% BibTeX for "Harness Engineering at ICML 2026" — a 45-paper survey.
% https://saikrishnarallabandi.github.io/judith/posts/harness-engineering-icml2026.html
% Author lists are partial where the source truncated them ("and others");
% arXiv/adjacent venues are labeled. Links verified against paper titles.

@misc{rallabandi2026harness,
  author       = {Rallabandi, Sai Krishna},
  title        = {The Harness Is the Point: Harness Engineering at {ICML} 2026},
  year         = {2026},
  howpublished = {\url{https://saikrishnarallabandi.github.io/judith/posts/harness-engineering-icml2026.html}},
  note         = {Field notes, On Rallabandi Radar}
}

@inproceedings{autotool2026,
  author       = {Jiaru Zou and Ling Yang and Mengdi Wang and others},
  title        = {{AutoTool: Dynamic Tool Selection and Integration for Agentic Reasoning}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/65574}
}

@inproceedings{gecko2026,
  author       = {Zeyu Zhang and Guohao Li and Zhenchang Xing and Liang Zheng and others},
  title        = {{Gecko: A Simulation Environment with Stateful Feedback for Refining Agent Tool Calls}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/65078}
}

@inproceedings{masorchestra2026,
  author       = {Zixuan Ke and Yifei Ming and Austin Xu and Shafiq Joty and others},
  title        = {{MAS-Orchestra: Understanding and Improving Multi-Agent Reasoning Through Holistic Orchestration and Controlled Benchmarks}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/66445}
}

@inproceedings{opensage2026,
  author       = {Hongwei Li and Zhun Wang and Dawn Song and others},
  title        = {{OpenSage: Self-programming Agent Generation Engine}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/66121}
}

@inproceedings{sage2026,
  author       = {Jun Liu and Peilin Liu and Ruicheng Zhang and Hai Jin and others},
  title        = {{SAGE: A Dataflow-Native Framework for Modular, Controllable, and Transparent LLM-Augmented Reasoning}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/63809}
}

@inproceedings{deepha2026,
  author       = {Zihao Wang and Muyao Li and Yitao Liang and others},
  title        = {{DeepHA: Scaling Action Chains Elicits Deep Hierarchical Agents}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/66241}
}

@inproceedings{agentic2026,
  author       = {Zhengbo Jiao and Shaobo Wang and Linfeng Zhang and others},
  title        = {{Agentic Proposing: Enhancing LLM Reasoning via Compositional Skill Synthesis}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/62973}
}

@misc{from2026,
  author       = {Mengzhuo Chen and Junjie Wang and Qing Wang and others},
  title        = {{From Failed Trajectories to Reliable LLM Agents: Diagnosing and Repairing Harness Flaws (HarnessFix)}},
  howpublished = {arXiv preprint},
  eprint       = {2606.06324},
  archivePrefix= {arXiv},
  year         = {2026},
  url          = {https://arxiv.org/abs/2606.06324}
}

@misc{the2026,
  author       = {Kyungmin Kim and Youngbin Choi and Sangdon Park and others},
  title        = {{The Interplay of Harness Design and Post-Training in LLM Agents}},
  howpublished = {arXiv preprint},
  eprint       = {2606.25447},
  archivePrefix= {arXiv},
  year         = {2026},
  url          = {https://arxiv.org/abs/2606.25447}
}

@inproceedings{aflow2026,
  title        = {{AFlow: Automating Agentic Workflow Generation}},
  booktitle    = {International Conference on Learning Representations (ICLR)},
  note         = {Oral},
  year         = {2025},
  url          = {https://arxiv.org/abs/2410.10762}
}

@inproceedings{agentnoisebench2026,
  author       = {Ruipeng Wang and Yuxin Chen and Tat-Seng Chua and others},
  title        = {{AgentNoiseBench: Benchmarking Robustness of Tool-Using LLM Agents Under Noisy Conditions}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/64355}
}

@inproceedings{beyond2026,
  author       = {Wonjoong Kim and Sangwu Park and Chanyoung Park and others},
  title        = {{Beyond the Final Answer: Evaluating the Reasoning Trajectories of Tool-Augmented Agents (TRACE)}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/64255}
}

@inproceedings{complexmcp2026,
  author       = {Yuanyang Li and Xue Yang and Hongyang Chen and others},
  title        = {{ComplexMCP: Evaluation of LLM Agents in Dynamic, Interdependent, and Large-Scale Tool Sandbox}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/66421}
}

@inproceedings{realtimetool2026,
  author       = {Xiaoxin Shi and Jiaxin Wan and Zengfeng Huang and others},
  title        = {{RealtimeTool: Parallel Decoding for Real-Time LLM Function Calling}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/62617}
}

@inproceedings{appworldul2026,
  author       = {Junzhi Chen and Harsh Trivedi and Ashish Sabharwal and others},
  title        = {{AppWorld-UL: Benchmarking Diverse Agent-User Interactions for Tool-Use}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/62856}
}

@inproceedings{dcore2026,
  title        = {{D-CORE: Incentivizing Task Decomposition in Large Reasoning Models for Complex Tool Use}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/61056}
}

@inproceedings{sciagentgym2026,
  title        = {{SciAgentGym: Benchmarking Multi-Step Scientific Tool-Use in LLM Agents}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/66785}
}

@inproceedings{vero2026,
  author       = {Varun Ursekar and Apaar Shanker and Veronica Chatrath and Yuan Xue and Samuel Denton},
  title        = {{VeRO: An Evaluation Harness for Agents to Optimize Agents}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/60518}
}

@inproceedings{autowebworld2026,
  author       = {Yifan Wu and Yiran Peng and Yuyu Luo and others},
  title        = {{AutoWebWorld: Synthesizing Infinite Verifiable Web Environments via Finite State Machines}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://arxiv.org/abs/2602.14296}
}

@inproceedings{cuarewardbench2026,
  author       = {Haojia Lin and Xiaoyu Tan and Xing Sun and others},
  title        = {{CUARewardBench: Benchmarking Reward Models on Computer-using Agent Trajectories}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/63367}
}

@inproceedings{enterpriseopsgym2026,
  author       = {Shiva Malay and Perampalli Shravan Nayak and Sai Rajeswar Mudumba and others},
  title        = {{EnterpriseOps-Gym: Environments and Evaluations for Stateful Agentic Planning and Tool Use in Enterprise Settings}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/64162}
}

@inproceedings{base2026,
  title        = {{Base Models Know How to Reason, Thinking Models Learn When}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://arxiv.org/abs/2510.07364}
}

@inproceedings{recognize2026,
  author       = {{NJU-NLP}},
  title        = {{Recognize Your Orchestrator: An Entropy Dynamics Perspective for LLM Multi-Agent Systems}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://arxiv.org/abs/2606.01351}
}

@misc{adaptorch2026,
  title        = {{AdaptOrch: Task-Adaptive Multi-Agent Orchestration in the Era of LLM Performance Convergence}},
  howpublished = {arXiv preprint},
  eprint       = {2602.16873},
  archivePrefix= {arXiv},
  year         = {2026},
  url          = {https://arxiv.org/abs/2602.16873}
}

@inproceedings{dsgym2026,
  title        = {{DSGym: A Standardized and Holistic Framework for Advancing Data Science Agents}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/66567}
}

@inproceedings{swecompass2026,
  title        = {{SWE-Compass: Towards Unified Evaluation of Agentic Coding Abilities for LLMs}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/64552}
}

@inproceedings{davincidev2026,
  author       = {Ji Zeng and Dayuan Fu and Pengfei Liu and others},
  title        = {{daVinci-Dev: Agent-native Mid-training for Software Engineering}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  note         = {Oral},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/oral/71032}
}

@inproceedings{jailbreak2026,
  author       = {Zhicheng Fang and Jingjie Zheng and Chenxu Fu and Wei Xu},
  title        = {{Jailbreak Foundry: From Papers to Runnable Attacks for Reproducible Benchmarking}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  note         = {Oral},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/oral/71103}
}

@inproceedings{quantifying2026,
  author       = {Rahul Marchand and Art Cathain and Harry Coppock and others},
  title        = {{Quantifying Frontier LLM Capabilities for Container Sandbox Escape (SandboxEscapeBench)}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  note         = {Oral},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/oral/71104}
}

@inproceedings{aorchestra2026,
  author       = {{FoundationAgents}},
  title        = {{AOrchestra: Automating Sub-Agent Creation for Agentic Orchestration}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://arxiv.org/abs/2602.03786}
}

@inproceedings{infopo2026,
  title        = {{InfoPO: Information-Driven Policy Optimization for User-Centric Agents}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://arxiv.org/abs/2603.00656}
}

@inproceedings{agentconductor2026,
  title        = {{AgentConductor: Topology Evolution for Multi-Agent Competition-Level Code Generation}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/66333}
}

@inproceedings{safeharbor2026,
  title        = {{SafeHarbor: Hierarchical Memory-Augmented Guardrail for LLM Agent Safety}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/64556}
}

@inproceedings{the2026b,
  title        = {{The Oversight Game: Learning to Cooperatively Balance an AI Agent's Safety and Autonomy}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://arxiv.org/abs/2510.26752}
}

@inproceedings{when2026,
  title        = {{When Benign Inputs Lead to Severe Harms: Unintended Harmful Behaviors in Computer-Use Agents}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/64242}
}

@inproceedings{anchor2026,
  title        = {{ANCHOR: Automated Alignment Auditing for CLI Agents Detecting Real-World Harmful Outcomes}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/63234}
}

@inproceedings{causalarmor2026,
  title        = {{CausalArmor: Efficient Indirect Prompt-Injection Guardrails via Causal Attribution}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/62590}
}

@inproceedings{constitutional2026,
  title        = {{Constitutional Black-Box Monitoring for Scheming in LLM Agents}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/65104}
}

@inproceedings{tracer2026,
  author       = {Ranganath Krishnan},
  title        = {{TRACER: Trajectory Risk Aggregation for Critical Episodes in Agentic Reasoning}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://arxiv.org/abs/2602.11409}
}

@inproceedings{air2026,
  title        = {{AIR: Improving Agent Safety through Incident Response}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://arxiv.org/abs/2602.11749}
}

@inproceedings{architecture2026,
  title        = {{Architecture Matters for Multi-Agent Security}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://arxiv.org/abs/2604.23459}
}

@inproceedings{agentspec2026,
  title        = {{AgentSpec: Customizable Runtime Enforcement for Safe and Reliable LLM Agents}},
  booktitle    = {International Conference on Software Engineering (ICSE)},
  year         = {2026},
  url          = {https://arxiv.org/abs/2503.18666}
}

@misc{progent2026,
  title        = {{Progent: Programmable Privilege Control for LLM Agents}},
  howpublished = {arXiv preprint},
  eprint       = {2504.11703},
  archivePrefix= {arXiv},
  year         = {2025},
  url          = {https://arxiv.org/abs/2504.11703}
}

@misc{safemcp2026,
  title        = {{SafeMCP: Proactive Power Regulation via Environment-Grounded Look-Ahead Reasoning}},
  howpublished = {arXiv preprint},
  eprint       = {2606.01991},
  archivePrefix= {arXiv},
  year         = {2026},
  url          = {https://arxiv.org/abs/2606.01991}
}

@inproceedings{swebench2026,
  author       = {Xiang Deng and Jeff Da and Brad Kenstler and others},
  title        = {{SWE-Bench Pro: Can AI Agents Solve Long-Horizon Software Engineering Tasks?}},
  booktitle    = {Proceedings of the International Conference on Machine Learning (ICML)},
  year         = {2026},
  url          = {https://icml.cc/virtual/2026/poster/61047}
}