PRISM is an agentic retrieval framework that uses large language models in a structured loop to improve evidence gathering for multi‑hop question answering. It splits retrieval into three specialized agents—a Question Analyzer, a Selector focused on precision, and an Adder focused on recall—whose iterative interaction yields a compact yet comprehensive evidence set. Experiments on HotpotQA, 2WikiMultiHopQA, MuSiQue, and MultiHopRAG show that PRISM consistently outperforms strong baselines by achieving higher retrieval accuracy and filtering out distracting content.
By Md Mahadi Hasan Nahid, Davood Rafiei
The paper introduces VICT, a method that leverages the internal structure of verifiable tasks to perform fine‑grained credit assignment for long‑horizon LLM agents. VICT exposes executable or evidence‑backed atoms from a task’s terminal verifier and traces them back to actions via dependency‑valid proof edges, redistributing advantage only along these edges. This approach improves performance on ALFWorld and WebShop compared to outcome‑only training and matches recent fine‑grained credit methods without requiring additional critics, labels, or inference‑time verifier access.
By Pengcheng Li, Zhengyang Zhang, Dongxu Zhang, Sui Huang, Shaohua Ma
arXiv:2601.13247v2 Announce Type: replace-cross
Abstract: Current Large Language Models (LLMs) exhibit a critical modal disconnect: they possess vast semantic knowledge but lack the procedural ground...
By Baochang Ren, Yunzhi Yao, Rui Sun, Shuofei Qiao, Ningyu Zhang, Huajun Chen
The paper proposes a label‑free 3D perception framework where roadside units (RSUs) act as unsupervised teachers for self‑driving cars. RSUs learn local 3D detectors from unlabeled data and broadcast predictions to passing vehicles, which use these as pseudo‑labels to train an ego‑centric detector. In a CARLA simulation, the method achieves 82.3% AP for vehicle detection, approaching a fully supervised upper bound of 94.4%, and demonstrates scalability and complementarity with existing ego‑centric approaches.
By Zhen Xu, Jinsu Yoo, Cristian Bautista, Zanming Huang, Tai-Yu Pan, Zhenzhen Liu, Katie Z Luo, Mark Campbell, Bharath Hariharan, Wei-Lun Chao
arXiv:2605.12015v3 Announce Type: replace-cross
Abstract: Reusable skills are becoming a common interface for extending large language model agents, packaging procedural guidance with access to files...
By Chang Jin, An Wang, Zeming Wei, Kai Wang, Biaojie Zeng, Qiaosheng Zhang, Chao Yang, Jingjing Qu, Xia Hu, Xingcheng Xu
The paper investigates how scaling video diffusion models affects performance on autonomous driving data, training models ranging from 1 M to 9 B parameters on up to 5,500 hours of driving footage. Validation loss follows predictable power‑law trends with both model size and training exposure, showing that longer training yields faster loss reduction than increasing model size, though larger models still achieve lower asymptotic loss. Guided by these findings, the authors train a 9 B‑parameter model that becomes the largest video diffusion model trained from scratch on driving data and sets a new open‑source state‑of‑the‑art on the nuScenes benchmark.
By Victor Besnier, Anh-Quan Cao, Elias Ramzi, Spyros Gidaris, Tuan-Hung Vu, Andrei Bursuc, Eloi Zablocki, Matthieu Cord
D3‑Gym is the first automatically constructed dataset that provides verifiable environments for scientific data‑driven discovery, comprising 565 tasks from 239 real scientific repositories across four disciplines. Each task includes a natural‑language instruction, an executable environment with pre‑installed dependencies, dataset previews, a reference solution, and an automatically synthesized evaluation script that achieves 87.5% agreement with human‑annotated gold standards. Training on trajectories sampled from D3‑Gym consistently improves Qwen3 models on ScienceAgentBench, and the platform also serves as a testbed for studying agentic optimization loops such as Autoresearch on real scientific workflows.
By Hanane Nour Moussa, Yifei Li, Zhuoyang Li, Yankai Yang, Cheng Tang, Tianshu Zhang, Nesreen K. Ahmed, Ali Payani, Ziru Chen, Huan Sun
arXiv:2608.27505v1 Announce Type: new
Abstract: Reinforcement learning from human feedback (RLHF) has become the dominant paradigm for aligning large language models (LLMs) with human preferences. Ho...
By Zifei Shan, Fangning Shao
Computer use agents (CUAs) are vision-language models that perceive a screen and act on a real operating system through mouse, keyboard, and terminal, and they are increasingly deployed to automate ev...
OpenAI launched ChatGPT Work on July 9, offering two distinct products: Work Cloud, accessed via chatgpt.com or the mobile apps, and Work Local, available through the desktop app that can run programs on a user’s computer. Work Cloud is a paid‑only feature for $20/month and up subscribers, providing advanced capabilities such as code execution with internet access, a headless Chrome browser, and a persistent filesystem across sessions. The author notes that while ChatGPT Work claims to be better for tasks with clear outcomes, many users find the distinction unnecessary and prefer the standard Chat interface for most use cases.
This paper presents an end-to-end, agent-driven pipeline for the LoD3 reconstruction of facade openings in 3D city models, producing directly usable CityGML-conform outputs. In contrast to existing ap...
The article "8 Tips for Writing Effective Agent Instructions" offers quick and simple guidance to improve the clarity and effectiveness of instructions given to agents. It presents a concise set of practical tips aimed at helping readers craft better agent instructions. The post is published on Towards Data Science.
By Payal Patel
The article discusses the appropriate contexts for using Claude Code versus Codex, two coding agents. It explains the strengths and ideal use cases for each tool, helping readers decide which agent to employ for specific programming tasks. The post provides guidance on selecting the most suitable coding assistant based on the nature of the work.
By Eivind Kjosbakken
The article reports that rumors of bugs in open‑source projects now trigger rapid security exploits, with automated agents probing for vulnerabilities within minutes of a patch being discussed. Anil Madhavapeddy demonstrates how modern coding agents can exploit even minimal hints, and rclone maintainer Nick Craig‑Wood notes a dramatic spike in security disclosures and delayed CVE assignments. The post highlights the need for new processes to protect open‑source communities from swift exploitation.
The article explains how the authors eliminated the need to review every agent action by selectively directing human attention to the most critical moments. It describes the process of stopping exhaustive reviews and instead routing human oversight where it truly matters. The result is a more efficient workflow that maintains quality without sacrificing throughput.
By Priyansh Bhardwaj
The article "From One Agent to a Team: Understanding Codex Subagents" offers a practical guide on how to create specialist agents and manage their collaboration using the Codex Command Line Interface. It explains the process of defining subagents and coordinating their tasks to build a cohesive team of AI agents. The guide is aimed at developers looking to extend Codex’s capabilities through modular, specialized agent workflows.
By Shuai Guo
The article explains how to connect a LangGraph AI agent to a Postgres database. It covers running the backend locally using Docker and deploying it in the cloud. The guide provides practical steps for setting up the database connection and managing the agent’s data storage.
By Soner Yıldırım
The paper introduces SARA, a framework that separates action induction from runtime authorization in tool‑augmented LLM agents. By treating these as distinct roles, SARA uses an Action Probe to record action provenance and only authorizes tool calls that align with the user objective and past successful executions. Experiments on AgentDojo and AgentDyn show that SARA reduces action‑to‑side‑effect risk to below 0.63% while preserving task performance.
By Xiaokun Guo, Zhen Xu, Dongdong Huo, Yanqiu Zhang, Wei Wang, Qinfu Yang, Dongjin Yu, Yu Wang
Daydreaming is an execution‑only attack that steals multi‑file agent skills by interacting with a black‑box task service. By adaptively crafting tasks and analyzing the returned results, the attacker reconstructs the hidden skill without ever requesting or revealing it. In experiments on seven skills and four victim models, Daydreaming recovers 86.8% of the original capability using only 32 victim calls on average, outperforming prior methods and demonstrating that hiding skill files and filtering direct disclosure are insufficient defenses.
By Yu-Lin Tsai, Yu-An Lu, Ci-Yang Tsai, Muxi Lyu, Raluca Ada Popa, Chia-Mu Yu
The paper introduces matched trajectory replay, a protocol that fixes answer states, evidence points, budgets, and action costs to evaluate how confidence signals influence agent actions. Using this method, the authors compare raw verbalized confidence with post‑hoc isotonic calibration across six model‑dataset pairs, finding that calibration can significantly improve accuracy of committed answers but may reduce coverage and increase retrieval usage. The study concludes that calibration helps interpret commitment risk but does not predict the benefit of additional retrieval, indicating the need for separate value‑of‑information estimates.
By Prateek Chhikara