arXiv AI
Humanize is a multi‑agent workflow for agentic coding that uses explicit, mechanically enforced decisions—called judgement engineering—to separate planning, implementation, review, and learning. A human approves a plan contract, a builder agent implements it in rounds, and a reviewer agent from another vendor decides completion, with deterministic hooks and 72 mechanical gates ensuring that a defect survives only if both models miss it. The system has been deployed in over 68 versions in 108 days, earned 1,468 GitHub stars, and powered applications such as a 567‑file gem5 build‑system migration, Kernel Design Agents that excelled in MLSys 2026 contests, and Humanize Olympiad Agents that achieved top scores in multiple international competitions and high rankings on PutnamBench and Lean‑Eval.
whyItMatters":"Humanize demonstrates that a structured, multi‑agent approach with explicit mechanical gates can improve reliability and performance in automated code generation, as evidenced by its adoption, star count, and success in competitive benchmarks."
The paper introduces HOA (Humanize Olympiad Agent), a workflow that orchestrates multiple LLM agent sessions via reviewer feedback, a structured wiki, and a verifier to tackle olympiad problems. Using GPT‑5.6 Sol, HOA achieves perfect scores on the IMO, IPhO, and IOI 2026, near‑full marks on the IChO 2026, and solves all 672 PutnamBench problems, while ranking first on the Lean‑Eval leaderboard. With open models like Kimi‑K3, HOA still attains full scores on major olympiads and a gold‑level IChO score, showing that systematic feedback can close gaps between closed‑source and open‑source agents.
By Zhengyang Zhang, Jing Xiong, Hongzhou Lin, Juihui Chung, Zijian Zhang, Song Han, Sihao Liu, Ligeng Zhu
arXiv:2610.18760v1 Announce Type: new
Abstract: Test-time scaling for large language models (LLMs) can involve generating multiple candidate solutions for selection or revising an earlier attempt. To...
By Zheye Deng, Sha Li, Changlong Yu, Xin Liu, Qin Lu, Priyanka Nigam, Yangqiu Song
arXiv:2606. 01490v1 Announce Type: cross Abstract: We present a controlled experiment evaluating 12 multi-agent LLM collaboration topologies for software architecture design.
By Nagarjuna Kanamarlapudi, Praveen K
arXiv:2604. 16706v2 Announce Type: replace Abstract: Automated evaluation of tool-using large language model (LLM) agents is widely assumed to be reliable, yet this assumption is rarely validated against human annotation.
By Bhaskar Gurram
AgentSWE is a new benchmark for agent software engineering that evaluates coding agents across three lifecycle stages: creating an agent from a natural‑language requirement, editing existing agent codebases, and optimizing toward a measurable target. The benchmark consists of 25 commission‑style tasks, where builders iterate on a few illustrative cases and are then tested on held‑out cases that cover the full requirement, with hard pass/fail gates or objective‑based grading. Experiments show that current coding agents fail to meet the requirements, revealing common failure modes such as overfitting, incomplete runtimes, and unsupported completion claims.
By Jiahao Wang, Hongjin Qian, Yuyang Hu, Jiajun Zhang, Zheng Liu
The paper introduces Pufibara, an agent harness designed to maintain engineering state and evidence across revisions in Modelica-based physical system modeling. It also presents a 232-task Modelica Agent Workflow Benchmark covering model repair, generation, and tuning, evaluated by an external benchmark-owned evaluator. Experiments show Pufibara outperforms Claude Code in task success and resource efficiency across two LLM backends.
By Zizhe Wang
Agents are increasingly asked to design hardware, and increasingly reported to succeed. Such reports establish that a design improved; they cannot establish why. An agent that improves an accelerator...
arXiv:2607. 15388v1 Announce Type: new Abstract: Many math- and science-oriented agent systems use hierarchical designs with specialized reviewer roles, assuming that a dedicated review stage should help turn wrong candidates into correct ones.
By Chih-Hsuan Yang, Jingyan Jiang, Vikram Vasudevan, Cheng-Hau Yang, Huihuo Zheng, Le Chen, Eliu A. Huerta, Venkatram Vishwanath, Ian T. Foster, Rajeev Thakur
The paper investigates whether AI agents truly understand computer architecture by comparing their performance when the same 15‑dimensional accelerator design space is presented either as meaningful architectural knobs or as anonymous variables. Using the AutoTuring framework, the authors find that agents with architectural knowledge outperform blind agents on a nine‑kernel FP16 GEMM benchmark, yet a critic loop can largely recover this advantage. The study highlights that architectural knowledge and structured critique act as substitutes rather than complements in improving agent performance.
By Ambika Sharan, Grigory Chirkov, Soheil Abbasloo
Dr. Claw is an open‑source AI scientist workspace that integrates existing command‑line coding agents into a single, auditable, human‑in‑the‑loop workflow. It uses persistent state objects, a reusable skill library, and multi‑executor coordination to link human decisions with AI execution, creating a traceable and recoverable loop for planning, execution, and writing. The authors demonstrate the system with an interactive scenario and a failure‑recovery walkthrough, and show that, when the underlying executor is held constant, Dr. Claw achieves higher research completeness while preserving an auditable process trail.
By Dingjie Song, Hanrong Zhang, Dawei Liu, Yixin Liu, Zongxia Li, Zhengqing Yuan, Siqi Zhang, Henry Peng Zou, Zhiling Yan, Yuxuan Zhang, Yanfang Ye, Philip S. Yu, Lichao Sun
arXiv:2607. 21268v1 Announce Type: cross Abstract: In many social-science research tasks, such as economics, LLM-based agents must produce outputs for which no cheap, task-complete, machine-readable correctness signal exists.
By Chen Zhu, Xiaolu Wang, Weilong Zhang
arXiv:2607. 03451v1 Announce Type: cross Abstract: While skill optimization for autonomous agents has gained traction, existing methods rely on complex pipelines.
By Yifei Shen, Bo Li, Xinjie Zhang