Agentic Hazard Mapper is a multi‑agent system that automates the creation of hazard maps by emulating professional cartographic workflows and design trade‑offs. The system adapts its design decisions to varying data, locations, and hazard contexts, and was evaluated through a formative study with map experts and a user study with hazard practitioners. Results showed that the generated maps improved spatial understanding and evacuation‑planning performance, reduced perceived workload and task completion time, and were generally preferred over municipal maps.
By Sidi Wu, Lifeng Lin, Yizi Chen, Anna Yokokubo, Lorenz Hurni
HarnessMM (Harness Minus Minus) is a method that improves harness token efficiency by using guarded decision reuse and recursive self‑improvement (RSI). It refines harness configurations from training feedback, reuses resolved decisions to replace later reasoning callbacks, and stores actions in Executable Decision Graphs. Experiments show significant token savings while maintaining or improving success rates on benchmarks such as HumanEval and DeepSeek Harness.
By Yuhua Jiang, Chang Xu, Feifei Gao, Li Zhao
The paper introduces an HPC help‑desk assistant that combines retrieval‑augmented generation over live documentation with a fine‑tuned Llama‑3‑8B model trained on historical ticket resolutions. The system merges answers from both sources, running entirely on‑premises to keep ticket data private. Evaluation on 400 held‑out tickets from Purdue’s Rosen Center shows that fine‑tuning on rewritten resolutions yields the best agreement with official resolutions, though the end‑to‑end merge does not outperform the individual tools.
By Elham J. Barezi, Primus C Kabuo, Niveditha Nerella
Compact‑Bench evaluates how context compaction affects long‑horizon coding agents by comparing full‑history and summary‑only continuations across 35 software‑engineering repairs, 146 agentic sessions, and 106 production chains. The benchmark tests 14 summarizer backbones and finds that stable damage ranges from 5‑64% of full‑history successes, with Claude Opus preserving the most repairs and DeepSeek‑Flash causing the most damage. It also shows that LLM summaries retain far fewer recurring details than mechanical extraction, and that failures can recur across all backbones and prompts, indicating that merely preserving information is insufficient for successful task continuation.
By Ruopei Sun, Jianfeng Cai, Kangwen Zhao, Wengang Zhou, Houqiang Li
Kolibri is an English‑German mixture‑of‑experts transformer with 78.1 B total parameters and 3.46 B active parameters per token, released under the Apache 2.0 licence. It was trained on 24 T tokens, with German comprising over 20 % of the data, and includes supervised fine‑tuning and reinforcement learning on more than 1.2 M curated tasks. The model activates only 4.4 % of its parameters per token and, among open models evaluated, sits on the Pareto frontier of quality and serving cost for both English and German.
By Aleph Alpha, :, Moritz Althaus, Artur Baranowski, Michael Barlow, Fabien C. Y. Benureau, Felix Berkenkamp, Amrita Bhattacherjee, Lukas Bl\"ubaum, Duncan Blythe, Augustijn de Boer, Bastian Boll, Thomas F. Burns, Bj\"orn Deiseroth, Niko D\"urr, B\'eni Bal\'azs Egressy, Niklas Finken, Niklas Freymuth, Fabian Gebhart, Max Hammer, Bastian Harren, Andreas Hartel, Johann Higl, Steffen Hirschmann, Pablo Iyu Guerrero, Yasser Jadidi, Andreas Jagusch, Mitja Kleider, Markus Klein, Jonas Knupp, Murali Kondragunta, Garwin Lechner, Dominik L\^e-Kellner, Jurek Leonhardt, Jiaze Li, Sohir Maskey, Omar Mehio, Pit Neitemeier, Letitia Parcalabescu, Alejandro Petry Pacheco, Pranav Ragupathy, Martin Reinhardt, Peter Rodenkirch Nfissi, Dylan Rodriquez, Leonard Salewski, Raziye Sari, Jordan Sassoon, Ivo Schaper, Ilhan Scheer, Frank Schneider, Philipp Scholl, Lukas Schott, Eric Schreiber, Pablo Schumacher, Alessio Serra, Philipp Siedler, Prabhu Teja Sivaprasad, Mohamed Soliman, Till Speicher, Volker Stampa, Simon Thel, Serge Thilges, Irina Vidal Migall\'on, Denis Wa{\ss}mann, Samuel Weinbach, Sascha Wirges, Michael W\"uthrich, Alexei Zacharov
The paper introduces MCP‑Injector, a fault‑injection middleware that simulates realistic failures in the Model Context Protocol used by large‑language‑model agents. Using data from over 57,000 issue and pull‑request threads, the authors calibrate a fault model and run 1,350 experiments across different backbone sizes, agent stacks, and fault conditions. Results show that permanent‑heavy faults drastically reduce task‑completion scores and that larger models (120B) do not gain fault tolerance, highlighting that framework‑level error handling is more critical than model scale.
By Antonio Ken Iannillo, Joshua Segun Owotogbe, Roberto Natella, Francesco Avallone, Kristina Kudryavtseva, Indika Kumara
SceneBloom is an agentic framework that generates realistic, varied, and simulation-ready indoor environments for embodied learning. It combines diffusion-based visual layout sampling with a contract-guided 3D compiler that uses extracted relations for geometric fitting and pose correction. The method outperforms baselines across five room types, achieving higher visual quality, physical stability, and better alignment with real room distributions, and improves object navigation policy success rates on real scans.
By Yuhan Liu, Lianhui Qin, Shenji Wan
The paper introduces Back-to-the-Future (BTTF), an agentic framework designed to automate post-simulation verification and waveform debugging in chip design. BTTF transforms large simulation dumps into a normalized SQLite database and uses a multi-agent orchestration engine to convert natural-language queries into schema-aware SQL, linking anomalies to versioned RTL code. In a benchmark of 150 queries, BTTF achieves 95.33% execution accuracy, demonstrating a viable path toward autonomous EDA verification.
By Je Yang, Ivan Lobov, Thomas Karpati
DriveOSS is a new real‑world multimodal dataset and benchmark for 3D anomaly segmentation and open‑set semantic segmentation in autonomous driving. It contains 12,661 synchronized frames from six RGB cameras, a 128‑beam LiDAR, and GPS/IMU, covering 52 driving scenes under various weather conditions. The data are split into in‑distribution (ID) scenes for training and out‑of‑distribution (OOD) scenes for validation and testing, with unknown objects labeled as an outlier class to challenge current 3D perception systems.
By Zhiran Yan, Savankumar Bhanderi, Sourabh Lolge, Rui Song, Girdharbhai Paneliya, Shiva Agrawal, Gordon Elger
WorldMirror is a real‑to‑sim framework that builds interactive digital twins from real‑world robot demonstrations by iteratively refining an executable world program. A coding agent decomposes the observed scene, plans reconstruction steps, and coordinates image and 3D generation tools, using verification feedback to correct errors. The resulting camera‑pose‑anchored scene representation preserves spatial relationships and enables physically consistent interaction and trajectory replay, as shown on the DROID dataset.
By Tianling Xu, Weijie Wan, Honghao Su, Kailun Su, Junyuan Tang, Huichi Zhou, Haozheng Zhang, Qi Xiong, Ping Luo, Wenbo Ding, Tianxing Chen, Yue Chen
RSure-Agent is a framework designed to improve remote sensing agents by verifying tool observations before they influence downstream reasoning. It introduces a verifiable observation protocol that requires tools to provide process evidence, and builds a task‑tool reliability prior from offline feedback to assess each tool configuration’s past performance. In evaluations on EarthBench, ThinkGeo, TerraLogic, and CHOICE‑420, RSure-Agent cuts error propagation by 21.3–25.9 percentage points and boosts overall accuracy, while also reducing the number of tool calls.
By Fuyuan Liu, Nayu Liu, Wenhao Yu, Peijin Wang, Yingchao Feng, Fanglong Yao, Yue Zhou, Xue Yang, Chao Ren, Liang Wan, Wei Feng
The paper surveys 3D generation techniques tailored for embodied AI and robotics, emphasizing that beyond visual realism, generated content must include kinematic structure, material properties, and interaction capabilities. It categorizes the literature into three roles: Data Generator, Simulation Environments, and Sim2Real Bridge, each addressing different needs in training and deployment. The survey highlights a shift toward interaction readiness and identifies key bottlenecks such as limited physical annotations, quality gaps, fragmented evaluation, and the sim‑to‑real divide.
By Tianwei Ye, Yifan Mao, Minwen Liao, Jian Liu, Chunchao Guo, Dazhao Du, Quanxin Shou, Fangqi Zhu, Song Guo
BARC-Bench is a new benchmark comprising 110 tasks across nine simulated workspace environments designed to expose the safety risks of underspecified user instructions for agents. The benchmark evaluates whether agents respect safety boundaries defined by workspace records, implicit environmental context, and unstated social norms, such as avoiding premature disclosure of an employee’s resignation. Results show that even the best model configurations fail to consistently act safely, with safe completion rates ranging from 9.1% to 72.7% and a 27.3% safety violation rate in the strongest setup.
By Kangkang Li, Kelvin Luu, Lily Ma, Tu Trinh, Ying Liu, Mohamed Elfeki
The paper examines how large language model (LLM) agents can use trust estimates in repeated multi‑party, multi‑issue negotiations where one participant covertly tries to block agreement. By introducing a framework where agents maintain pairwise trust scores across successive games, the study compares trust‑aware and trust‑agnostic agents. Results show a trust–agreement gap: while trust‑aware agents quickly spot the adversary and trust can raise agreement rates by up to 23 percentage points, this benefit only materializes when agents are explicitly guided to incorporate trust into their decision‑making.
By Amit Sarker, Moumita Choudhury, Saaduddin Mahmud, Shlomo Zilberstein
The paper proposes a hierarchical framework for distilling large language model agents, distinguishing between the policy that generates interaction histories and the policy that produces token prefixes for actions. It explores various combinations of teacher-generated and student-generated data across multiple models and environments, finding that pairing student-induced histories with teacher-generated action prefixes yields the strongest training signal, while fully on-policy data is only beneficial when mixed with other configurations. The study also shows that the specific pairing of teacher and student proportions at each level significantly impacts performance, and that strategically front-loading teacher-generated contexts and delaying on-policy contexts improves results.
By Ya Gao, Zhengtao Zou, Pekka Marttinen
The paper investigates how AI agents can reliably recover from changes in authorization while executing tasks that involve external tools. It defines distinct obligations—checkpointing, permission to dispatch, evidence of external effects, and permission to reconcile—and evaluates them through a finite conformance study on 336 histories across multiple hosts, schedules, and provider contracts. The study demonstrates that all paired host cells produce identical outcomes and that various retry strategies (cached, revalidation, stable-key) successfully pass the majority of histories, while a component study of InfinityOS further isolates the impact of authority, contracts, and observations on recovery.
By Deekshith Marla, Kushagra Bhatnagar
TACS‑RF is a framework that uses a large‑language‑model coding agent (Claude Opus 4.8) to drive industry‑standard EDA tools for designing RF and mmWave modules. The agent autonomously designed a 240‑251 GHz sub‑THz transmitter in a 90 nm SiGe BiCMOS process, creating sub‑blocks such as a 60 GHz VCO, power drivers, a frequency doubler, a bowtie antenna, and matching networks. The resulting transmitter delivers +5.1 to +6.3 dBm over an 11 GHz tuning range, with phase noise of –80 to –88 dBc/Hz at 1 MHz offset while consuming 361 mW of DC power.
By Tristan Liang, Aravind Nagulu
Phronix introduces a capability‑centric learning paradigm that lets language‑model agents autonomously acquire, validate, and refine reusable capabilities from their own executions. Through a predictor‑validator‑corrector loop, it stores every interaction as a capability, enabling agents to solve recurring tasks on the first attempt in 93‑95% of cases and generate 7.6‑30.8 durable capabilities per task without user instruction. In practical benchmarks, Phronix reduces token consumption, improves reliability, and boosts task‑solving rates of cheaper models to match those of the most expensive ones.
By Hariharan Devarajan, Neeraj Rajesh, Aniruddha Marathe, Nikoli Dryden, Kin Ng, Michela Taufer
PersonaScienceBench is a new benchmark comprising 200 tasks drawn from five existing benchmarks across four scientific domains. It introduces simulated researchers with distinct persona profiles who provide context and feedback during interactions, while the task objectives and grading remain unchanged. The benchmark evaluates scientific agents on how they collaborate with diverse researchers, assessing task performance, fact recovery, and self-repetition across eighteen model configurations.
By Siyang Wu, Weizhi Zhang, Chen Yang, Esther Xu Fei, Jorin Overwiening, Xu Pan, Jing Tang, Yixuan He, Kelvin C. Mo, Peiwen Li, Defu Cao, Xu Hu, Haixin Wang, Taoran Li, Haoming Li, Zheyuan Deng, Jingwen Zhang, Pengsong Zhang, Haichao Zhang, Guangze Luo, Yi Daniel Lu, Zihuai He, Chenxi Qiu, Dianzhuo Wang, Zhejian Peng, Shushu Liang, Wenyi Lu, Yifan Wang, Jintao Huang, Yijun Wang, Xiaomin Li, Yuexing Hao
The paper proposes a general framework for analysing the behaviour of large language model-driven agents. It connects data-driven behavioural discovery with analysis and experimental investigation, using categories derived from recorded activity to measure behaviours and preserve their positions in the process. The framework includes statistical assessment in held-out contexts and experiments such as prompt steering to test how behaviours and outcomes change under varied conditions, illustrated by a study of iterative self-reflection.
By Tianqiang Yan, Lizhen Qu