation-methodology-and-safety, section-4},
}

@misc{harnessx2026,
  title = {{HarnessX: A Composable, Adaptive, and Evolvable Agent Harness Foundry}},
  author = {Chen, Tingyang and Lu, Shuo and Zhao, Kang and Meng, Weicheng and Teng, Hanlin and Li, Tianhao and Li, Chao and Liu, Xule and Liang, Jian and Zhang, Zhizhong and Xie, Yuan and Qu, Heng and Shao, Kun and Luan, Jian},
  year = {2026},
  eprint = {2606.14249},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2606.14249},
  url = {https://arxiv.org/abs/2606.14249v3},
  urldate = {2026-10-02},
  keywords = {category-harness-optimization-and-search-workflows, reading-corpus},
}

@inproceedings{ace2025,
  title = {{Agentic Context Engineering: Evolving Contexts for Self-Improving Language Models}},
  author = {Zhang, Qizheng and Hu, Changran and Upasani, Shubhangi and Ma, Boyuan and Hong, Fenglu and Kamanuru, Vamsidhar and Rainton, Jay and Wu, Chen and Ji, Mengmeng and Li, Hanchen and Thakker, Urmish and Zou, James and Olukotun, Kunle},
  year = {2026},
  url = {https://openreview.net/forum?id=eC4ygDs02R},
  urldate = {2026-10-10},
  keywords = {category-memory-context-skills-and-experience-reuse, reading-corpus},
  booktitle = {The Fourteenth International Conference on Learning Representations},
  note = {Survey source: arXiv:2510.04618v3},
}

@misc{dreamrsi2026,
  title = {{Dream-RSI: Recursive Self-Improvement through Evolving Worlds}},
  author = {Zheng, Tong and Wu, Xidong and Zhang, Zheng and He, Zhankui and Zhang, Chaoyi and Coleman, Benjamin and Wei, Ruoqiao and Bai, Di and Liu, Haolin and Liu, Rui and Wang, Xue and Zhuan, Yue and Kang, Wang-Cheng and Xiang, Renkai and Huang, Heng and Cheng, Xinwu and Guo, Yunsong},
  year = {2026},
  eprint = {2609.14858},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.14858},
  url = {https://arxiv.org/abs/2609.14858v2},
  urldate = {2026-10-10},
  keywords = {category-self-modification-and-meta-level-improvement, section-3},
  note = {Survey source: arXiv:2609.14858v1},
}

@inproceedings{reasoningbank2025,
  title = {{ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory}},
  author = {Ouyang, Siru and Yan, Jun and Hsu, I-Hung and Chen, Yanfei and Jiang, Ke and Wang, Zifeng and Han, Rujun and Le, Long and Daruki, Samira and Tang, Xiangru and Tirumalashetty, Vishy and Lee, George and Rofouei, Mahsan and Lin, Hangfei and Han, Jiawei and Lee, Chen-Yu and Pfister, Tomas},
  year = {2026},
  url = {https://openreview.net/forum?id=jL7fwchScm},
  urldate = {2026-10-10},
  keywords = {category-memory-context-skills-and-experience-reuse, reading-corpus},
  booktitle = {The Fourteenth International Conference on Learning Representations},
  note = {Survey source: arXiv:2509.25140v2},
}

@inproceedings{tot2023,
  title = {{Tree of Thoughts: Deliberate Problem Solving with Large Language Models}},
  author = {Yao, Shunyu and Yu, Dian and Zhao, Jeffrey and Shafran, Izhak and Griffiths, Thomas L. and Cao, Yuan and Narasimhan, Karthik R},
  year = {2023},
  url = {https://openreview.net/forum?id=5Xc1ecxO1h},
  urldate = {2026-10-10},
  keywords = {category-task-time-search-and-output-refinement-adjacent-work, reading-corpus},
  booktitle = {Thirty-seventh Conference on Neural Information Processing Systems},
  note = {Survey source: arXiv:2305.10601v2},
}

@inproceedings{lats2023,
  title = {{Language Agent Tree Search Unifies Reasoning, Acting, and Planning in Language Models}},
  author = {Zhou, Andy and Yan, Kai and Shlapentokh-Rothman, Michal and Wang, Haohan and Wang, Yu-Xiong},
  year = {2024},
  url = {https://openreview.net/forum?id=njwv9BsGHF},
  urldate = {2026-10-10},
  keywords = {category-task-time-search-and-output-refinement-adjacent-work, reading-corpus},
  booktitle = {Forty-first International Conference on Machine Learning},
  note = {Survey source: arXiv:2310.04406v3},
}

@misc{tthe2026,
  title = {{TTHE: Test-Time Harness Evolution}},
  author = {Nie, Jun and Zhang, Yonggang and Song, Jun and Cai, Qianshu and Yu, Dahai and Guo, Yike and Tian, Xinmei and Han, Bo},
  year = {2026},
  eprint = {2607.08124},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2607.08124},
  url = {https://arxiv.org/abs/2607.08124v1},
  urldate = {2026-10-02},
  keywords = {category-harness-optimization-and-search-workflows, section-2, section-3},
}

@inproceedings{dynamic_cheatsheet2025,
  title = {{Dynamic Cheatsheet: Test-Time Learning with Adaptive Memory}},
  author = {Suzgun, Mirac and Yuksekgonul, Mert and Bianchi, Federico and Jurafsky, Dan and Zou, James},
  year = {2026},
  url = {https://aclanthology.org/2026.eacl-long.333/},
  urldate = {2026-10-10},
  keywords = {category-memory-context-skills-and-experience-reuse, reading-corpus},
  doi = {10.18653/v1/2026.eacl-long.333},
  pages = {7080--7106},
  booktitle = {Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)},
  publisher = {Association for Computational Linguistics},
  note = {Survey source: arXiv:2504.07952v1},
}

@misc{skillweaver2025,
  title = {{SkillWeaver: Web Agents can Self-Improve by Discovering and Honing Skills}},
  author = {Zheng, Boyuan and Fatemi, Michael Y. and Jin, Xiaolong and Wang, Zora Zhiruo and Gandhi, Apurva and Song, Yueqi and Gu, Yu and Srinivasa, Jayanth and Liu, Gaowen and Neubig, Graham and Su, Yu},
  year = {2025},
  eprint = {2504.07079},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2504.07079},
  url = {https://arxiv.org/abs/2504.07079v1},
  urldate = {2026-10-02},
  keywords = {category-memory-context-skills-and-experience-reuse, reading-corpus},
}

@misc{frontiereng2026,
  title = {{Frontier-Eng: Benchmarking Self-Evolving Agents on Real-World Engineering Tasks with Generative Optimization}},
  author = {Chi, Yizhe and Hong, Deyao and Jiang, Dapeng and Luo, Tianwei and Yang, Kaisen and Zhang, Boshi and Cao, Zhe and Fan, Xiaoyan and He, Bingxiang and Hao, Han and Jin, Weiyang and Lei, Dianqiao and Liu, Qingle and Qian, Houde and Wang, Bowen and Wang, Situ and Zheng, Youjie and Zhou, Yifan and Xiao, Calvin and Cai, Eren and Na, Qinhuai},
  year = {2026},
  eprint = {2604.12290},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2604.12290},
  url = {https://arxiv.org/abs/2604.12290v2},
  urldate = {2026-10-02},
  keywords = {category-downstream-task-benchmarks, section-4},
}

@misc{evobench2026,
  title = {{Evo-Bench: Can Language Models Improve Agent Harness?}},
  author = {Huang, Lisheng and Yang, Chen and Zhou, Hao and Song, Huatong and Chen, Zongchao and Le, Ran and Song, Yang and Zhao, Wayne Xin and Zhang, Tao},
  year = {2026},
  eprint = {2608.09096},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.09096},
  url = {https://arxiv.org/abs/2608.09096v2},
  urldate = {2026-10-02},
  keywords = {category-benchmarks-for-harness-optimization-and-adaptation, section-4},
}

@misc{envharness2026,
  title = {{EnvHarness: Awakening Static Worlds for Agent Learning}},
  author = {Huang, Chengsong and Wang, Zifeng and Han, Rujun and Yan, Jun and Chen, Yanfei and CuiZhu, Zoey and Jiang, Ke and Xia, Peng and Yu, Han and Zhuang, Yufan and Ming, Yifei and Pan, Jiaqi and Mishra, Bhavana Dalvi and Huang, Jiaxin and Gokturk, Burak and Pfister, Tomas and Lee, Chen-Yu},
  year = {2026},
  eprint = {2608.19880},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.19880},
  url = {https://arxiv.org/abs/2608.19880v1},
  urldate = {2026-10-02},
  keywords = {category-agent-architectures-and-harness-infrastructure, reading-corpus},
}

@inproceedings{engdesign2025,
  title = {{Toward Engineering AGI: Benchmarking the Engineering Design Capabilities of LLMs}},
  author = {Guo, Xingang and Li, Yaxin and Kong, XiangYi and JIANG, YILAN and Zhao, Xiayu and Gong, Zhihua and Zhang, Yufan and Li, Daixuan and Sang, Tianle and Zhu, Beixiao and Jun, Gregory and Huang, Yingbing and Liu, Yiqi and Xue, Yuqi and Kundu, Rahul Dev and Lim, Qi and Zhao, Yizhou and Granger, Luke and Younis, Mohamed and Keivan, Darioush and Sabharwal, Nippun and Sinha, Shreyanka and Agarwal, Prakhar and Vandyck, Kojo and Mai, Hanlin and Wang, Zichen and Venkatesh, Aditya and Barik, Ayush and Yang, Jiankun and Yue, Chongying and He, Jingjie and Wang, Libin and Xu, Licheng and Chen, Hao and Wang, Jinwen and Xu, Liujun and Shetty, Rushabh and Guo, Ziheng and Song, Dahui and Jha, Manvi and Liang, Weijie and Yan, Weiman and Zhang, Bryan and Bhandary Karnoor, Sahil and Zhang, Jialiang and Pandya, Rutva and Gong, Xinyi and Ganesh, Mithesh and Shi, Feize and Xu, Ruiling and Zhang, Yifan and Ouyang, Yanfeng and Qin, Lianhui and Rosenbaum, Elyse and Snyder, Corey and Seiler, Peter and Dullerud, Geir and Zhang, Xiaojia and Cheng, Zuofu and Hanumolu, Pavan Kumar and Huang, Jian and Kulkarni, Mayank and Namazifar, Mahdi and Zhang, Huan and Hu, Bin},
  year = {2025},
  url = {https://proceedings.neurips.cc/paper_files/paper/2025/file/664f777548205fb6e0cbb0965e8d2e16-Paper-Datasets_and_Benchmarks_Track.pdf},
  urldate = {2026-10-10},
  keywords = {category-downstream-task-benchmarks, section-4},
  booktitle = {Advances in Neural Information Processing Systems},
  note = {Survey source: arXiv:2509.16204v2},
  doi = {10.52202/085713-2372},
  editor = {D. Belgrave and C. Zhang and H. Lin and R. Pascanu and P. Koniusz and M. Ghassemi and N. Chen},
  publisher = {Curran Associates, Inc.},
  volume = {38, Main Conference},
}

@misc{apexagents2026,
  title = {{APEX-Agents}},
  author = {Vidgen, Bertie and Mann, Austin and Fennelly, Abby and Stanly, John Wright and Rothman, Lucas and Burstein, Marco and Benchek, Julien and Ostrofsky, David and Ravichandran, Anirudh and Sur, Debnil and Venugopal, Neel and Hsia, Alannah and Robinson, Isaac and Huang, Calix and Varones, Olivia and Khan, Daniyal and Haines, Michael and Bridges, Austin and Boyle, Jesse and Twist, Koby and Richards, Zach and Mahapatra, Chirag and Foody, Brendan and Nitski, Osvald},
  year = {2026},
  eprint = {2601.14242},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2601.14242},
  url = {https://arxiv.org/abs/2601.14242v3},
  urldate = {2026-10-02},
  keywords = {category-downstream-task-benchmarks, section-4},
}

@inproceedings{gdpval2025,
  title = {{{GDP}val: Evaluating {AI} Model Performance on Real-World Economically Valuable Tasks}},
  author = {Patwardhan, Tejal and Dias, Rachel and Proehl, Elizabeth and Kim, Grace and Wang, Michele and Watkins, Olivia and Fishman, Simón Posada and Aljubeh, Marwan and Thacker, Phoebe and Fauconnet, Laurance and Kim, Natalie S. and Miserendino, Samuel and Chabot, Gildas and Li, David and Chao, Patrick and Sharman, Michael and Barr, Alexandra and Glaese, Amelia and Tworek, Jerry},
  year = {2026},
  url = {https://openreview.net/forum?id=hcuEdq6eKD},
  urldate = {2026-10-10},
  keywords = {category-downstream-task-benchmarks, section-4},
  booktitle = {The Fourteenth International Conference on Learning Representations},
  note = {Survey source: arXiv:2510.04374v1},
}

@misc{jobbench2026,
  title = {{JobBench: Aligning Agent Work With Human Will}},
  author = {Li, Yuetai and Feng, Yichen and Xu, Zhangchen and Ma, Zixian and Zheng, Kaiyuan and Jiang, Fengqing and Sun, Xinghua and Shao, Rulin and Chen, Zichen and Huang, Yue and Han, Xinyang and Lee, Brian and Xu, Kayla and Zeng, Shenglai and Hua, Hang and Zhang, Xiangliang and Alomair, Basel and Krishna, Ranjay and Zettlemoyer, Luke and Koh, Pang Wei and Ramasubramanian, Bhaskar and Niu, Luyao and Yue, Xiang and Poovendran, Radha},
  year = {2026},
  eprint = {2605.26329},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2605.26329},
  url = {https://arxiv.org/abs/2605.26329v1},
  urldate = {2026-10-02},
  keywords = {category-downstream-task-benchmarks, section-4},
}

@misc{continualharness2026,
  title = {{Continual Harness: Online Adaptation for Self-Improving Foundation Agents}},
  author = {Karten, Seth and Zhang, Joel and Upaa Jr, Tersoo and Feng, Ruirong and Li, Wenzhe and Shi, Chengshuai and Jin, Chi and Vodrahalli, Kiran},
  year = {2026},
  eprint = {2605.09998},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2605.09998},
  url = {https://arxiv.org/abs/2605.09998v1},
  urldate = {2026-10-02},
  keywords = {category-harness-optimization-and-search-workflows, section-3},
}

@misc{evoharnessbench2026,
  title = {{EvoHarnessBench: Can Your Agents Keep Pace with an Evolving Harness?}},
  author = {Ke, Zixuan and Patil, Vaidehi and Shi, Haizhou and Li, Yang and Liu, Ye and Shekkizhar, Sarath and Koul, Anurag and Wang, Jiayu and Nguyen, Xuan Phi and Yavuz, Semih and Bansal, Mohit and Joty, Shafiq},
  year = {2026},
  eprint = {2609.04280},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.04280},
  url = {https://arxiv.org/abs/2609.04280v3},
  urldate = {2026-10-02},
  keywords = {category-benchmarks-for-harness-optimization-and-adaptation, section-4},
}

@misc{evolvemem2026,
  title = {{EvolveMem:Self-Evolving Memory Architecture via AutoResearch for LLM Agents}},
  author = {Liu, Jiaqi and Ye, Xinyu and Xia, Peng and Zheng, Zeyu and Xie, Cihang and Ding, Mingyu and Yao, Huaxiu},
  year = {2026},
  eprint = {2605.13941},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2605.13941},
  url = {https://arxiv.org/abs/2605.13941v1},
  urldate = {2026-10-02},
  keywords = {category-memory-context-skills-and-experience-reuse, reading-corpus},
}

@inproceedings{adaptive_autoharness2026,
  title = {{Adaptive Auto-Harness: Sustained Self-Improvement on Open-Ended Task Streams}},
  author = {Liu, Zewen and Shi, Zhan and Sang, Yisi and He, Bing and Lin, Minhua and Wei, Tianxin and Wang, Dakuo and Dumoulin, Benoit and Jin, Wei and Lu, Hanqing},
  year = {2026},
  url = {https://openreview.net/forum?id=UDdlNs6TxU},
  urldate = {2026-10-10},
  keywords = {category-harness-optimization-and-search-workflows, reading-corpus},
  booktitle = {NeurIPS 2026 Workshop on Continual Learning for Enterprise AI Agents},
  note = {Survey source: arXiv:2606.01770v2; Accepted workshop paper; forthcoming},
}

@inproceedings{rho2026,
  title = {{Evolving Agents in the Dark: Retrospective Harness Optimization via Self-Preference}},
  author = {Pan, Wenbo and Liu, Shujie and Lin, Chin-Yew and Zeng, Jingying and Tang, Xianfeng and Zhou, Xiangyang and Lu, Yan and Jia, Xiaohua},
  year = {2026},
  eprint = {2606.05922},
  archiveprefix = {arXiv},
  url = {https://arxiv.org/abs/2606.05922v3},
  urldate = {2026-10-10},
  keywords = {category-harness-optimization-and-search-workflows, section-2},
  booktitle = {Findings of the Association for Computational Linguistics: EMNLP 2026},
  note = {Accepted for publication; forthcoming; Survey source: arXiv:2606.05922v3},
}

@misc{agentkb2025,
  title = {{Agent KB: Leveraging Cross-Domain Experience for Agentic Problem Solving}},
  author = {Tang, Xiangru and Qin, Tianrui and Peng, Tianhao and Zhou, Ziyang and Shao, Daniel and Du, Tingting and Wei, Xinming and Xia, Peng and Wu, Fang and Zhu, He and Zhang, Ge and Liu, Jiaheng and Wang, Xingyao and Hong, Sirui and Wu, Chenglin and Cheng, Hao and Wang, Chi and Zhou, Wangchunshu},
  year = {2025},
  eprint = {2507.06229},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2507.06229},
  url = {https://arxiv.org/abs/2507.06229v5},
  urldate = {2026-10-10},
  keywords = {category-memory-context-skills-and-experience-reuse, reading-corpus},
  note = {Earlier workshop version: ACL 2026 KnowFM (OpenReview: vyGxyPM2a0); this entry retains the expanded arXiv v5 author list and survey source},
}

@misc{harness_handbook2026,
  title = {{Harness Handbook: Making Evolving Agent Harnesses Readable,Navigable, and Editable}},
  author = {Wang, Ruhan and Shi, Yucheng and Li, Zongxia and Li, Zhongzhi and Yu, Yue and Yang, Junyao and Panaganti, Kishan and Mi, Haitao and Zhou, Dongruo and Leoweiliang},
  year = {2026},
  eprint = {2607.13285},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2607.13285},
  url = {https://arxiv.org/abs/2607.13285v1},
  urldate = {2026-10-02},
  keywords = {category-agent-architectures-and-harness-infrastructure, section-2},
}

@inproceedings{hgm2025,
  title = {{Huxley-G\"odel Machine: Human-Level Coding Agent Development by an Approximation of the Optimal Self-Improving Machine}},
  author = {Wang, Wenyi and Pi{\k{e}}kos, Piotr and Nanbo, Li and Laakom, Firas and Chen, Yimeng and Ostaszewski, Mateusz and Zhuge, Mingchen and Schmidhuber, Jürgen},
  year = {2026},
  url = {https://openreview.net/forum?id=T0EiEuhOOL},
  urldate = {2026-10-10},
  keywords = {category-self-modification-and-meta-level-improvement, section-2, section-3, section-4},
  booktitle = {The Fourteenth International Conference on Learning Representations},
  note = {Survey source: arXiv:2510.21614v3},
}

@inproceedings{terminalbench2026,
  title = {{Terminal-Bench: Benchmarking Agents on Hard, Realistic Tasks in Command Line Interfaces}},
  author = {Merrill, Mike A. and Shaw, Alexander Glenn and Carlini, Nicholas and Li, Boxuan and Raj, Harsh and Bercovich, Ivan and Shi, Lin and Shin, Jeong Yeon and Walshe, Thomas and Buchanan, E. Kelly and Shen, Junhong and Ye, Guanghao and Lin, Haowei and Poulos, Jason and Wang, Maoyu and Nezhurina, Marianna and Lu, Di and Mastromichalakis, Orfeas Menis and Xu, Zhiwei and Chen, Zizhao and Liu, Yue and Zhang, Robert and Chen, Leon Liangyu and Kashyap, Anurag and Uslu, Jan-Lucas and Li, Jeffrey and Wu, Jianbo and Yan, Minghao and Bian, Song and Sharma, Vedang and Sun, Ke and Dillmann, Steven and Anand, Akshay and Lanpouthakoun, Andrew and Koopah, Bardia and Hu, Changran and Guha, Etash Kumar and Dreiman, Gabriel H. S. and Zhu, Jiacheng and Krauth, Karl and Zhong, Li and Muennighoff, Niklas and Amanfu, Robert Kwesi and Tan, Shangyin and Pimpalgaonkar, Shreyas and Aggarwal, Tushar and Lin, Xiangning and Lan, Xin and Zhao, Xuandong and Liang, Yiqing and Wang, Yuanli and Wang, Zilong and Zhou, Changzhi and Heineman, David and Liu, Hange and Trivedi, Harsh and Yang, John and Lin, Junhong and Shetty, Manish and Yang, Michael and Omi, Nabil and Raoof, Negin and Li, Shanda and Zhuo, Terry Yue and Lin, Wuwei and Dai, Yiwei and Wang, Yuxin and Chai, Wenhao and Zhou, Shang and Wahdany, Dariush and She, Ziyu and Hu, Jiaming and Dong, Zhikang and Zhu, Yuxuan and Cui, Sasha and Saiyed, Ahson and Kolbeinsson, Arinbjörn and Rytting, Christopher Michael and Marten, Ryan and Wang, Yixin and Jitsev, Jenia and Dimakis, Alex and Konwinski, Andy and Schmidt, Ludwig},
  year = {2026},
  url = {https://openreview.net/forum?id=a7Qa4CcHak},
  urldate = {2026-10-10},
  keywords = {category-downstream-task-benchmarks, section-4},
  booktitle = {The Fourteenth International Conference on Learning Representations},
  note = {Survey source: arXiv:2601.11868v1},
}

@inproceedings{awm2024,
  title = {{Agent Workflow Memory}},
  author = {Wang, Zora Zhiruo and Mao, Jiayuan and Fried, Daniel and Neubig, Graham},
  year = {2025},
  url = {https://proceedings.mlr.press/v267/wang25bx.html},
  urldate = {2026-10-10},
  keywords = {category-memory-context-skills-and-experience-reuse, reading-corpus},
  booktitle = {Forty-second International Conference on Machine Learning},
  note = {Survey source: arXiv:2409.07429v1},
  pages = {63897--63911},
  volume = {267},
  publisher = {PMLR},
}

@misc{metaskillevolve2026,
  title = {{MetaSkill-Evolve: Recursive Self-Improvement of LLM Agents via Two-Timescale Meta-Skill Evolution}},
  author = {Wang, Zefeng and Yan, Minxi and Bi, Jinhe and Yan, Sikuan and Tresp, Volker and Ma, Yunpu},
  year = {2026},
  eprint = {2607.05297},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2607.05297},
  url = {https://arxiv.org/abs/2607.05297v1},
  urldate = {2026-10-02},
  keywords = {category-self-modification-and-meta-level-improvement, section-3},
}

@misc{financial_evolution_audit2026,
  title = {{Auditing Self-Evolution in Financial Agents: Capability Gains, Security Drift, and Execution-Interface Mismatch}},
  author = {Li, Jialong and Zhu, Jialing},
  year = {2026},
  eprint = {2608.17684},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.17684},
  url = {https://arxiv.org/abs/2608.17684v1},
  urldate = {2026-10-02},
  keywords = {category-generalization-evaluation-methodology-and-safety, reading-corpus},
}

@misc{path_rsi2026,
  title = {{The Path to Recursive Self-Improving Agents: Foundation, Framework, and Future Directions}},
  author = {Liu, Shuaiqi and Lin, Zhengkai and Zhang, Yuxiang and Ren, Yuanyi and Wu, Yue and Li, Yongbin and Wang, Zheng and Fu, Zhihang and Ye, Jieping},
  year = {2026},
  url = {https://www.preprints.org/manuscript/202608.0051},
  urldate = {2026-10-02},
  keywords = {category-surveys-definitions-and-conceptual-frameworks, section-1, section-3},
  note = {Review preprint},
}

@misc{towards_rsi_survey,
  title = {{Towards AI That Improves Itself: A Survey of Recursive Self-Improvement}},
  author = {Li, Hanjing and Chen, Qiguang and Zhang, Chenyuan and Qiu, Qionglin and Meng, Fanqing and Hu, Mengkang and Qin, Libo and Zhang, Min},
  year = {n.d.},
  url = {https://lee1003-lee.github.io/Awesome-RSI-Research/},
  urldate = {2026-10-02},
  keywords = {category-surveys-definitions-and-conceptual-frameworks, reading-corpus},
  note = {Author-hosted manuscript; 2026 snapshot},
}

@misc{harness_engineering_survey2026,
  title = {{Agent Harness Engineering: A Survey}},
  author = {Li, Junjie and Xiao, Xi and Zhang, Yunbei and Liu, Chen and Zhao, Lin and Liao, Xiaoying and Ji, Yingrui and Wang, Janet and Ge, Yingqiang and Xu, Weijie and Fang, Xi and Xu, Xiang and Zhao, Tianchen and Kim, Youngeun and Hamm, Jihun and Wang, Tianyang and Reddy, Chandan K.},
  year = {2026},
  url = {https://picrew.github.io/LLM-Harness/},
  urldate = {2026-10-02},
  keywords = {category-surveys-definitions-and-conceptual-frameworks, section-1},
  note = {Author-hosted survey manuscript; publication venue not established},
}

@misc{harness_agents_survey2026,
  title = {{Agent Harness for Large Language Model Agents: A Survey}},
  author = {Meng, Qianyu and Wang, Yanan and Chen, Liyi and Li, Yihang and Wu, Wei and Jiang, Wenyuan and Wang, Qimeng and Lu, Chengqiang and Gao, Yan and Wu, Yi and Hu, Yao},
  year = {2026},
  url = {https://www.preprints.org/manuscript/202604.0428/v3},
  urldate = {2026-10-10},
  keywords = {category-surveys-definitions-and-conceptual-frameworks, section-1},
  note = {Preprint, version 3},
}

@misc{harvey_lab2026,
  title = {{Open-Sourcing Harvey's Long Horizon Legal Agent Benchmark}},
  author = {Grupen, Niko and Pereyra, Gabe and Pereyra, Julio},
  year = {2026},
  url = {https://www.harvey.ai/blog/introducing-harveys-legal-agent-benchmark},
  urldate = {2026-10-02},
  keywords = {category-official-benchmark-and-implementation-resources, section-4},
  note = {Benchmark announcement},
}

@misc{harnesscompass2026,
  keywords = {category-harness-optimization-and-search-workflows, section-2},
  title = {{HarnessCompass: Guiding Automatic Harness Evolution toward Generalizable and Effective Agent Harnesses}},
  author = {Zhang, Luan and Zhou, Ruochen and Song, Dandan and Chen, Zhengyu and Tian, Yuhang and Yang, Jun and Ma, Huipeng and Li, Chenhao and Feng, Guangyuan and Li, Xudong and Jin, Yizhou and Xu, Yan},
  year = {2026},
  eprint = {2608.01918},
  archiveprefix = {arXiv},
  url = {https://arxiv.org/abs/2608.01918v1},
  urldate = {2026-10-02},
}

@misc{harnessoptbench2026,
  keywords = {category-benchmarks-for-harness-optimization-and-adaptation, section-2, section-4},
  title = {{HarnessOpt-Bench: Evaluating LLMs at Harness Optimization}},
  author = {Ursekar, Varun and Shanker, Apaar and Maurya, Yash and Yasser, Shehab and Kalmath, Vijay S. and Chatrath, Veronica and Xue, Yuan},
  year = {2026},
  eprint = {2608.06301},
  archiveprefix = {arXiv},
  url = {https://arxiv.org/abs/2608.06301v1},
  urldate = {2026-10-02},
}

@misc{ahe_prompt2026,
  keywords = {category-official-benchmark-and-implementation-resources, section-2},
  title = {{AHE: Released Evolution-Agent Prompt}},
  author = {{AHE contributors}},
  year = {2026},
  url = {https://github.com/china-qijizhifeng/agentic-harness-engineering/blob/8b2a55d97590363fe50c3cc6b5e833b020a4bb4c/agents/evolve_agent/evolve_prompt.md},
  urldate = {2026-10-02},
  note = {Versioned implementation; commit 8b2a55d},
}

@misc{ahe_code2026,
  keywords = {category-official-benchmark-and-implementation-resources, section-2},
  title = {{AHE: Evolution Query and Cumulative History Implementation}},
  author = {{AHE contributors}},
  year = {2026},
  url = {https://github.com/china-qijizhifeng/agentic-harness-engineering/blob/8b2a55d97590363fe50c3cc6b5e833b020a4bb4c/evolve.py},
  urldate = {2026-10-02},
  note = {Versioned implementation; commit 8b2a55d},
}

@misc{rsiexam2026,
  keywords = {category-official-benchmark-and-implementation-resources, section-4},
  title = {{RSI-Exam: Official Benchmark and Task Collection}},
  author = {{RSI-Exam contributors}},
  year = {n.d.},
  url = {https://rsi-exam.ai/},
  urldate = {2026-10-02},
  note = {Benchmark resource; 2026 resource snapshot, publication date not stated},
}

@misc{rsiexam_memory2026,
  keywords = {category-official-benchmark-and-implementation-resources, section-4},
  title = {{RSI-Exam: LLM Agent Memory Architecture Design}},
  author = {{RSI-Exam contributors}},
  year = {n.d.},
  url = {https://rsi-exam.ai/tasks/locomo_longterm_memory.html},
  urldate = {2026-10-02},
  note = {Task specification and public optimization traces; 2026 resource snapshot, publication date not stated},
}

@misc{rsiexam_lean2026,
  keywords = {category-official-benchmark-and-implementation-resources, section-4},
  title = {{RSI-Exam: Lean 4 Proof-Search LLM Agent Workflow Design}},
  author = {{RSI-Exam contributors}},
  year = {n.d.},
  url = {https://rsi-exam.ai/tasks/lean_formal_proof_workflow_design.html},
  urldate = {2026-10-02},
  note = {Task specification and public optimization traces; 2026 resource snapshot, publication date not stated},
}

@misc{rsiexam_discovery2026,
  keywords = {category-official-benchmark-and-implementation-resources, section-4},
  title = {{RSI-Exam: Scientific Discovery Agent Harness Design}},
  author = {{RSI-Exam contributors}},
  year = {n.d.},
  url = {https://rsi-exam.ai/tasks/discoveryworld_agent_harness_low2.html},
  urldate = {2026-10-02},
  note = {Task specification; 2026 resource snapshot, publication date not stated},
}

@misc{scale_rsibench2026,
  keywords = {category-official-benchmark-and-implementation-resources, section-4},
  title = {{RSI Bench: Official Benchmark}},
  author = {{Scale AI}},
  year = {n.d.},
  url = {https://labs.scale.com/rsi-benchmark/},
  urldate = {2026-10-02},
  note = {Benchmark resource; 2026 resource snapshot, publication date not stated},
}

@misc{scale_swarm2026,
  keywords = {category-official-benchmark-and-implementation-resources, section-4},
  title = {{RSI Bench: Agent Swarm Optimization}},
  author = {{Scale AI}},
  year = {n.d.},
  url = {https://labs.scale.com/rsi-benchmark/tasks/agent-swarm-optimization},
  urldate = {2026-10-02},
  note = {Task specification; 2026 resource snapshot, publication date not stated},
}

@misc{scale_data2026,
  keywords = {category-official-benchmark-and-implementation-resources, section-4},
  title = {{RSI Bench: Nano-GPT Data Curation}},
  author = {{Scale AI}},
  year = {n.d.},
  url = {https://labs.scale.com/rsi-benchmark/tasks/nano-gpt-data-curation},
  urldate = {2026-10-02},
  note = {Task specification; 2026 resource snapshot, publication date not stated},
}

@inproceedings{stop2024,
  title = {{Self-Taught Optimizer ({STOP}): Recursively Self-Improving Code Generation}},
  author = {Zelikman, Eric and Lorch, Eliana and Mackey, Lester and Kalai, Adam Tauman},
  year = {2024},
  url = {https://openreview.net/forum?id=46Zgqo4QIU},
  urldate = {2026-10-10},
  keywords = {category-self-modification-and-meta-level-improvement, section-3},
  booktitle = {First Conference on Language Modeling},
  note = {Survey source: arXiv:2310.02304v3},
}

@inproceedings{godelagent2025,
  title = {{G{\"o}del Agent: A Self-Referential Agent Framework for Recursively Self-Improvement}},
  author = {Yin, Xunjian and Wang, Xinyi and Pan, Liangming and Lin, Li and Wan, Xiaojun and Wang, William Yang},
  booktitle = {Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2025},
  month = {jul},
  address = {Vienna, Austria},
  publisher = {Association for Computational Linguistics},
  pages = {27890--27913},
  doi = {10.18653/v1/2025.acl-long.1354},
  url = {https://aclanthology.org/2025.acl-long.1354/},
  urldate = {2026-10-02},
  keywords = {category-self-modification-and-meta-level-improvement, section-3},
}

@misc{aide2025,
  title = {{AIDE: AI-Driven Exploration in the Space of Code}},
  author = {Jiang, Zhengyao and Schmidt, Dominik and Srikanth, Dhruv and Xu, Dixing and Kaplan, Ian and Jacenko, Deniss and Wu, Yuxiang},
  year = {2025},
  eprint = {2502.13138},
  archiveprefix = {arXiv},
  url = {https://arxiv.org/abs/2502.13138v1},
  urldate = {2026-10-02},
  keywords = {category-prompt-workflow-and-program-optimization, section-2},
}

@misc{aide2_2026,
  title = {{Recursive self-improvement of AI research agents}},
  author = {Srikanth, Dhruv and Zhao, Bingchen and Xu, Dixing and Wu, Yuxiang and Jiang, Zhengyao},
  year = {2026},
  eprint = {2609.26457},
  archiveprefix = {arXiv},
  url = {https://arxiv.org/abs/2609.26457v1},
  urldate = {2026-10-02},
  keywords = {category-self-modification-and-meta-level-improvement, section-2},
}

@misc{memoharness2026,
  title = {{MemoHarness: Agent Harnesses That Learn from Experience}},
  author = {Huang, Yue and Wang, Wenjie and Bao, Han and Ma, Yuchen and Luo, Xiaonan and Nian, Yi and Zhuang, Haomin and Liu, Zheyuan and Zhao, Yue and Zhang, Xiangliang},
  year = {2026},
  eprint = {2607.14159},
  archiveprefix = {arXiv},
  url = {https://arxiv.org/abs/2607.14159v1},
  urldate = {2026-10-02},
  keywords = {category-harness-optimization-and-search-workflows, section-2},
}

@misc{filesystemmemory2026,
  title = {{Filesystem-Based Memory for LLM Agents: Organization, Evolution, and Sustainability}},
  author = {Zhou, Sizhe and Yu, Sheldon and Wei, Hui and Wu, Junda and Ouyang, Siru and Jiao, Yizhu and Pan, Shijia and McAuley, Julian and Zhang, Yu and Yu, Tong and Han, Jiawei},
  year = {2026},
  eprint = {2607.26637},
  archiveprefix = {arXiv},
  url = {https://arxiv.org/abs/2607.26637v1},
  urldate = {2026-10-02},
  keywords = {category-memory-context-skills-and-experience-reuse, section-2},
}

@article{agenttreesearch2024,
  title = {{Tree Search for Language Model Agents}},
  author = {Koh, Jing Yu and McAleer, Stephen Marcus and Fried, Daniel and Salakhutdinov, Ruslan},
  year = {2025},
  url = {https://openreview.net/forum?id=QF0N3x2XVm},
  urldate = {2026-10-10},
  keywords = {category-task-time-search-and-output-refinement-adjacent-work, section-2},
  journal = {Transactions on Machine Learning Research},
  issn = {2835-8856},
  note = {Survey source: arXiv:2407.01476v1},
}

@misc{harnesszero2026,
  title = {{Harness-Zero: Harness Distillation via Agent-as-Harness}},
  author = {Ye, Haoran and Lu, Yuxing and Dong, Haonan and Su, Zhaochen and Song, Guojie},
  year = {2026},
  eprint = {2609.24974},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.24974},
  url = {https://arxiv.org/abs/2609.24974v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-model-and-policy-training-harness-co-evolution-and-distillation, model--harness co-evolution and self-training},
}

@misc{multitask_selfharness2026,
  title = {{Self-Evolving Harness on Multiple Tasks with the Agent as Its Own Optimizer}},
  author = {Xu, Qiankai},
  year = {2026},
  eprint = {2609.38372},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.38372},
  url = {https://arxiv.org/abs/2609.38372v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-harness-optimization-and-search-workflows, harness optimization and search workflows},
}

@misc{dynaharness2026,
  title = {{DynaHarness: A Dynamic Physical Harness for Self-Evolving Robot Agents}},
  author = {Deng, Haoyuan and Liu, Jiebin and Zhang, Tengxiao and Yan, Langning and Cao, Hongye and Wang, Ziwei},
  year = {2026},
  eprint = {2609.40306},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.40306},
  url = {https://arxiv.org/abs/2609.40306v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-agent-architectures-and-harness-infrastructure, agent architectures and harness infrastructure},
}

@misc{harnessevolve2026,
  title = {{HarnessEvolve: Learning from Reference Trajectories for Reliable Agent Self-Evolution}},
  author = {Jiang, Wen and Chu, Mingmin and Tian, Yimeng and Zhang, Qianxin and Yang, Haofei and Yang, Rui and Liu, Yang and Lv, Tao and Li, Fangming},
  year = {2026},
  eprint = {2609.00829},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.00829},
  url = {https://arxiv.org/abs/2609.00829v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-harness-optimization-and-search-workflows, harness optimization and search workflows},
}

@misc{reliable_harness_theory2026,
  title = {{A Theory of Reliable Self-Evolution for Agent Harnesses}},
  author = {Cai, Qianshu and Zhang, Yonggang and Nie, Jun and Ran, Maohao and Zheng, Huajiang and Song, Jun and Tian, Xinmei and Guo, Yike and Xue, Wei},
  year = {2026},
  eprint = {2609.08175},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.08175},
  url = {https://arxiv.org/abs/2609.08175v2},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 2},
  keywords = {category-generalization-evaluation-methodology-and-safety, generalization, evaluation methodology, and safety},
}

@misc{harness_learning_limits2026,
  title = {{Harness Evolution as Learning: Approximation, Generalization, and Optimization Limits of Self-Improving Personal Agents}},
  author = {Gan, Zeyu and Gong, Zixuan and Liu, Yong},
  year = {2026},
  eprint = {2609.36892},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.36892},
  url = {https://arxiv.org/abs/2609.36892v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-generalization-evaluation-methodology-and-safety, generalization, evaluation methodology, and safety},
}

@misc{skillopt2026,
  title = {{SkillOpt: Executive Strategy for Self-Evolving Agent Skills}},
  author = {Yang, Yifan and Gong, Ziyang and Huang, Weiquan and Yang, Qihao and Zhou, Ziwei and Huang, Zisu and Li, Yan and Gao, Xuemei and Dai, Qi and Liu, Bei and Qiu, Kai and Yang, Yuqing and Chen, Dongdong and Yang, Xue and Luo, Chong},
  year = {2026},
  eprint = {2605.23904},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2605.23904},
  url = {https://arxiv.org/abs/2605.23904v2},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 2},
  keywords = {category-memory-context-skills-and-experience-reuse, memory, context, skills, and experience reuse},
}

@misc{skillissue2026,
  title = {{Skill Issue: Lessons from Optimizing Repository SKILLs for Coding Agents}},
  author = {Kozyrev, Mykhailo and Kozyrev, Andrei and Podkopaev, Anton},
  year = {2026},
  eprint = {2609.12742},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.12742},
  url = {https://arxiv.org/abs/2609.12742v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-generalization-evaluation-methodology-and-safety, generalization, evaluation methodology, and safety},
}

@misc{r2flow2026,
  title = {{R$^2$ Flow: Recursive Self-Improvement via Recursive Skill Evolution}},
  author = {Zhang, Mingda and Huang, Qiang and Li, Yanjin and Wang, Zijia and Lin, Qika and Tang, Xiaoying and Shen, Tiesunlong},
  year = {2026},
  eprint = {2609.33867},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.33867},
  url = {https://arxiv.org/abs/2609.33867v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-model-and-policy-training-harness-co-evolution-and-distillation, self-modification and meta-level improvement},
}

@misc{rsi_onpolicy_distillation2026,
  title = {{Recursive Self-Improvement via On-Policy Distillation for Reasoning}},
  author = {Yin, Shangjian and Zhao, Zehao and Asadi, Kavosh and Liu, Rui and Lu, Yuchen and Mei, Shike and Cui, Hang and Simon, Luke and Shi, Zhouxing and Firooz, Hamed},
  year = {2026},
  eprint = {2609.30652},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.30652},
  url = {https://arxiv.org/abs/2609.30652v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-model-and-policy-training-harness-co-evolution-and-distillation, model--harness co-evolution and self-training},
}

@misc{liteevo2026,
  title = {{LiteEvo: Automated, Cost-Efficient Harness Evolution for Generalization to Unseen Tasks}},
  author = {Choi, Euntae and Song, Sumin and Yoo, Sungjoo},
  year = {2026},
  eprint = {2609.33146},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.33146},
  url = {https://arxiv.org/abs/2609.33146v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-harness-optimization-and-search-workflows, harness optimization and search workflows},
}

@misc{robot_harness_distillation2026,
  title = {{Recursive Harness Distillation across Agents for Robot Manipulation}},
  author = {Kim, Seungyeon and Lee, Junhoo and Kim, Minkyu and Kim, Baekseung and Kwak, Nojun},
  year = {2026},
  eprint = {2609.33378},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.33378},
  url = {https://arxiv.org/abs/2609.33378v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-memory-context-skills-and-experience-reuse, memory, context, skills, and experience reuse},
}

@misc{agent_harness_extraction2026,
  title = {{Agent Harness Distillation: Inference-Time Harness Extraction and Exploitation in Autonomous Multi-Agent Systems}},
  author = {Cui, Yu and Yang, Wuli and Shi, Yirui and Xia, Junhao and Jiang, Hui and Gao, Lei and Bao, Chenfu},
  year = {2026},
  eprint = {2607.28147},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2607.28147},
  url = {https://arxiv.org/abs/2607.28147v4},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 4},
  keywords = {category-generalization-evaluation-methodology-and-safety, generalization, evaluation methodology, and safety},
}

@misc{hardware_harness_evolution2026,
  title = {{Automatic Harness Evolution for Hardware Design Verification: Can LLMs Consolidate Gains Across Discovered Harnesses?}},
  author = {Seyoum, Kidus and Mittur, Ajay},
  year = {2026},
  eprint = {2609.28908},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.28908},
  url = {https://arxiv.org/abs/2609.28908v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-generalization-evaluation-methodology-and-safety, generalization, evaluation methodology, and safety},
}

@misc{last_harness2026,
  title = {{The Last Harness You'll Ever Build}},
  author = {Seong, Haebin and Yin, Li and Zhang, Haoran and Shi, Zhan},
  year = {2026},
  eprint = {2604.21003},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2604.21003},
  url = {https://arxiv.org/abs/2604.21003v3},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 3},
  keywords = {category-self-modification-and-meta-level-improvement, self-modification and meta-level improvement},
}

@misc{scholarevolve2026,
  title = {{Learning from Research: Toward Lifelong Agent Harness Evolution}},
  author = {Yang, Jingbo and Lai, Kwei-Herng and Wang, Xiaowen and Harari, Yaar and Gabrilovich, Evgeniy and Chang, Shiyu},
  year = {2026},
  eprint = {2609.40169},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.40169},
  url = {https://arxiv.org/abs/2609.40169v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-harness-optimization-and-search-workflows, harness optimization and search workflows},
}

@misc{robot_scale_selection2026,
  title = {{Scale and Selection: What Makes Automatic Harness Evolution Work for Visual-Interface Robot Agents}},
  author = {Wei, Zhijie and Tan, Ferris and Wang, Jinghui},
  year = {2026},
  eprint = {2609.39304},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.39304},
  url = {https://arxiv.org/abs/2609.39304v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-generalization-evaluation-methodology-and-safety, generalization, evaluation methodology, and safety},
}

@misc{evogen_harness2026,
  title = {{EvoGen-Harness: Learning Where and How to Evolve Image-Generation Harnesses}},
  author = {Luo, Jiabin and Liu, Yinan and Meng, Chunlei and Guo, Yufei},
  year = {2026},
  eprint = {2610.00383},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.00383},
  url = {https://arxiv.org/abs/2610.00383v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-harness-optimization-and-search-workflows, harness optimization and search workflows},
}

@misc{videorsi2026,
  title = {{Video-RSI: Recursive Self-Improvement of Video Understanding Agents via Harness Evolution}},
  author = {Luo, Bingjun and Guo, Jialin and Li, Siqi},
  year = {2026},
  eprint = {2609.37950},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.37950},
  url = {https://arxiv.org/abs/2609.37950v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-harness-optimization-and-search-workflows, harness optimization and search workflows},
}

@misc{rsigame2026,
  title = {{RSIGame: Autonomous Agentic Game Development with Recursive Self-improvement}},
  author = {Wu, Wenyi and Fu, Minghao and You, Jieyu and Zhou, Kun and Liu, Siqi and Salvi, Aayush and Lin, Yiheng and Zhang, Ce and Lan, Xiaohan and Zhu, Jiahui and Zhong, Yujie and She, Qi and Huang, Biwei},
  year = {2026},
  eprint = {2609.39045},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.39045},
  url = {https://arxiv.org/abs/2609.39045v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-model-and-policy-training-harness-co-evolution-and-distillation, model--harness co-evolution and self-training},
}

@misc{schema2026,
  title = {{Schema: Discovering Unknown Environments via Agentic Program Induction}},
  author = {Zeng, Guanning and Wang, Jiani and Ma, Wenjie and Yin, Shaofeng and Wang, Chenyang and Liu, Shichen and Kanazawa, Angjoo and Ni, Wode and Li, Xiuyu and Zanette, Andrea and Feng, Haiwen},
  year = {2026},
  eprint = {2609.39140},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.39140},
  url = {https://arxiv.org/abs/2609.39140v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-agent-architectures-and-harness-infrastructure, agent architectures and harness infrastructure},
}

@misc{activesaddler2026,
  title = {{ActiveSaddler: Automated Curriculum Learning for Agent Harness Optimization}},
  author = {Park, Sungho and Kim, Wonjoong and Zhang, Jue and Han, Wook-Shin and Gao, Pengfei and Park, Chanyoung and Yao, Yongqiang and Fu, Rao and Nallipogu, Elsie and Lin, Qingwei and Rühle, Victor},
  year = {2026},
  eprint = {2610.00906},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.00906},
  url = {https://arxiv.org/abs/2610.00906v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-harness-optimization-and-search-workflows, harness optimization and search workflows},
}

@misc{navharness2026,
  title = {{NavHarness: Towards Lifelong Embodied Navigation}},
  author = {Zhao, Xunyi and Zhou, Jian and Lin, Sihao and Zhou, Gengze and Li, Zerui and Yan, Xinyu and Liu, Jiajun and van den Hengel, Anton and Wu, Qi},
  year = {2026},
  eprint = {2609.34276},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.34276},
  url = {https://arxiv.org/abs/2609.34276v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-memory-context-skills-and-experience-reuse, memory, context, skills, and experience reuse},
}

@inproceedings{coevolving_correction2026,
  title = {{Co-Evolving Harnesses and Models: On-Policy Correction Helps Weaker Models Catch Up Where Imitation Fails}},
  author = {Yu, Zhou and Bi, Bin and Pentyala, Shiva Kumar and Mehrotra, Shubham and Chaudhuri, Sougata and Bhagavath, Shilpa and Chen, Zeyuan and Xu, Ran and Zhu, James and Asur, Sitaram and Mui, Phil L},
  year = {2026},
  url = {https://openreview.net/forum?id=jgc0U3zRFp},
  urldate = {2026-10-10},
  note = {Survey source: arXiv:2609.09134v1; Accepted workshop paper; forthcoming},
  keywords = {category-model-and-policy-training-harness-co-evolution-and-distillation, model--harness co-evolution and self-training},
  booktitle = {NeurIPS 2026 Workshop on Continual Learning for Enterprise AI Agents},
}

@misc{minimal_mle_harness2026,
  title = {{How Much of a Harness Does a Strong Agent Need for Autonomous ML Engineering?}},
  author = {Brilliantov, Kirill and Hernández-Cano, Alejandro and Abbé, Emmanuel},
  year = {2026},
  eprint = {2609.40303},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.40303},
  url = {https://arxiv.org/abs/2609.40303v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-generalization-evaluation-methodology-and-safety, generalization, evaluation methodology, and safety},
}

@inproceedings{coharness2026,
  title = {{Co-Harness: Co-Evolving Harnesses and Model for Agents}},
  author = {Chen, Zhengyu and Xiao, Teng and Zhu, Huaisheng and Yuan, Yige and Zhang, Luan and Wang, Jingang},
  year = {2026},
  url = {https://openreview.net/forum?id=qDC8LmkpSR},
  urldate = {2026-10-10},
  note = {Survey source: arXiv:2607.22688v1},
  keywords = {category-model-and-policy-training-harness-co-evolution-and-distillation, model--harness co-evolution and self-training},
  booktitle = {COLM 2026 The 2nd Workshop on Lifelong Agents: Learning, Aligning, and Evolving},
}

@misc{harbor2026,
  title = {{HARBOR: Automated Harness Optimization}},
  author = {Sengupta, Biswa and Wang, Jinhua},
  year = {2026},
  eprint = {2604.20938},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2604.20938},
  url = {https://arxiv.org/abs/2604.20938v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-harness-optimization-and-search-workflows, harness optimization and search workflows},
}

@misc{sclate2026,
  title = {{SCLATE: a Substrate for Continual-Learning Agent Training and Evaluation}},
  author = {Jung, Youngmok and Salekin, Sirajul and Tran, Henry and Movellan, Javier and Huang, Zhao and Bilkhu, Manjot},
  year = {2026},
  eprint = {2609.32391},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.32391},
  url = {https://arxiv.org/abs/2609.32391v2},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 2},
  keywords = {category-agent-architectures-and-harness-infrastructure, agent architectures and harness infrastructure},
}

@misc{longhorizon_harness2026,
  title = {{LongHorizon-Harness: Advancing Long-Horizon Agents for Real-World Tasks}},
  author = {Ma, Ziyu and Huang, Hailang and Zou, Shun and Wang, Yong and Yang, Shidong and Hu, Yiming and Wei, Fei and Chu, XiangXiang},
  year = {2026},
  eprint = {2608.01964},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.01964},
  url = {https://arxiv.org/abs/2608.01964v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-agent-architectures-and-harness-infrastructure, agent architectures and harness infrastructure},
}

@misc{memcodex2026,
  title = {{MemCodex: Self-Programming Hierarchical Memory for Language Agents}},
  author = {Wang, Xiaoqiang and Liu, Bang},
  year = {2026},
  eprint = {2609.39765},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.39765},
  url = {https://arxiv.org/abs/2609.39765v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-memory-context-skills-and-experience-reuse, memory, context, skills, and experience reuse},
}

@misc{genmem2026,
  title = {{GenMem: Generative Symbolic Memory for Self-Evolving Harness}},
  author = {Jiang, Xinke and Feng, Tao and Xu, Weixuan and Zhang, Zhixin and Yang, Zhibang and Zhang, Wentao and Zhu, Runchuan and Chu, Xu and Zhao, Junfeng and Wang, Yasha},
  year = {2026},
  eprint = {2609.34633},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.34633},
  url = {https://arxiv.org/abs/2609.34633v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-memory-context-skills-and-experience-reuse, memory, context, skills, and experience reuse},
}

@misc{chime2026,
  title = {{CHIME: Credit-Aware Hierarchical Memory Evolution for Long-Horizon Agentic Planning}},
  author = {Ye, Yongshi and Lan, Tian and Jiang, Feihu and Ye, Muyang and Zhu, Bin and Jia, Qianghuai and Wang, Longyue and Xu, Zhao and Luo, Weihua and Shi, Xiaodong},
  year = {2026},
  eprint = {2609.02074},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.02074},
  url = {https://arxiv.org/abs/2609.02074v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-memory-context-skills-and-experience-reuse, memory, context, skills, and experience reuse},
}

@misc{selfindex2026,
  title = {{Self-Evolving Search Index}},
  author = {Lee, Sangam and Lee, Wonjae and Kim, Sunghwan and Kim, Deogyong and Kim, Jaehoon and Nam, Daye and Kang, SeongKu and Lee, Dongha},
  year = {2026},
  eprint = {2609.19656},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.19656},
  url = {https://arxiv.org/abs/2609.19656v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-memory-context-skills-and-experience-reuse, memory, context, skills, and experience reuse},
}

@misc{cobraskills2026,
  title = {{COBRA-Skills: Contextual Bandit-Guided Evolution for Agent Skill Optimization}},
  author = {Lu, Pingchen and Wang, Xiangyi and Li, Xiang and Mao, Jie and Qu, Zikun and Luo, Junfeng and Shu, Yao and Low, Bryan Kian Hsiang and Dai, Zhongxiang},
  year = {2026},
  eprint = {2609.11682},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.11682},
  url = {https://arxiv.org/abs/2609.11682v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-harness-optimization-and-search-workflows, harness optimization and search workflows},
}

@misc{sagegate2026,
  title = {{SAGE: A Statistical Acceptance Gate for Self-Evolving Agents}},
  author = {Wang, Yihao and Xia, Linhan and Liu, Rui and Zhang, Zhaofeng and Wu, Hongyu and Yang, Yang and He, Jinglu and Guo, Yu and Lei, Kai},
  year = {2026},
  eprint = {2609.36043},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.36043},
  url = {https://arxiv.org/abs/2609.36043v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-generalization-evaluation-methodology-and-safety, generalization, evaluation methodology, and safety},
}

@misc{sbco2026,
  title = {{SBCO: Self-Supervised, Verifier-Grounded Harness Optimization For Planning Agents}},
  author = {Kulkarni, Vivek and Paul, Sudipta and Kumar, Aounon and Tzou, Nicholas and Chappidi, Srinivas},
  year = {2026},
  eprint = {2608.10157},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.10157},
  url = {https://arxiv.org/abs/2608.10157v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-harness-optimization-and-search-workflows, harness optimization and search workflows},
}

@misc{selfop2026,
  title = {{SelfOp: An Optimization Algorithm for Self-Improving Security Agents}},
  author = {Ullah, Saad and Kaya, Yigitcan and Kruegel, Christopher and Vigna, Giovanni and Stringhini, Gianluca},
  year = {2026},
  eprint = {2609.22792},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.22792},
  url = {https://arxiv.org/abs/2609.22792v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-prompt-workflow-and-program-optimization, prompt, workflow, and program optimization},
}

@misc{envrethink2026,
  title = {{Breaking the Environment Wall: A Unified Framework for Preparing and Evolving Agent-Native Environments}},
  author = {Wu, Yukai and Yang, Yuanjing and Zhou, Le and Han, Shaokun and Wang, Haoyu and Tang, Zirui and Zhu, Xuzhou and Zheng, Weihuang and Pan, Maxm and Zhou, Xuanhe and Wu, Fan},
  year = {2026},
  eprint = {2609.29773},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.29773},
  url = {https://arxiv.org/abs/2609.29773v2},
  urldate = {2026-10-10},
  note = {Survey source: arXiv:2609.29773v1},
  keywords = {category-agent-architectures-and-harness-infrastructure, agent architectures and harness infrastructure},
}

@misc{autorecsys2026,
  title = {{Auto-RecSys: Harnessing Autonomous Research Agents for Industry-Scale Recommender System}},
  author = {Li, Ming and Li, Dai and Ning, Xuying and Sun, Bo and Li, Rui and Zhang, Yi and Gong, Silvia and Cao, Xuan and Li, Rui and Carapcea, Cornelia and Zhang, Qunshu and Wang, Zhigang and Xia, Yinglong and Wang, Andy},
  year = {2026},
  eprint = {2609.10922},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.10922},
  url = {https://arxiv.org/abs/2609.10922v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-agent-architectures-and-harness-infrastructure, agent architectures and harness infrastructure},
}

@misc{seabench2026,
  title = {{SEABench: Benchmarking Endogenous Misalignment In Self-Evolving Agents}},
  author = {Das, Saswat and Viswanathan, Parvati and Donnelly, Daniel and Huang, Chang and Abdelnabi, Sahar and Fioretto, Ferdinando},
  year = {2026},
  eprint = {2609.35596},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.35596},
  url = {https://arxiv.org/abs/2609.35596v2},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 2},
  keywords = {category-generalization-evaluation-methodology-and-safety, generalization, evaluation methodology, and safety},
}

@misc{proctor2026,
  title = {{LLM-as-a-Judge Is Not an Oracle: Why Self-Improving Agents Need Deterministic Guardrails}},
  author = {Wahi, Vansh},
  year = {2026},
  eprint = {2609.02246},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.02246},
  url = {https://arxiv.org/abs/2609.02246v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-generalization-evaluation-methodology-and-safety, generalization, evaluation methodology, and safety},
}

@misc{enumgrpo2026,
  title = {{Cost-Aware Optimization for Agentic Query Execution}},
  author = {Nie, Lunyiu and Xia, Yilin and Liu, Yiren and Jermaine, Christopher and Chaudhuri, Swarat},
  year = {2026},
  eprint = {2606.03152},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2606.03152},
  url = {https://arxiv.org/abs/2606.03152v2},
  urldate = {2026-10-10},
  note = {Survey source: arXiv:2606.03152v1},
  keywords = {category-prompt-workflow-and-program-optimization, prompt, workflow, and program optimization},
}

@inproceedings{memevolve2025,
  title = {{MemEvolve: Meta-Evolution of Agent Memory Systems}},
  author = {Zhang, Guibin and Ren, Haotian and Zhan, Chong and Wang, Junhao and Zhu, He and Zhou, Wangchunshu and Yan, Shuicheng},
  year = {2026},
  url = {https://openreview.net/forum?id=qpkG0eKx4v},
  urldate = {2026-10-10},
  note = {Survey source: arXiv:2512.18746v1},
  keywords = {category-memory-context-skills-and-experience-reuse, memory, context, skills, and experience reuse},
  booktitle = {Forty-third International Conference on Machine Learning},
}

@misc{evomemory2025,
  title = {{Evo-Memory: Benchmarking LLM Agent Test-time Learning with Self-Evolving Memory}},
  author = {Wei, Tianxin and Sachdeva, Noveen and Coleman, Benjamin and He, Zhankui and Bei, Yuanchen and Ning, Xuying and Ai, Mengting and Li, Yunzhe and He, Jingrui and Chi, Ed H. and Wang, Chi and Chen, Shuo and Pereira, Fernando and Kang, Wang-Cheng and Cheng, Derek Zhiyuan},
  year = {2025},
  eprint = {2511.20857},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2511.20857},
  url = {https://arxiv.org/abs/2511.20857v2},
  urldate = {2026-10-10},
  note = {Survey source: arXiv:2511.20857v1},
  keywords = {category-memory-context-skills-and-experience-reuse, memory, context, skills, and experience reuse},
}

@inproceedings{reme2025,
  title = {{Remember Me, Refine Me: A Dynamic Procedural Memory Framework for Experience-Driven Agent Evolution}},
  author = {Cao, Zouying and Deng, Jiaji and Yu, Li and Zhou, Weikang and Liu, Zhaoyang and Ding, Bolin and Zhao, Hai},
  year = {2026},
  url = {https://aclanthology.org/2026.findings-acl.829/},
  urldate = {2026-10-10},
  note = {Survey source: arXiv:2512.10696v1},
  keywords = {category-memory-context-skills-and-experience-reuse, memory, context, skills, and experience reuse},
  doi = {10.18653/v1/2026.findings-acl.829},
  pages = {16803--16822},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  publisher = {Association for Computational Linguistics},
}

@misc{milo2026,
  title = {{MILO: Automated Harness Discovery via Orchestrated Multi-Agent Evolution}},
  author = {Jana, Prithwish and Goswami, Mononito and Liu, Hao and Li, Xinyu and Huang, Langlin and Huang, Zhehui and Huang, Zhishen and Bl{\"o}baum, Patrick and Deoras, Anoop and Jain, Purak and Kanakaris, Nikos and Genc, Sahika},
  year = {2026},
  eprint = {2609.38349},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.38349},
  url = {https://arxiv.org/abs/2609.38349v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-harness-optimization-and-search-workflows, harness optimization and search workflows},
}

@misc{jitagent2026,
  title = {{JIT-Agent: Scaling Harness Intelligence via Just-in-Time Harness Evolution}},
  author = {Zhang, Guibin and Lu, Leo and Xie, Fangzhou and Zhu, Kang and Wang, Junhao and Xie, Zhifei and Yu, Zhaochen and Liu, Zihang and Sun, Zhongxiang and Li, Qiankun and Liao, Yue and Chang, Heng and Hu, Xiaobin and Ren, Qibing and Zhou, Wangchunshu and Hu, Chuanrui and Deng, Yafeng and Yan, Shuicheng},
  year = {2026},
  eprint = {2608.25593},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.25593},
  url = {https://arxiv.org/abs/2608.25593v2},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 2},
  keywords = {category-harness-optimization-and-search-workflows, harness optimization and search workflows},
}

@misc{harde2026,
  title = {{HARDE: Optimizing Agent Harnesses for Runtime Risk Detection and Execution Control}},
  author = {Liu, Zhuo and Li, Moxin and Ma, Zhixin and Shi, Wentao and Wang, Wenjie and Feng, Fuli},
  year = {2026},
  eprint = {2609.38291},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.38291},
  url = {https://arxiv.org/abs/2609.38291v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-harness-optimization-and-search-workflows, harness optimization and search workflows},
}

@misc{harnessrisk2026,
  title = {{HarnessRisk: A Lifecycle-Oriented Benchmark for Agent Harness Safety}},
  author = {Bai, Yajing and Duan, Jinhao and Peng, Jie and Wu, Xianfeng and Liu, Sijia and Wang, Song and Chen, Tianlong},
  year = {2026},
  eprint = {2608.17597},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.17597},
  url = {https://arxiv.org/abs/2608.17597v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-benchmarks-for-harness-optimization-and-adaptation, benchmarks for harness optimization and adaptation},
}

@inproceedings{harness_disclosure2026,
  title = {{Stop Comparing {LLM} Agents Without Disclosing the Harness}},
  author = {Zhang, Yunbei and Wang, Janet and Ge, Yingqiang and Xu, Weijie and Hamm, Jihun and Reddy, Chandan K.},
  year = {2026},
  url = {https://openreview.net/forum?id=ffKHSraOIK},
  urldate = {2026-10-10},
  note = {Survey source: arXiv:2605.23950v1},
  keywords = {category-generalization-evaluation-methodology-and-safety, generalization, evaluation methodology, and safety},
  booktitle = {ICML 2026 Second Workshop on Agents in the Wild: Safety, Security, and Beyond},
}

@misc{system_scaling2026,
  title = {{From Model Scaling to System Scaling: Scaling the Harness in Agentic AI}},
  author = {Gu, Shangding},
  year = {2026},
  eprint = {2605.26112},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2605.26112},
  url = {https://arxiv.org/abs/2605.26112v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-surveys-definitions-and-conceptual-frameworks, surveys, definitions, and conceptual frameworks},
}

@inproceedings{hasp2026,
  title = {{Harnessing LLM Agents with Skill Programs}},
  author = {Liu, Hongjun and Ming, Yifei and Joty, Shafiq and Zhao, Chen},
  year = {2026},
  eprint = {2605.17734},
  archiveprefix = {arXiv},
  url = {https://arxiv.org/abs/2605.17734v2},
  urldate = {2026-10-10},
  note = {Accepted for publication; forthcoming; Survey source: arXiv:2605.17734v2},
  keywords = {category-model-and-policy-training-harness-co-evolution-and-distillation, model and policy training, harness co-evolution, and distillation},
  booktitle = {Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing},
}

@misc{model_or_harness2026,
  title = {{Model or Harness? An Interaction-Centric Taxonomy for Localizing Agent Failures}},
  author = {Raj, Harsh and Gupta, Vipul and Mahmoud, Anas and Dumitru, Razvan-Gabriel and Yi, Darvin and Sabharwal, Aakash and He, Yunzhong},
  year = {2026},
  eprint = {2607.28802},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2607.28802},
  url = {https://arxiv.org/abs/2607.28802v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-generalization-evaluation-methodology-and-safety, generalization, evaluation methodology, and safety},
}

@misc{primeagent2026,
  title = {{Prime Agent: A Self-Improving RLM Harness}},
  author = {Karten, Seth and Zhang, Alex L. and Thomas, Kevin and M{\"u}ller, Sebastian and Bakouch, Elie and Auras, Daniel and Senghaas, Mika and Obeid, Fares and Dunas, Konstantin and Hagemann, Johannes and Jaghouar, Sami},
  year = {2026},
  eprint = {2608.23552},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.23552},
  url = {https://arxiv.org/abs/2608.23552v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-agent-architectures-and-harness-infrastructure, agent architectures and harness infrastructure},
}

@misc{harness_effects2026,
  title = {{Beyond the Model: Demystifying Harness Effects in Software Engineering Agents}},
  author = {Hu, Haichuan and Zhang, Quanjun and Yu, Shengcheng and Chen, Zhifei and Luo, Tianyu and Fang, Chunrong and Chen, Zhenyu and Xiao, Liang},
  year = {2026},
  eprint = {2609.32459},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.32459},
  url = {https://arxiv.org/abs/2609.32459v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-generalization-evaluation-methodology-and-safety, generalization, evaluation methodology, and safety},
}

@misc{task_aware_harness2026,
  title = {{Task-Aware Harness Provisioning for LLM Agents in Mission-Critical Infrastructure Operations}},
  author = {Lin, Liangtao and Zhang, Qingang and Zhu, Zhaomeng and Zhang, Tianwei and Wen, Yonggang},
  year = {2026},
  eprint = {2608.17433},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.17433},
  url = {https://arxiv.org/abs/2608.17433v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-agent-architectures-and-harness-infrastructure, agent architectures and harness infrastructure},
}

@misc{meta_engineering2026,
  title = {{Meta-Engineering Harnesses for AI-Native Software Production: A Contract-Driven Adversarial Verification Architecture with Early Deployment Report}},
  author = {Sengupta, Satadru and Briggs, Tamunokorite and Myshakivskyi, Ivan},
  year = {2026},
  eprint = {2605.25665},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2605.25665},
  url = {https://arxiv.org/abs/2605.25665v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-agent-architectures-and-harness-infrastructure, agent architectures and harness infrastructure},
}

@misc{darwinx2026,
  title = {{DarwinX: Evolving Agent Harnesses Through Natural Selection}},
  author = {Zhang, Yifan and Dai, Yutong and Tan, Juntao and Yang, Luyu and Mullur, Rishi and Hoang, Thai and Hu, Zhiyuan and Zhu, James and Mui, Phil and Savarese, Silvio and Xu, Ran and Chen, Zeyuan},
  year = {2026},
  eprint = {2608.07545},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.07545},
  url = {https://arxiv.org/abs/2608.07545v1},
  urldate = {2026-10-02},
  keywords = {category-harness-optimization-and-search-workflows, section-2},
}

@misc{evolvenet2026,
  title = {{EvolveNet: Collaborative Harness Evolution for Agent Self-Improvement}},
  author = {Nie, Jun and Zhang, Yonggang and Cai, Qianshu and Cheung, Yiu-ming and Tian, Xinmei and Han, Bo},
  year = {2026},
  eprint = {2608.04968},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.04968},
  url = {https://arxiv.org/abs/2608.04968v1},
  urldate = {2026-10-02},
  note = {arXiv preprint, version 1},
  keywords = {category-harness-optimization-and-search-workflows, harness optimization and search workflows},
}

@misc{selfevolving_coding_survey2026,
  title = {{Self-Evolving Coding Agents}},
  author = {Zhou, Hao and Hu, Haichuan and Luo, Tianyu and Shang, Ye and Fang, Chunrong and Chen, Zhenyu and Xiao, Liang and Zhang, Quanjun},
  year = {2026},
  eprint = {2608.03392},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.03392},
  url = {https://arxiv.org/abs/2608.03392v4},
  note = {arXiv preprint, version 4},
  urldate = {2026-10-04},
  keywords = {category-surveys-definitions-and-conceptual-frameworks, prior-survey, introduction-candidate, domain-specific-survey, section-4},
}

@misc{task_completion_harness_survey2026,
  title = {{From Question Answering to Task Completion: A Survey on Agent System and Harness Design}},
  author = {Guo, Jianyuan and Hao, Zhiwei and Wang, Chengcheng and Fan, Cheng and Luo, Tingzhang and Li, Hongguang and Gao, Ying and Mei, Hefei and Peng, Jiankun and Xu, Rongjian and Dong, Minjing and Wu, Han and Zheng, Mengyu and Han, Kai and Wang, Shiqi and Xu, Chang and Wang, Yunhe},
  year = {2026},
  eprint = {2606.20683},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2606.20683},
  url = {https://arxiv.org/abs/2606.20683v1},
  note = {arXiv preprint, version 1},
  urldate = {2026-10-04},
  keywords = {category-surveys-definitions-and-conceptual-frameworks, prior-survey, introduction-candidate, harness-architecture-survey, section-2, section-4},
}

@misc{adas_survey2026,
  title = {{Automated Design of Agentic Systems: A Survey of Algorithms for Searching, Optimizing, and Evolving LLM Agents, Workflows, and Prompts}},
  author = {Mad\v{z}ar, Maksim and Mekterovi\'{c}, Igor},
  year = {2026},
  url = {https://www.preprints.org/manuscript/202606.0238/v1},
  note = {Preprints.org preprint, version 1; posted 2 June 2026; not peer-reviewed},
  urldate = {2026-10-04},
  keywords = {category-surveys-definitions-and-conceptual-frameworks, prior-survey, introduction-candidate, automated-design-survey, section-2},
}

@inproceedings{workflow_optimization_survey2026,
  title = {{From Static Templates to Dynamic Runtime Graphs: A Survey of Workflow Optimization for LLM Agents}},
  author = {Yue, Ling and Bhandari, Kushal Raj and Ko, Ching-Yun and Patel, Dhaval and Lin, Shuxin and Zhou, Nianjun and Gao, Jianxi and Chen, Pin-Yu and Pan, Shaowu},
  year = {2026},
  eprint = {2603.22386},
  archiveprefix = {arXiv},
  url = {https://arxiv.org/abs/2603.22386v1},
  note = {Accepted for publication; forthcoming; Survey source: arXiv:2603.22386v1},
  urldate = {2026-10-10},
  keywords = {category-surveys-definitions-and-conceptual-frameworks, prior-survey, introduction-candidate, workflow-optimization-survey, section-2},
  booktitle = {Findings of the Association for Computational Linguistics: EMNLP 2026},
}

@misc{experience_selfimprovement_survey2026,
  title = {{Self-Improving Agents in the Era of Experience: A Survey of Self- to Meta-Evolution}},
  author = {Jiang, Che and Zhong, Jincheng and Fu, Yu and Tian, Kai and Yang, Junlin and Zhao, Kaikai and Wang, Yuchong and Luo, Tianwei and Wang, Weizhi and Zuo, Yuxin and Jia, Guoli and Lv, Xingtai and Lei, Dianqiao and Zeng, Sihang and Wang, Yuru and Yuan, Zhenzhao and Long, Xinwei and Hua, Ermo and Ren, Can and Jiang, Xin and Xie, Shulei and Zheng, Yuanchun and Sun, Youbang and Qi, Biqing and Ding, Ning and Zhang, Kaiyan and Zhou, Bowen},
  year = {2026},
  url = {https://openreview.net/forum?id=IUltZSgLMm},
  note = {Public manuscript on OpenReview; author-released PDF dated 25 June 2026},
  urldate = {2026-10-04},
  keywords = {category-surveys-definitions-and-conceptual-frameworks, prior-survey, introduction-candidate, experience-and-meta-evolution-survey, section-2, section-3},
}

@misc{agent_systems_harness_survey2026,
  title = {{Agent Systems with Harness Engineering}},
  author = {Tang, Xinyu and Peng, Han and Chen, Guoxin and Shi, Yuze and Su, Zitao and Liu, Peiyu and Zhao, Wayne Xin and Li, Yawen and Xue, Zhe},
  year = {2026},
  url = {https://openreview.net/forum?id=nM5tDHrQsx},
  note = {Public manuscript on OpenReview; author-released PDF dated 18 May 2026},
  urldate = {2026-10-04},
  keywords = {category-surveys-definitions-and-conceptual-frameworks, prior-survey, introduction-candidate, harness-architecture-survey, section-2, section-4},
}

@misc{midharness2026,
  title = {{Mid-Harness: Scaling Actions Between Model and Harness for Terminal Agents}},
  author = {Kang, Minki and Hachiuma, Ryo and Zhang, Shaokun and Radhakrishnan, Subhashree and Fu, Yonggan and Jiang, Jindong and Liu, Mingjie and Hosseini-Asl, Ehsan and Dong, Yi and Wang, Yu-Chiang Frank and Lee, Byung-Kwan},
  year = {2026},
  eprint = {2609.39982},
  archiveprefix = {arXiv},
  url = {https://arxiv.org/abs/2609.39982v1},
  note = {arXiv preprint, version 1; submitted 30 September 2026},
  urldate = {2026-10-04},
  keywords = {category-task-time-search-and-output-refinement, adjacent-work, action-verification, verifier-distillation, section-2},
}

@misc{wu_rsi_survey2026,
  title = {{Recursive Self-Improvement in AI: A Survey}},
  author = {Wu, Hao and Xu, Fan and Zhang, Fan and Xu, Zhipeng and Zhang, Yongheng and Li, Yinghui and Zhao, Yizhou and Zhao, Penghao and Yao, Weiran and Li, Zengxiang and Gao, Yuan and Gao, Zhangyang and Wang, Hanbin and Tao, Ruihan and Wang, Wenjie and Li, Qi and Fang, Junfeng and Ma, Xingjun and Li, Qiankun and Luo, Xiao and Zheng, Yefeng and Liu, Zhiwei and Wang, Stephen and Wang, Kun and Liu, Yang and Jiang, Yugang and Wang, Liang and Liu, Nan and Yu, Philip S. and Xiong, Caiming and Yang, Qiang and Wen, Qingsong},
  year = {2026},
  url = {https://www.preprints.org/manuscript/202609.2680/v1},
  note = {Preprints.org preprint, version 1; posted 30 September 2026; not peer-reviewed},
  urldate = {2026-10-04},
  keywords = {category-surveys-definitions-and-conceptual-frameworks, prior-survey, introduction-candidate, recursive-evidence, section-3},
}

@misc{openrsi_index,
  title = {{OpenRSI Index}},
  author = {{OpenRSI Foundation}},
  year = {n.d.},
  howpublished = {Official benchmark website and repository},
  url = {https://index.openrsi.foundation/},
  note = {Preview v0.1; official resource, not a verified research-paper publication; repository: \url{https://github.com/OpenRSI-Foundation/OpenRSI-Index}; accessed 4 October 2026},
  urldate = {2026-10-04},
  keywords = {category-official-benchmark-and-implementation-resources, official-resource, benchmark, supplemental-resource, section-4},
}

@misc{mltoolbench2026,
  title = {{MLToolBench: Learning Tool-Augmented Agents for Machine Learning Development}},
  author = {Yu, Xin and Zhang, Lizhu and Bai, Jiamu and Wu, Yanhong and Wang, Zellux and Li, Serena and Li, Weiwei and Xue, Lingzhou and Fan, Xiangjun and Peng, Bo},
  year = {2026},
  eprint = {2609.36679},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.36679},
  url = {https://arxiv.org/abs/2609.36679v2},
  urldate = {2026-10-10},
  note = {Survey source: arXiv:2609.36679v1},
  keywords = {adjacent-or-supporting-work, model and policy training, harness co-evolution, and distillation},
}

@misc{pbsd2026,
  title = {{Preference-Based Self-Distillation: Beyond KL Matching via Reward Regularization}},
  author = {Yu, Xin and Liao, Liuchen and Zhang, Yiwen and Yu, Yingchen and Xue, Lingzhou and Guo, Qinzhen},
  year = {2026},
  eprint = {2605.05040},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2605.05040},
  url = {https://arxiv.org/abs/2605.05040v2},
  urldate = {2026-10-10},
  note = {Survey source: arXiv:2605.05040v1},
  keywords = {adjacent-or-supporting-work, model and policy training, harness co-evolution, and distillation},
}

@misc{servicecontrol2026,
  title = {{When Should Service Agents Reconsider? Difficulty-Routed Control in Customer-Service Operations}},
  author = {Chen, Qian and Liu, Chengyuan and Yu, Xin},
  year = {2026},
  eprint = {2607.01426},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2607.01426},
  url = {https://arxiv.org/abs/2607.01426v1},
  urldate = {2026-10-04},
  note = {arXiv preprint, version 1},
  keywords = {adjacent-or-supporting-work, agent architectures and harness infrastructure},
}

@misc{curveshift2026,
  title = {{CurveShift: Is Agent Progress Scalar? Separating Level from Shape}},
  author = {Xing, Hanwen and Wang, Pengyun and Meng, BingXu and Alhamoud, Kumail and Li, Xiang and Wang, Jicheng and Yu, Xin and Han, Xinyang and Li, Xiaomin and Torr, Philip and Hao, Yuexing},
  year = {2026},
  eprint = {2608.00355},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.00355},
  url = {https://arxiv.org/abs/2608.00355v1},
  urldate = {2026-10-04},
  note = {arXiv preprint, version 1},
  keywords = {adjacent-or-supporting-work, generalization, evaluation methodology, and safety},
}

@misc{chan2026intelligenceexplosion,
  title = {{What if automating AI R\&D triggers an intelligence explosion?}},
  author = {Chan, Alan and Winter, Christoph and Barto, Andrew and Pachocki, Jakub and Hinton, Geoffrey and Horvitz, Eric and Bengio, Yoshua and Song, Dawn and Clark, Jack and Greaves, Hilary and Korinek, Anton and Hammond, Samuel and Graepel, Thore and Bariach, Ben and Torr, Philip H. S. and McIlraith, Sheila A. and Clune, Jeff and Manning, Sam and Sastry, Girish and Davidson, Tom and Eth, Daniel and Mindermann, S{\"o}ren},
  year = {2026},
  eprint = {2609.36054},
  archiveprefix = {arXiv},
  primaryclass = {cs.CY},
  doi = {10.48550/arXiv.2609.36054},
  url = {https://arxiv.org/abs/2609.36054v1},
  urldate = {2026-10-05},
  note = {arXiv preprint, version 1},
  keywords = {category-surveys-definitions-and-conceptual-frameworks, position-paper, ai-r-and-d-automation, motivation, section-1},
}

@misc{lean_coevolution2026,
  title = {{Self-Modifying Lean Proof Agents with Verifier-Grounded Benchmark Coevolution}},
  author = {Li, Yuqing and Wu, Zeguan and Gan, Yu and Liu, Junyu},
  year = {2026},
  eprint = {2607.17352},
  archiveprefix = {arXiv},
  primaryclass = {cs.AI},
  doi = {10.48550/arXiv.2607.17352},
  url = {https://arxiv.org/abs/2607.17352v1},
  urldate = {2026-10-05},
  note = {arXiv preprint, version 1},
  keywords = {category-self-modification-and-meta-level-improvement, section-4, mathematical-reasoning, formal-proof},
}

@inproceedings{mermaidflow2025,
  title = {{MermaidFlow: Redefining Agentic Workflow Generation via Safety-Constrained Evolutionary Programming}},
  author = {Zheng, Chengqi and Chen, Jianda and Lyu, Yueming and Ng, Wen Zheng Terence and Zhang, Haopeng and Ong, Yew-Soon and Tsang, Ivor and Yin, Haiyan},
  year = {2025},
  eprint = {2505.22967},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2505.22967},
  url = {https://arxiv.org/abs/2505.22967v1},
  urldate = {2026-10-10},
  note = {Non-archival workshop; Survey source: arXiv:2505.22967v1},
  booktitle = {ICML 2025 Workshop on Multi-Agent Systems in the Era of Foundation Models: Opportunities, Challenges and Futures},
}

@misc{sew2025,
  title = {{SEW: Self-Evolving Agentic Workflows for Automated Code Generation}},
  author = {Liu, Siwei and Fang, Jinyuan and Zhou, Han and Wang, Yingxu and Meng, Zaiqiao},
  year = {2025},
  eprint = {2505.18646},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2505.18646},
  url = {https://arxiv.org/abs/2505.18646v2},
  urldate = {2026-10-05},
  note = {arXiv preprint, version 2},
}

@inproceedings{maas2025,
  title = {{Multi-agent Architecture Search via Agentic Supernet}},
  author = {Zhang, Guibin and Niu, Luyang and Fang, Junfeng and Wang, Kun and Bai, Lei and Wang, Xiang},
  year = {2025},
  booktitle = {Proceedings of the 42nd International Conference on Machine Learning},
  volume = {267},
  series = {Proceedings of Machine Learning Research},
  pages = {75834--75852},
  publisher = {PMLR},
  url = {https://proceedings.mlr.press/v267/zhang25bi.html},
  urldate = {2026-10-10},
  note = {Case-analysis source version: \href{https://arxiv.org/abs/2502.04180v2}{arXiv:2502.04180v2}},
}

@misc{hyevo2026,
  title = {{HyEvo: Self-Evolving Hybrid Agentic Workflows for Efficient Reasoning}},
  author = {Xu, Beibei and Ye, Yutong and Shen, Chuyun and Zhou, Yingbo and Chen, Cheng and Chen, Mingsong},
  year = {2026},
  eprint = {2603.19639},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2603.19639},
  url = {https://arxiv.org/abs/2603.19639v1},
  urldate = {2026-10-05},
  note = {arXiv preprint, version 1},
}

@inproceedings{flowevo2026,
  title = {{FlowEvo: Self-Evolving Agents through the Co-Evolution of Workflows and Executable Skills}},
  author = {Ren, Zeyu and Yue, Ling and Li, Ran and Wang, Yishu and Xu, Shengxiang and Liu, Hanmo and Pan, Shaowu and Di, Shimin},
  year = {2026},
  doi = {10.48550/arXiv.2607.21596},
  url = {https://openreview.net/forum?id=hU2N7IIkcE},
  urldate = {2026-10-10},
  note = {Survey source: arXiv:2607.21596v2},
  booktitle = {Third Conference on Language Modeling (COLM)},
}

@misc{codeskill2026,
  title = {{CODESKILL: Learning Self-Evolving Skills for Coding Agents}},
  author = {Li, Yanzhou and Zhang, Yiran and Zhang, Xiaoyu and Liu, Xiaoxia and Liu, Yang},
  year = {2026},
  eprint = {2605.25430},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2605.25430},
  url = {https://arxiv.org/abs/2605.25430v2},
  urldate = {2026-10-05},
  note = {arXiv preprint, version 2},
}

@misc{gse2026,
  title = {{Learning Globally Reusable Skills for Coding Agents}},
  author = {Yang, Chen and Tian, Jiashuo and Wang, Ziqi and Liu, Xinyin and Ye, Meiru and Chen, Junjie},
  year = {2026},
  eprint = {2608.06153},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.06153},
  url = {https://arxiv.org/abs/2608.06153v1},
  urldate = {2026-10-05},
  note = {arXiv preprint, version 1},
}

@misc{skillforge2026,
  title = {{SkillForge: Self-Distilling Agents for Project-Specific Issue Resolution}},
  author = {Chen, Silin and Li, Han and Gu, Xiaodong and Shi, Yuling and Guan, Haibing},
  year = {2026},
  eprint = {2608.18933},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.18933},
  url = {https://arxiv.org/abs/2608.18933v1},
  urldate = {2026-10-05},
  note = {arXiv preprint, version 1},
}

@misc{aibuildai2_2026,
  title = {{AIBuildAI-2: A Knowledge-Enhanced Agent for Automatically Building AI Models}},
  author = {Zhang, Ruiyi and Qin, Peijia and Cao, Qi and Zhang, Li and Xie, Pengtao},
  year = {2026},
  eprint = {2605.27873},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2605.27873},
  url = {https://arxiv.org/abs/2605.27873v1},
  urldate = {2026-10-05},
  note = {arXiv preprint, version 1},
}

@misc{wda2026,
  title = {{Streamlined Reflective Evolution for Task-Adaptive Self-Refinement Pipelines}},
  author = {Zhou, Xiaofan and Cheng, Lu},
  year = {2026},
  eprint = {2609.32458},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.32458},
  url = {https://arxiv.org/abs/2609.32458v1},
  urldate = {2026-10-05},
  note = {arXiv preprint, version 1},
}

@misc{skillevolean2026,
  title = {{SkillEvoLean: Mutation-enhanced skill evolution for Lean provers}},
  author = {Zhou, Kuo and Yang, Zixiong and Zhang, Lu},
  year = {2026},
  eprint = {2610.01799},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.01799},
  url = {https://arxiv.org/abs/2610.01799v2},
  urldate = {2026-10-10},
  note = {Survey source: arXiv:2610.01799v1},
}

@misc{evotrainer2026,
  title = {{EvoTrainer: Co-Evolving LLM Policies and Training Harnesses for Autonomous Agentic Reinforcement Learning}},
  author = {Chen, Guhong and Shi, Yingcheng and Li, Yongbin and Li, Binhua and Xu, Xander and Wei, Hu and Ni, Shiwen and Yang, Min and Ye, Jieping},
  year = {2026},
  eprint = {2606.03108},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2606.03108},
  url = {https://arxiv.org/abs/2606.03108v2},
  urldate = {2026-10-05},
  note = {arXiv preprint, version 2},
}

@article{agentevo2026,
  title = {{AgentEvo: cost-aware agentic workflow generation via adaptive multi-stage evolution}},
  author = {Xiao, Bo and Liu, Haiyang and Wang, Yitong and Hou, Yanning and Xu, Ke and Zhang, Xingyi},
  year = {2026},
  journal = {Complex \& Intelligent Systems},
  volume = {12},
  number = {9},
  pages = {208},
  doi = {10.1007/s40747-026-02325-0},
  url = {https://doi.org/10.1007/s40747-026-02325-0},
  urldate = {2026-10-05},
}

@misc{drevo2026,
  title = {{DREvo: Distilling Recalibrated Historical Experience for Harness Self-Evolution}},
  author = {Guo, Hanghui and Shi, Weijie and Chen, Zhangze and Xu, Shengxiang and Wang, Yishu and Zhang, Yimei and Ni, Wangze and Zhu, Jia and Di, Shimin},
  year = {2026},
  eprint = {2607.26722},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2607.26722},
  url = {https://arxiv.org/abs/2607.26722v2},
  urldate = {2026-10-10},
  note = {Survey source: arXiv:2607.26722v1},
  keywords = {domain-harness-evolution},
}

@misc{taskcoevolve2026,
  title = {{Task-CoEvolve: Efficient Harness Optimization via Adaptive Validation Task Selection}},
  author = {Miyai, Atsuyuki and Aizawa, Kiyoharu and Yamasaki, Toshihiko},
  year = {2026},
  eprint = {2608.20169},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.20169},
  url = {https://arxiv.org/abs/2608.20169v2},
  urldate = {2026-10-05},
  note = {arXiv preprint, version 2},
  keywords = {domain-harness-evolution},
}

@misc{hase2026,
  title = {{Harness-Aware Self-Evolving: Co-Evolving Model Weights, Harness, and Task Solutions}},
  author = {Luo, Haochen and Huang, Yi and Luo, Sichun and Liu, Fengyuan and Li, Lei and Hu, Zefa and Feng, Junlan and Liu, Qi},
  year = {2026},
  eprint = {2607.03935},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2607.03935},
  url = {https://arxiv.org/abs/2607.03935v1},
  urldate = {2026-10-05},
  note = {arXiv preprint, version 1},
  keywords = {domain-harness-evolution},
}

@misc{hcl2026,
  title = {{Harness Continual Learning: Continual Adaptation Beyond Model Parameters}},
  author = {Kang, Borui and Gu, Jinrui and Lv, Junhan and Li, Wenbin and Wang, Lei and Gao, Yang},
  year = {2026},
  eprint = {2608.19013},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.19013},
  url = {https://arxiv.org/abs/2608.19013v2},
  urldate = {2026-10-10},
  note = {Survey source: arXiv:2608.19013v1},
  keywords = {domain-harness-evolution},
}

@misc{zetta2026,
  title = {{Zetta $\zeta$: An Efficient Closed-Loop Embodied Harness for Self-Evolving Physical Intelligence}},
  author = {Ding, Xin and Mi, Liang and Huang, Mingzhe and Wang, Zixuan and Zhang, Chao and Hao, Zixu and Chen, Fu and Li, Xiangyu and Zheng, Yikai and Guo, Yaoyu and Wang, Weijun and Li, Kun and Wu, Hao and Liu, Yunxin and Cao, Ting},
  year = {2026},
  eprint = {2608.16590},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.16590},
  url = {https://arxiv.org/abs/2608.16590v1},
  urldate = {2026-10-05},
  note = {arXiv preprint, version 1},
  keywords = {domain-harness-evolution},
}

@misc{harnesspai2026,
  title = {{HarnessPAI: An Evolving Harness for Physical AI}},
  author = {Wang, Xin and Wu, Wenhao and Zhang, Menghao and Wang, Zhi and Shao, Kun and Luan, Jian and Li, Yang and Li, Qing and Gu, Shangding and Zhou, Huichi and Shi, Shuqing and Ni, Fei and Lu, Shuo and Meng, Weicheng and Li, Kang and Wu, Jin and Zhao, Kang and Guo, Shangmin and Li, Gen and Tang, Yongqiang and Zhang, Zhizhong and Xie, Yuan and Qu, Heng},
  year = {2026},
  eprint = {2609.29166},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.29166},
  url = {https://arxiv.org/abs/2609.29166v1},
  urldate = {2026-10-05},
  note = {arXiv preprint, version 1},
  keywords = {domain-harness-evolution},
}

@misc{shaper2026,
  title = {{Self-Evolving Embodied Agents via Skill-Harness Evolution}},
  author = {Wang, Peidong and Ma, Zhiming and Chang, Ying and Luo, Xufang and Zhang, Yiqun and Wang, Zihan and Yang, Xiaocui and Feng, Shi and Yang, Yuqing and Li, Dongsheng},
  year = {2026},
  eprint = {2608.11350},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.11350},
  url = {https://arxiv.org/abs/2608.11350v2},
  urldate = {2026-10-10},
  note = {Survey source: arXiv:2608.11350v1},
  keywords = {domain-harness-evolution},
}

@inproceedings{comfyclaw2026,
  title = {{COMFYCLAW: Self-Evolving Skill Harnesses for Image Generation Workflows}},
  author = {Li, Zongxia and Liu, Dawei and Liu, Fuxiao and Zhou, Yuhang and Wu, Xiyang and Chen, Jingxi and Xie, Jing and Wu, Xiaomin and Sun, Lichao},
  year = {2026},
  eprint = {2607.01709},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2607.01709},
  url = {https://arxiv.org/abs/2607.01709v1},
  urldate = {2026-10-10},
  note = {Survey source: arXiv:2607.01709v1; Workshop acceptance verified; final proceedings metadata not confirmed},
  keywords = {domain-harness-evolution},
  booktitle = {ECCV 2026 Workshop on Multimodal Large Language Models for Unified Comprehension and Generation},
}

@misc{autoref2026,
  title = {{AutoRef: Harness Optimization for Agentic Multi-Reference Image Generation}},
  author = {Oshima, Yuta and Onoda, Ku and Iwasawa, Yusuke and Suzuki, Masahiro and Matsuo, Yutaka and Furuta, Hiroki},
  year = {2026},
  eprint = {2609.35530},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.35530},
  url = {https://arxiv.org/abs/2609.35530v1},
  urldate = {2026-10-05},
  note = {arXiv preprint, version 1},
  keywords = {domain-harness-evolution},
}

@misc{videoweaver2026,
  title = {{VideoWeaver: Evaluating and Evolving Skills for Agentic Long Video Generation}},
  author = {Wei, Jianhui and Zhang, Yan and Tan, Jie and Zhu, Hengchuan and Zhang, Xiaotian and Chen, Ziyi and Zhang, Daoan and Xu, Wei and Jin, Yeying and Liu, Zuozhu},
  year = {2026},
  eprint = {2606.08091},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2606.08091},
  url = {https://arxiv.org/abs/2606.08091v2},
  urldate = {2026-10-05},
  note = {arXiv preprint, version 2},
  keywords = {domain-harness-evolution},
}

@misc{cineforge2026,
  title = {{CineForge: Self-Improving Agents for Long-Horizon Video Generation}},
  author = {Liu, Junxiang and Wang, Lin and Shi, Haiyu and Ma, Hongxu and Yang, Xiaoyu and Chen, Chunjie and Xu, Xiaoxiao and Zhan, Kaiqiao and Wang, Boao and Shi, Shuizhou and Zhu, Tianyun and Li, Jie and Li, Jiangtong},
  year = {2026},
  eprint = {2608.29621},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.29621},
  url = {https://arxiv.org/abs/2608.29621v1},
  urldate = {2026-10-05},
  note = {arXiv preprint, version 1},
  keywords = {domain-harness-evolution},
}

@misc{vidharness2026,
  title = {{VidHarness: Evolving Agent Harnesses for Cost-Efficient Long Video Understanding}},
  author = {Liang, Susan and Wu, Jianmin and Dong, Daxiang},
  year = {2026},
  eprint = {2609.38413},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.38413},
  url = {https://arxiv.org/abs/2609.38413v1},
  urldate = {2026-10-05},
  note = {arXiv preprint, version 1},
  keywords = {domain-harness-evolution},
}

@misc{videoevolve2026,
  title = {{VideoEvolve: Evolving Agent Harnesses for Video Temporal Grounding}},
  author = {Luo, Bingjun and Fan, Yuhuan and Guo, Jialin and Li, Siqi},
  year = {2026},
  eprint = {2610.01766},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.01766},
  url = {https://arxiv.org/abs/2610.01766v1},
  urldate = {2026-10-05},
  note = {arXiv preprint, version 1},
  keywords = {domain-harness-evolution},
}

@misc{proverinterface2026,
  title = {{Growing an Agent/Prover Interface: Evolutionary Tool Design for Cost-Efficient Theorem Proving in Rocq and Lean}},
  author = {Viennot, Jules and Baudart, Guillaume and Lelarge, Marc},
  year = {2026},
  eprint = {2609.39544},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.39544},
  url = {https://arxiv.org/abs/2609.39544v3},
  urldate = {2026-10-10},
  note = {Survey source: arXiv:2609.39544v1},
  keywords = {domain-harness-evolution, section-4},
}

@misc{proofevolve2026,
  title = {{ProofEvolve: Neuro-Symbolic Evolution for Formal Automated Theorem Proving}},
  author = {Ye, Wenqian and Guan, Ziwei and Xie, Eric and Liu, Bohan and Modi, Shivani and Zhang, Buyun and Wen, Ellie Dingqiao and Kautz, Henry and Zhang, Aidong},
  year = {2026},
  eprint = {2608.26334},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.26334},
  url = {https://arxiv.org/abs/2608.26334v1},
  urldate = {2026-10-05},
  note = {arXiv preprint, version 1},
  keywords = {domain-harness-evolution, section-4},
}

@misc{disco2026,
  title = {{Repo-To-Skill: Distilling GitHub Repositories Into AI4AI Skills}},
  author = {Chen, Jianlyu and Hu, Yuyang and Qian, Hongjin and Liu, Jiawei and Wei, Wenqing and Chen, Xiaolong and Lian, Defu and Dou, Zhicheng and Li, Chaozhuo and Ye, Qiwei and Liu, Zheng},
  year = {2026},
  eprint = {2609.02749},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.02749},
  url = {https://arxiv.org/abs/2609.02749v1},
  urldate = {2026-10-05},
  note = {arXiv preprint, version 1},
  keywords = {domain-harness-evolution, section-4},
}

@misc{comprehensive_selfevolving_survey2025,
  title = {{A Comprehensive Survey of Self-Evolving AI Agents: A New Paradigm Bridging Foundation Models and Lifelong Agentic Systems}},
  author = {Fang, Jinyuan and Peng, Yanwen and Zhang, Xi and Wang, Yingxu and Yi, Xinhao and Zhang, Guibin and Xu, Yi and Wu, Bin and Liu, Siwei and Li, Zihao and Ren, Zhaochun and Aletras, Nikos and Wang, Xi and Zhou, Han and Meng, Zaiqiao},
  year = {2025},
  eprint = {2508.07407},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2508.07407},
  url = {https://arxiv.org/abs/2508.07407v2},
  note = {arXiv preprint, v2},
  urldate = {2026-10-10},
  keywords = {prior-survey, screened-2026-10-10},
}

@misc{ai4ai_survey2026,
  title = {{AI4AI Survey: From Long-Horizon Agents to Recursive Self-Improvement---Definitions, Reliable Horizons, and Open Problems}},
  author = {Wu, Kai and Lyu, Hao and Luo, Zhen and Wang, Chaofan and Ye, Siyu and Lin, Jinghao and Ji, Xiaozhong and Jiang, Boyuan and Wang, Shengzhi and Wang, Zihan and Ye, Yiwen and Wang, Hao and Wang, Zimu and Liu, Wenzhe and Wang, Ruobing and Cai, Kai and Xiong, Mingliang and Fang, Wen and Liu, Mingqing and Zhang, Yifan and Yang, Lei and Hu, Xiaobin and Liu, Qingwen},
  year = {2026},
  url = {https://www.preprints.org/manuscript/202608.2108/v1},
  doi = {10.20944/preprints202608.2108.v1},
  note = {Preprints.org preprint, version 1; not peer-reviewed},
  urldate = {2026-10-10},
  keywords = {prior-survey, screened-2026-10-10},
}

@misc{reliable_selfevolving_survey2026,
  title = {{Diving into Reliable Self-Evolving Agents: A Survey}},
  author = {Wang, Kaiqi and Hou, Wenjin and Yan, Yuchen and Jia, Hongrui and Zhong, Zhisheng and Ren, Botao and Chen, Yifei and Zhang, Songyang and Shen, Yongliang and Xiao, Jun and Yang, Yi and Zhuang, Yueting and Fan, Hehe},
  year = {2026},
  url = {https://www.preprints.org/manuscript/202609.0913/v1},
  note = {Preprints.org preprint, version 1; not peer-reviewed},
  urldate = {2026-10-10},
  keywords = {prior-survey, screened-2026-10-10},
}

@misc{systematic_coevolution_survey2026,
  title = {{A Systematic Survey of Self-Evolving Agents: From Model-Centric to Environment-Driven Co-Evolution}},
  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 Yang, Chang and Wang, Xinrun and Huang, Xiao and Zhang, Qinggang and Su, Jinsong},
  year = {2026},
  url = {https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6626878},
  doi = {10.2139/ssrn.6626878},
  note = {SSRN preprint; author repository also links a TechRxiv version},
  urldate = {2026-10-10},
  keywords = {prior-survey, screened-2026-10-10},
}

@misc{graph_selfevolving_survey2026,
  title = {{Self-Evolving Agents as Dynamic Graph Transformation: A Survey and New Perspective}},
  author = {Xu, Yuanyuan and Zhang, Wenjie and Chen, Yin and Lin, Xuemin and Zhang, Ying},
  year = {2026},
  eprint = {2608.18104},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.18104},
  url = {https://arxiv.org/abs/2608.18104v1},
  note = {arXiv preprint, v1},
  urldate = {2026-10-10},
  keywords = {prior-survey, screened-2026-10-10},
}

@misc{safety_selfevolution_survey2026,
  title = {{Safety in Self-Evolving Agents: A Survey}},
  author = {Chen, Jiahao and Feng, Zhou and Ma, Oubo and Yan, Yichen and Lin, Ruixiao and Zhang, Hangtao and Du, Linkang and An, Hengyu and Yang, Yong and Liu, Jun and Li, Junhao and Xu, Naen and Zhou, Chunyi and Su, Yuan and Jin, Zehao and Ma, Qianli and Qi, Leyi and Wang, Yiming and Ma, Zhe and Pu, Yuwen and Du, Mengyao and Song, Yuanyi and Huang, Enhao and Fu, Zhihui and Wang, Jun and Li, Jinfeng and Chen, Yuefeng and Xue, Hui and Li, Yiming and Du, Tianyu and Ji, Shouling},
  year = {2026},
  eprint = {2610.00093},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.00093},
  url = {https://arxiv.org/abs/2610.00093v1},
  note = {arXiv preprint, v1},
  urldate = {2026-10-10},
  keywords = {prior-survey, screened-2026-10-10},
}

@misc{humanoid_selfevolution_survey2026,
  title = {{Self-Evolving AI for Humanoids: Mechanisms, Safety, and Evaluation of Post-Deployment Self-Improvement}},
  author = {Nguyen, Loc X. and Raha, Avi Deb and Le, Huy Q. and Huh, Eui-Nam and Niyato, Dusit and Hong, Choong Seon},
  year = {2026},
  eprint = {2609.13236},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2609.13236},
  url = {https://arxiv.org/abs/2609.13236v1},
  note = {arXiv preprint, v1},
  urldate = {2026-10-10},
  keywords = {prior-survey, screened-2026-10-10},
}

@misc{ec_trustworthy_survey2026,
  title = {{Evolutionary Computation for Trustworthy AI: From Attacks and Defenses to Self-Evolving Era}},
  author = {Dong, Junhao and Wang, Chenkai and Lin, Xuanhui and Gong, Mingrong and Wang, Siyu and Wen, Yuqing and Liu, Jiao and Huang, Catherine and Yen, Gary G. and Yao, Xin and Ong, Yew-Soon},
  year = {2026},
  eprint = {2610.02996},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.02996},
  url = {https://arxiv.org/abs/2610.02996v1},
  note = {arXiv preprint, v1},
  urldate = {2026-10-10},
  keywords = {prior-survey, screened-2026-10-10},
}

@misc{mesh_harness2026,
  title = {{MESH-Harness: Self-Improving Agent Harnesses via Bandit-Guided Compositional Evolution}},
  author = {Shang, Zhiwei and Huo, Yu and Gong, Mingrong and Yan, An and Qu, Zikun and Dong, Junhao and Low, Bryan Kian Hsiang and Wu, Chenglin and Dai, Zhongxiang},
  year = {2026},
  eprint = {2610.05300},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.05300},
  url = {https://arxiv.org/abs/2610.05300v1},
  note = {arXiv preprint, v1},
  urldate = {2026-10-10},
  keywords = {harness-evolution-candidate, screened-2026-10-10},
}

@misc{harnesssr2026,
  title = {{Learning to Design SR Solvers: Harness Optimization for Symbolic Regression}},
  author = {Li, Qian and Hu, Yuxiao and Duan, Qihao and Yan, Jinyue and Chen, Yuntian},
  year = {2026},
  eprint = {2610.16966},
  archivePrefix = {arXiv},
  doi = {10.48550/arXiv.2610.16966},
  url = {https://arxiv.org/abs/2610.16966v1},
  urldate = {2026-10-18},
  note = {arXiv preprint, version 1; method: Sections 3.1-3.5; Appendix G.2}
}

@misc{harnessttt2026,
  title = {{HarnessTTT: Learn to Optimize Harness at Test-Time for Scientific Discovery}},
  author = {Ma, Yingzi and Cao, Yulong and Li, Nanxi and Wu, Fang and Zhang, Yizhe and Zou, James and Choi, Yejin and Xiao, Chaowei},
  year = {2026},
  eprint = {2610.17370},
  archivePrefix = {arXiv},
  doi = {10.48550/arXiv.2610.17370},
  url = {https://arxiv.org/abs/2610.17370v1},
  urldate = {2026-10-18},
  note = {arXiv preprint, version 1; method: Section 3, Sections 3.1-3.2 and Appendix C}
}

@misc{harnessnet2026,
  title = {{HarnessNet: Feedback-driven Hierarchical Joint Harness Evolution for Multi-agent Systems}},
  author = {Qiu, Zhaoyuan and Seneviratne, Sachith and Xia, Yu and Halgamuge, Saman},
  year = {2026},
  eprint = {2610.15221},
  archivePrefix = {arXiv},
  doi = {10.48550/arXiv.2610.15221},
  url = {https://arxiv.org/abs/2610.15221v1},
  urldate = {2026-10-18},
  note = {arXiv preprint, version 1; method: Sections 3.1-3.3, Algorithms 1-4}
}

@misc{harness_squared2026,
  title = {{Harness^2: Recursive Agent Harnessing for an Open World}},
  author = {Xu, Ruiyao and Chen, Yanfei and CuiZhu, Zhongying and Mishra, Bhavana Dalvi and Ming, Yifei and Yu, Han and Han, Rujun and Lee, Chen-Yu and Pfister, Tomas},
  year = {2026},
  eprint = {2610.14224},
  archivePrefix = {arXiv},
  doi = {10.48550/arXiv.2610.14224},
  url = {https://arxiv.org/abs/2610.14224v1},
  urldate = {2026-10-18},
  note = {arXiv preprint, version 1; method: Sections 2.1-2.3; Appendix D.3}
}

@misc{workflowops2026,
  title = {{WorkflowOps: Learning Agent Collaboration Priors for Multi-Agent Workflow Orchestration}},
  author = {Cheng, Qi and Chen, Shengyu and Cheng, Wei and Chen, Zhengzhang and Jia, Xiaowei and Wang, Haoyu and Chen, Haifeng},
  year = {2026},
  eprint = {2610.07860},
  archivePrefix = {arXiv},
  doi = {10.48550/arXiv.2610.07860},
  url = {https://arxiv.org/abs/2610.07860v1},
  urldate = {2026-10-18},
  note = {arXiv preprint, version 1; method: Sections 3-4, especially Sections 4.2-4.3}
}

@inproceedings{cborchestrator2026,
  title = {{CB-Orchestrator: Adaptive Workflow Optimization for LLM Agents via Contextual Bandits}},
  author = {Sun, Jiahang and Shang, Zhiwei and Liu, Zhipiao and Yang, Hongwei and Xie, Guoqing and Qiu, Shuang and Dai, Zhongxiang},
  year = {2026},
  eprint = {2610.13064},
  archivePrefix = {arXiv},
  doi = {10.48550/arXiv.2610.13064},
  url = {https://arxiv.org/abs/2610.13064v1},
  urldate = {2026-10-18},
  note = {Method source: arXiv:2610.13064v1; Sections 3.1-3.4 and Appendix A.11; workshop version: \url{https://daizhongxiang.github.io/papers/90_CB_Orchestrator_Adaptive_Wo.pdf}; exact first public date unresolved},
  booktitle = {ICML 2026 Workshop on Compositional Learning: Safety, Interpretability, and Agents}
}

@misc{lawa2026,
  title = {{Optimize Once, Execute Fast: Latency-Aware Multi-Agent Workflow Learning for Recurrent Queries}},
  author = {Zhang, Enpei and Qu, Feiyu and Huang, Zheng and Zhou, Dawei and Chen, Elynn and Yan, Yujun},
  year = {2026},
  eprint = {2610.14481},
  archivePrefix = {arXiv},
  doi = {10.48550/arXiv.2610.14481},
  url = {https://arxiv.org/abs/2610.14481v1},
  urldate = {2026-10-18},
  note = {Method source: arXiv:2610.14481v1; Sections 3.1-3.5; Algorithm 1}
}

@inproceedings{dago2026,
  title = {{Fusion is the New Mutation: Bandit-Guided Evolution on Workflow Graphs}},
  author = {Shang, Zhiwei and Sun, Jiahang and Gong, Mingrong and Kong, Mingze and Qu, Zikun and Lu, Pingchen and Dong, Junhao and Liu, Zhipiao and Yang, Hongwei and Xie, Guoqing and Shu, Yao and Dai, Zhongxiang},
  year = {2026},
  eprint = {2610.05284},
  archivePrefix = {arXiv},
  doi = {10.48550/arXiv.2610.05284},
  url = {https://arxiv.org/abs/2610.05284v1},
  urldate = {2026-10-18},
  note = {Method source: arXiv:2610.05284v1; Sections 3-4, especially Fusion, evaluation, and feedback update; workshop version: \url{https://daizhongxiang.github.io/papers/54_Fusion_is_the_New_Mutation_.pdf}; exact first public date unresolved},
  booktitle = {ICML 2026 Workshop on Compositional Learning: Safety, Interpretability, and Agents}
}

@misc{inflowop2026,
  title = {{Pay for the Fault, Not the Flow: Label-Free In-Flow Multi-Agent Workflow Optimization}},
  author = {Guo, Xuehang and Wang, Haoyu and Chen, Shengyu and Chen, Zach and Cheng, Wei and Wang, Qingyun and Chen, Haifeng},
  year = {2026},
  eprint = {2610.01017},
  archivePrefix = {arXiv},
  doi = {10.48550/arXiv.2610.01017},
  url = {https://arxiv.org/abs/2610.01017v2},
  urldate = {2026-10-18},
  note = {Method source: arXiv:2610.01017v2; Sections 3.1-3.2}
}

@misc{gso2026,
  title = {{Do Self-Evolving Skills Generalize to Held-Out Tasks?}},
  author = {Piao, Xihao and Wang, Zifeng and Chen, Zhen},
  year = {2026},
  eprint = {2609.39148},
  archivePrefix = {arXiv},
  doi = {10.48550/arXiv.2609.39148},
  url = {https://arxiv.org/abs/2609.39148v2},
  urldate = {2026-10-18},
  note = {Method source: arXiv:2609.39148v2; Sections 3.1-3.3 and 4.1-4.2}
}

@misc{skillv2026,
  title = {{Skill-V: Verifiable Self-Evolving Skill Library for Interactive Agents}},
  author = {Ma, Jie and Qian, Zhipeng and Ma, Yufei and Liang, Zihan and Ji, Jiayi and Cai, Qingpeng and Chen, Ben and Jiang, Peng and Sun, Xiaoshuai},
  year = {2026},
  eprint = {2610.11781},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.11781},
  url = {https://arxiv.org/abs/2610.11781v1},
  note = {arXiv preprint, v1},
  urldate = {2026-10-17},
  keywords = {category-model-and-policy-training-harness-co-evolution-and-distillation, screened-2026-10-17},
}

@misc{evoalloc2026,
  title = {{EvoAlloc: A Self-Evolving Resource Allocation Agent for Efficient Program Evolution}},
  author = {Dai, Yanning and Wang, Yuhui and Li, Nanbo and Wang, Wenyi and Schmidhuber, Jürgen},
  year = {2026},
  eprint = {2610.12086},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.12086},
  url = {https://arxiv.org/abs/2610.12086v1},
  note = {arXiv preprint, v1},
  urldate = {2026-10-17},
  keywords = {category-self-modification-and-meta-level-improvement, screened-2026-10-17},
}

@misc{sepm2026,
  title = {{SEPM: Self-Evolving Procedural Memory Grounded in Task Graphs for Multi-Agent Collaboration}},
  author = {Liu, Yi and Zhao, Junwei and Sun, Li and Zhang, Jiansong and Su, Sen},
  year = {2026},
  eprint = {2610.14515},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.14515},
  url = {https://arxiv.org/abs/2610.14515v1},
  note = {arXiv preprint, v1},
  urldate = {2026-10-17},
  keywords = {category-memory-context-skills-and-experience-reuse, screened-2026-10-17},
}

@misc{harmonyharness2026,
  title = {{Harmony Harness: Multi-Objective Self-Evolution of Agent Harnesses}},
  author = {Yu, Weichen and Miculicich, Lesly and Mishra, Bhavana Dalvi and Chen, Jiefeng and Li, Shuo and Fredrikson, Matt and Pfister, Tomas and Le, Long T.},
  year = {2026},
  eprint = {2610.15150},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.15150},
  url = {https://arxiv.org/abs/2610.15150v1},
  note = {arXiv preprint, v1},
  urldate = {2026-10-17},
  keywords = {category-harness-optimization-and-search-workflows, screened-2026-10-17},
}

@misc{videoevoharness2026,
  title = {{VideoEvoHarness: Self Evolving Agent Harness from Unlabeled Videos for Video Understanding}},
  author = {Xu, Weixin and Liu, Yifan and Wang, Yi and Rong, Mingfei and Yang, Zhenyu and Zhong, Zhisheng and Qian, Shengsheng and Xu, Changsheng},
  year = {2026},
  eprint = {2610.15389},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.15389},
  url = {https://arxiv.org/abs/2610.15389v1},
  note = {arXiv preprint, v1},
  urldate = {2026-10-17},
  keywords = {category-additional-task-domain-harness-evolution, screened-2026-10-17},
}

@misc{videoharness_srsi2026,
  title = {{VideoHarness-SRSI: Structured Recursive Self-Improvement for Video Understanding Harness}},
  author = {Lin, Zihao and Dong, Haoyu and Xu, Haobo and Chang, Chi-Chih and Ning, Xuying and Xia, Yinglong and Yang, Shuang and Zhao, Zhuokai and Song, Xinying and Liu, Dongfang and Huang, Lifu and Song, Yang and Wang, Qifan},
  year = {2026},
  eprint = {2610.19424},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.19424},
  url = {https://arxiv.org/abs/2610.19424v1},
  note = {arXiv preprint, v1},
  urldate = {2026-10-17},
  keywords = {category-additional-task-domain-harness-evolution, screened-2026-10-17},
}

@misc{scientific_release_gating2026,
  title = {{Structure Before Intelligence: Gating What Agentic Scientific Workflows Release}},
  author = {Acharya, Ashish and Amin, Aman Al and Qin, Hong},
  year = {2026},
  eprint = {2610.19507},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.19507},
  url = {https://arxiv.org/abs/2610.19507v1},
  note = {arXiv preprint, v1},
  urldate = {2026-10-17},
  keywords = {category-generalization-evaluation-methodology-and-safety, screened-2026-10-17},
}

@misc{l3m2026,
  title = {{L3M: A Verifiable Harness and Runtime for Self-Extending and Self-Improving Agents}},
  author = {Eisenach, Carson and George, Robert Joseph and Quenneville-Belair, Vincent and Gwimm, Kate and Foster, Dean},
  year = {2026},
  eprint = {2610.19591},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.19591},
  url = {https://arxiv.org/abs/2610.19591v1},
  note = {arXiv preprint, v1},
  urldate = {2026-10-17},
  keywords = {category-agent-architectures-and-harness-infrastructure, screened-2026-10-17},
}

@misc{evolvingbrain2026,
  title = {{EvolvingBrain: An Adaptive Skill System for Multi-modal Agents Self-Evolution in Open Worlds}},
  author = {Liu, Runxuan and Shi, Jingqi and Wang, Baoxin and Zhao, Honghong and Xu, Haoyuan and Sun, Yueqing and Liang, Jiafeng and He, Tao and Wu, Dayong and Gu, Qi and Wang, Shijin and Liu, Ming and Qin, Bing},
  year = {2026},
  eprint = {2610.20786},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.20786},
  url = {https://arxiv.org/abs/2610.20786v1},
  note = {arXiv preprint, v1},
  urldate = {2026-10-17},
  keywords = {category-memory-context-skills-and-experience-reuse, screened-2026-10-17},
}

@misc{oraclefree_dsl2026,
  title = {{A Self-Improving Agent that Grows its Own DSL: Oracle-Free Operator Induction for Exact Question Answering}},
  author = {Chuang, Yun-Shiuan and Ge, Yirou and Li, Yixi and Bian, Yiming and Pan, Zijie and Ballard, Jenna L. and Kona, Uma and Pang, Linsey and Mehrotra, Prakhar},
  year = {2026},
  eprint = {2610.20809},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.20809},
  url = {https://arxiv.org/abs/2610.20809v1},
  note = {arXiv preprint, v1},
  urldate = {2026-10-17},
  keywords = {category-memory-context-skills-and-experience-reuse, screened-2026-10-17},
}

@misc{memscope2026,
  title = {{MemScope: Evolving Agent Memory Architectures through Localized Code Repair}},
  author = {Su, Miao and Guo, Yucan and Li, Zixuan and Bai, Long and Hou, Zhongni and Jin, Xiaolong and Guo, Jiafeng},
  year = {2026},
  eprint = {2610.21170},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.21170},
  url = {https://arxiv.org/abs/2610.21170v1},
  note = {arXiv preprint, v1},
  urldate = {2026-10-17},
  keywords = {category-memory-context-skills-and-experience-reuse, screened-2026-10-17},
}

@misc{rsiflow2026,
  title = {{RSIFlow: Adaptive Self-Improvement Workflows with an Evolving Meta-Agent}},
  author = {Zhang, Huan and Chen, Mingju and Zhou, Dongxu and Lv, Can and Cui, Sen and Wu, Faguo and Zhou, Shiji},
  year = {2026},
  eprint = {2610.21791},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.21791},
  url = {https://arxiv.org/abs/2610.21791v1},
  note = {arXiv preprint, v1},
  urldate = {2026-10-17},
  keywords = {category-self-modification-and-meta-level-improvement, screened-2026-10-17},
}

@misc{metashare2026,
  title = {{MetaShare: Failure-Aware Collective Self-Improvement for Open-Ended Agents}},
  author = {Bhutani, Vishal and Gupta, Prasang and Chundru, Sumanth and Paul, Kevin},
  year = {2026},
  eprint = {2610.21887},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.21887},
  url = {https://arxiv.org/abs/2610.21887v1},
  note = {arXiv preprint, v1},
  urldate = {2026-10-17},
  keywords = {category-harness-optimization-and-search-workflows, screened-2026-10-17},
}

@misc{cig2026,
  title = {{Causal Improvement Graph for Agentic Harness Optimization}},
  author = {Zhang, Junjie and Liu, Shunyu and Wang, Haoyu and Lin, Ting-En and Li, Yongbin and Tao, Dacheng},
  year = {2026},
  eprint = {2610.05039},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.05039},
  url = {https://arxiv.org/abs/2610.05039v1},
  note = {arXiv preprint, v1},
  urldate = {2026-10-10},
  keywords = {harness-evolution-candidate, screened-2026-10-10},
}

@misc{sorsi2026,
  title = {{Second-Order Problem Solving for Recursive Self-Improvement in Formal Verification}},
  author = {Jiang, Yuxuan and Vempaty, Aditya and Jagmohan, Ashish},
  year = {2026},
  eprint = {2610.05701},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.05701},
  url = {https://arxiv.org/abs/2610.05701v2},
  note = {arXiv preprint, v2},
  urldate = {2026-10-10},
  keywords = {harness-evolution-candidate, screened-2026-10-10},
}

@misc{verse2026,
  title = {{VERSE: Verified Self-Evolving Optimizer for Agent Harnesses}},
  author = {Wang, Zekai and Ge, Yingqiang and Wang, Zekun and Wang, Hai and Xu, Yuhui and Frandsen, Joshua and Fu, Shancong and Wilson, Ashia C. and Reddy, Chandan K.},
  year = {2026},
  eprint = {2610.02616},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.02616},
  url = {https://arxiv.org/abs/2610.02616v1},
  note = {arXiv preprint, v1},
  urldate = {2026-10-10},
  keywords = {harness-evolution-candidate, screened-2026-10-10},
}

@misc{mass_rsi2026,
  title = {{Recursive Self-Improvement through Multi-Agent Self-Supervision}},
  author = {Lee, Hyunin and Xu, Jinglue and Seely, Jeffrey and Lee, Donghyun and Sojoudi, Somayeh and Zaharia, Matei and Tang, Yujin},
  year = {2026},
  eprint = {2610.12176},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.12176},
  url = {https://arxiv.org/abs/2610.12176v1},
  note = {arXiv preprint, v1},
  urldate = {2026-10-10},
  keywords = {harness-evolution-candidate, screened-2026-10-10},
}

@misc{emho2026,
  title = {{EMHO: EMbodied Agent Harness Optimization via Experience Traces}},
  author = {Lee, Hyun Jung and Kim, Jungtaek and Jeong, Jongwon and Kam, Tae-Eui and Kim, Donghyun and Lee, Yong Jae},
  year = {2026},
  eprint = {2610.08432},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.08432},
  url = {https://arxiv.org/abs/2610.08432v1},
  note = {arXiv preprint, v1},
  urldate = {2026-10-10},
  keywords = {harness-evolution-candidate, screened-2026-10-10},
}

@misc{worldabduct2026,
  title = {{Safe, Persistent, and Evolving Agent Harness for Understanding Partially Observable Worlds}},
  author = {Gao, Yisen and Guo, Yue and Zong, Qing and Guo, Yiwen and Song, Yangqiu},
  year = {2026},
  eprint = {2610.11552},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.11552},
  url = {https://arxiv.org/abs/2610.11552v1},
  note = {arXiv preprint, v1},
  urldate = {2026-10-10},
  keywords = {harness-evolution-candidate, screened-2026-10-10},
}
@inproceedings{muse2026,
  title = {{Towards Self-Evolving Agents: Enabling Autonomy through Interactive Experience Refinement}},
  author = {Yang, Cheng and Yang, Xuemeng and Wen, Licheng and Fu, Daocheng and Mei, Jianbiao and Wu, Rong and Cai, Pinlong and Shen, Yufan and Deng, Nianchen and Xu, Jia and Shi, Botian and Qiao, Yu and Li, Haifeng},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  month = {jul},
  pages = {30424--30451},
  publisher = {Association for Computational Linguistics},
  eprint = {2510.08002},
  archiveprefix = {arXiv},
  doi = {10.18653/v1/2026.findings-acl.1522},
  url = {https://aclanthology.org/2026.findings-acl.1522/},
  note = {MUSE; earlier preprint titled Learning on the Job: An Experience-Driven Self-Evolving Agent for Long-Horizon Tasks, arXiv v1; final author list follows the published PDF},
  urldate = {2026-10-19},
  keywords = {category-memory-context-skills-and-experience-reuse, screened-2026-10-19},
}

@inproceedings{contda2026,
  title = {{Beyond Static Toolsets: Self-Evolving LLM Tool Agents via Continual Documentation Adaptation}},
  author = {Wu, Bin and Meij, Edgar and Yilmaz, Emine},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  month = {jul},
  pages = {21519--21539},
  publisher = {Association for Computational Linguistics},
  doi = {10.18653/v1/2026.findings-acl.1082},
  url = {https://aclanthology.org/2026.findings-acl.1082/},
  note = {ContDa; published proceedings version inspected; no earlier preprint verified},
  urldate = {2026-10-19},
  keywords = {category-harness-optimization-and-search-workflows, screened-2026-10-19},
}

@inproceedings{agentfactory_subagents2026,
  title = {{AgentFactory: A Self-Evolving Framework Through Executable Subagent Accumulation and Reuse}},
  author = {Zhang, Zhang and Lu, Shuqi and Qian, Hongjin and He, Di and Liu, Zheng},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 3: System Demonstrations)},
  year = {2026},
  month = {jul},
  pages = {819--828},
  publisher = {Association for Computational Linguistics},
  eprint = {2603.18000},
  archiveprefix = {arXiv},
  doi = {10.18653/v1/2026.acl-demo.81},
  url = {https://aclanthology.org/2026.acl-demo.81/},
  note = {Earlier arXiv v1 and published demonstration paper are one study; distinct from AgentFactory: Towards Automated Agentic System Design and Optimization by Enci Zhang et al.},
  urldate = {2026-10-19},
  keywords = {category-memory-context-skills-and-experience-reuse, screened-2026-10-19},
}

@misc{failure_feedback_directedness2026,
  title = {{Does Failure Feedback Direct Agent Self-Modification?}},
  author = {He, Rong},
  year = {2026},
  eprint = {2610.14662},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.14662},
  url = {https://arxiv.org/abs/2610.14662v1},
  note = {arXiv preprint, v1; no formal venue verified},
  urldate = {2026-10-19},
  keywords = {category-generalization-evaluation-methodology-and-safety, screened-2026-10-19},
}

@misc{rsi_two_perspectives2026,
  title = {{From Existing Methods to New Research Ideas: Two Perspectives on Recursive Self-Improvement}},
  author = {Zhou, Leijun and Yu, Yanke and Liu, Zhihao and Xiang, Qu and Hu, Tiancheng and Dong, Yiru and Wu, Yihan and Xiang, Yangxi and Wu, Xuejun and Chen, Xi and Qian, Buyue and Zheng, Yaowei and Hu, Junhao},
  year = {2026},
  eprint = {2610.14625},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.14625},
  url = {https://arxiv.org/abs/2610.14625v1},
  note = {arXiv preprint, v1; survey and conceptual background, not a new harness intervention},
  urldate = {2026-10-19},
  keywords = {category-surveys-definitions-and-conceptual-frameworks, screened-2026-10-19},
}

@misc{evoharness_rl2026,
  title = {{EvoHarness-RL: Learning Runtime Harness Coordination for Self-Evolving Agents}},
  author = {Ning, Xuying and Fu, Dongqi and Wei, Tianxin and Bei, Yuanchen and Yang, Xiyuan and Xu, Wujiang and Song, Yueqi and Li, Bingxuan and Li, Zihao and Zeng, Hanqing and Shen, Xiang and Wang, Yajuan and Wu, Yifan and Wang, Qifan and Liu, Jiayi and Li, Hong and Xia, Yinglong and Fan, Xiangjun and Tong, Hanghang and He, Jingrui},
  year = {2026},
  eprint = {2608.05446},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.05446},
  url = {https://arxiv.org/abs/2608.05446v2},
  note = {arXiv v2, revised 2026-10-05; author-reported acceptance to LLA at COLM 2026 workshop; earlier v1 title Learning Self-Evolving Runtime Harness for Long-Horizon LLM Agents},
  urldate = {2026-10-19},
  keywords = {category-model-and-policy-training-harness-co-evolution-and-distillation, screened-2026-10-19},
}

@misc{memskill2026,
  title = {{MemSkill: Learning and Evolving Memory Skills for Self-Evolving Agents}},
  author = {Zhang, Haozhen and Long, Quanyu and Bao, Jianzhu and Feng, Tao and Zhang, Weizhi and Yue, Haodong and Wang, Wenya},
  year = {2026},
  eprint = {2602.02474},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2602.02474},
  url = {https://arxiv.org/abs/2602.02474v2},
  note = {arXiv v2, revised 2026-05-24; NeurIPS 2026 acceptance is author-reported in the official repository, not independently verified against proceedings},
  urldate = {2026-10-20},
  keywords = {category-memory-context-skills-and-experience-reuse, screened-2026-10-20},
}

@misc{erskill2026,
  title = {{ERSkill: Evolving for Skill-Guided Adaptive Memory Retrieval}},
  author = {Chen, Haolong and Zhang, Liang and Li, Zhuo and Xue, Lei and Zhu, Guangxu},
  year = {2026},
  eprint = {2608.12720},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2608.12720},
  url = {https://arxiv.org/abs/2608.12720v2},
  note = {arXiv v2, revised 2026-09-28; final author spelling Guangxu Zhu follows the primary full text rather than abstract metadata Guanrxu Zhu; no formal venue verified},
  urldate = {2026-10-20},
  keywords = {category-memory-context-skills-and-experience-reuse, screened-2026-10-20},
}

@misc{sesa2026,
  title = {{Self-Play Meets Skill Evolution: Self-Evolving Search Agents that Pose, Solve, and Remember}},
  author = {Fu, Zenghuang and Li, Zhaoyang and Ai, Qiuyuan and Wu, Haoyu and Wu, Minghui and Zhao, Chenxu and Wang, Ante and He, Guannan and Wang, Changwei},
  year = {2026},
  eprint = {2607.29468},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2607.29468},
  url = {https://arxiv.org/abs/2607.29468v1},
  note = {arXiv v1; no later arXiv version or formal venue verified},
  urldate = {2026-10-20},
  keywords = {category-model-and-policy-training-harness-co-evolution-and-distillation, screened-2026-10-20},
}

@misc{harness_ceiling2026,
  title = {{Harness Evolution Hits a Ceiling: When Weight Training Should Begin}},
  author = {Tian, Yuan and Hu, Bing and Wang, Hao and Lu, Binghang and Wu, Fang},
  year = {2026},
  eprint = {2610.11655},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.11655},
  url = {https://arxiv.org/abs/2610.11655v1},
  note = {arXiv v1; no later arXiv version or formal venue verified},
  urldate = {2026-10-20},
  keywords = {category-model-and-policy-training-harness-co-evolution-and-distillation, screened-2026-10-20},
}

@misc{self_evolving_agent_harness2026,
  title = {{Self-Evolving Agent Harness}},
  author = {Wei, Chuyang and Guan, Haoxiang and Zhu, Minghang and Zhou, Kexin and He, Jiyan and Zhou, Xiren and Chen, Huanhuan and Shen, Yifei},
  year = {2026},
  eprint = {2610.18813},
  archiveprefix = {arXiv},
  doi = {10.48550/arXiv.2610.18813},
  url = {https://arxiv.org/abs/2610.18813v1},
  note = {arXiv v1; conceptual framework and related-work synthesis, not a new benchmark intervention; no formal venue verified},
  urldate = {2026-10-20},
  keywords = {category-surveys-definitions-and-conceptual-frameworks, screened-2026-10-20},
}