Flama is an open‑source Python framework that unifies the development and deployment of production‑ready web APIs, machine‑learning services, and large‑language‑model (LLM) applications. Built on ASGI, it offers an async‑first, type‑driven programming model with seven subsystems—including dependency injection, a pluggable schema layer, automatic CRUD generation, a portable binary model format, a multi‑backend LLM server, a Rust‑accelerated core, and a Model Context Protocol module. The framework also provides built‑in JWT authentication, pagination, background tasks, WebSocket and streaming support, OpenAPI generation, and a CLI for running, packaging, and inspecting models.
By Jos\'e A. Perdiguero L\'opez, Miguel A. Dur\'an-Olivencia
The paper introduces Harness Continual Learning (HCL), a paradigm where an agent’s state evolves through prompts, memories, tools, skills, and routing rules while keeping the underlying foundation model frozen. HCL defines harness-level forgetting and proposes a guarded evolution process involving a Continual Optimizer and Evaluator to ensure improvements without losing prior behavior. Experiments across textual reasoning, multimodal perception, and open‑world interaction show over 10% performance gains and demonstrate how the stability–plasticity trade‑off can be explicitly tuned.
By Borui Kang, Jinrui Gu, Junhan Lv, Wenbin Li, Lei Wang, Yang Gao
The paper introduces Interval POMDP Shielding for agents with imperfect perception, aiming to prevent unsafe actions when sensor readings may be misclassified. By estimating perception uncertainty from finite labeled data, the authors construct confidence intervals and model the system as a finite Interval Partially Observable Markov Decision Process. They propose an algorithm that computes a conservative belief set, enabling a runtime shield that guarantees, with high probability, that any action allowed by the shield meets a specified safety lower bound. Experiments on four case studies demonstrate that this shielding approach outperforms state‑of‑the‑art baselines in safety.
By William Scarbro, Ravi Mangal
The paper introduces EvoResearcher, a training‑free, inference‑time protocol that enables a frozen large language model to perform cost‑bounded self‑reflection and early stopping. By iterating through generate → self‑critique → revise steps until a maximum depth or a CONFIRMED sentinel is reached, the model can self‑verify its answers within a strict compute budget. The protocol incorporates four self‑reflective meta‑reward components—correctness, efficiency, reflection depth, and tool‑call diversity—implemented as prompt‑level mechanisms, and is validated on Big‑Bench Hard, GSM8K, and MATH benchmarks, achieving comparable accuracy while terminating 82‑88% of items early with only about 2.1 generations per question.
By Wei Yu, Suxing Liu, Minjie Yu, Jiahao Wang, Zhijian Zheng, Haocheng Deng, Bing Li
ChainWorld is a new benchmark that composes atomic OSWorld tasks into long‑horizon desktop workloads, creating 347 chains of length two to four that compare two renderings of the same task sequence. In single‑turn evaluation all tasks are presented together in one prompt, while in multi‑turn evaluation tasks are revealed one at a time. Across four current computer‑use agents, maximum chain completion is 31 %, with multi‑turn evaluation improving completion for three models but both protocols remaining challenging and exposing different failure profiles.
By Vincent Siu, Manasi Sharma, Dawn Song, Daniel Yue Zhang, Chenguang Wang, Ying Liu
The paper introduces rEDMRec, a method that compresses a large language model’s reasoning about user preferences and item comparisons into a compact, editable memory. This memory, organized into four channels—long‑term preference, short‑term context, item perception, and counterfactual hard‑negative comparisons—can be updated by an LLM controller and queried by a lightweight student LLM for ranking, eliminating the need to re‑run the expensive teacher model for each request. Experiments on ML‑1M, Amazon Beauty, and Steam datasets show that rEDMRec consistently outperforms zero‑shot, few‑shot, RAG, and GraphRAG baselines, achieving up to a 13.3% improvement in HR@1 on ML‑1M.
By Minh Hoang Nguyen, Tung Le, Huy Tien Nguyen
The paper argues that many failures in large language model–based multi‑agent systems stem from concurrency control issues rather than coordination or communication problems. It highlights how concurrent reads and writes to shared state, coupled with long inference windows, lead to stale data, lost updates, and inconsistent outcomes. The authors propose that MAS frameworks incorporate explicit concurrency controls—such as conflict detection, isolation guarantees, and structured resource access—to make concurrency a primary design concern.
By Xin Yang, Letian Li, Zimo Ji, Terry Jingchen Zhang, Wenyuan Jiang
The paper proposes a systematic framework for creating a "Map of Datasets in Engineering Design and Systems Engineering" (EDSE) to address the fragmented and inaccessible nature of existing datasets. It introduces a multi‑dimensional taxonomy that classifies datasets by domain, lifecycle stage, data type, and format, and presents an interactive discovery tool built on a knowledge graph data model. The authors analyze the current data landscape, identify underrepresented areas such as early‑stage design and system architecture, and suggest strategies for curation and sustainability to build a dynamic, community‑driven resource.
By H. Sinan Bank, Daniel R. Herber
The paper studies optimal placement of millimeter-wave base stations in a realistic, non-convex campus layout using deep reinforcement learning. It compares four DRL methods—single-agent DQN, multi-agent partitioned DQN, single-agent DDPG, and multi-agent partitioned DDPG—and finds that the multi-agent DDPG approach achieves full coverage, a Jain's fairness index of 0.94, and superior performance in dense scenarios with 400 users. The multi-agent DDPG also converges more efficiently than single-agent methods.
By Omar Rady, Mohamed Ayman, Ali Arafa, Mohamed Shalma
Adversarial Review (AR) is a minimal cooperative code‑review protocol that employs a main coding agent, a reviewer, and a critic. The reviewer evaluates code while the critic audits the review through structured disagreement before the main agent edits. On multiple benchmarks (LiveCodeBench, SWE‑PRBench, SWE‑bench Verified), AR achieves higher pass rates or F1 scores than larger multi‑agent baselines, demonstrating that effective code review can be achieved with only three agents and minimal, evidence‑grounded disagreement.
By Eric S. Qiu, Joyce Gill
The paper introduces a production-ready framework that connects e‑commerce search and CRM systems via AI‑powered Product Research Agents. These agents detect users with exploratory purchase intent, perform multi‑agent research using behavioral data, external knowledge, and catalog information, and then send personalized product recommendations through WhatsApp. In a 23‑day deployment, the system sent about 15,000 notifications, achieving higher click‑through rates than standard campaigns and generating downstream purchases and GMV gains.
By Mandar Kulkarni, Pooja A., Samir Shah
The paper titled "Preference Reasoning under Indeterminacy in Large Language Models" discusses how large language models (LLMs) are becoming decision‑making agents and the importance of reasoning over preferences for alignment, coordination, and collective intelligence. It highlights that real‑world preference reasoning is inherently indeterminate, with incomplete information and potentially no valid solutions. The authors formalize this challenge into epistemic indeterminacy (incomplete or expressive preferences) and structural indeterminacy (non‑existence of solutions under standard social choice concepts), and demonstrate that current LLMs fail to differentiate between determined and undetermined cases, showing miscalibrated reasoning even in verification tasks.
By Hadi Hosseini, Samarth Khanna, Xiyuan Wang
DentAgent is an evidence‑centric multi‑agent framework designed for multimodal dental reasoning. It coordinates five specialized agents that process domain knowledge, radiographs, intraoral photographs, and 3D dental data, converting observations into structured evidence records. The Evidence Blackboard tracks coverage, gaps, and conflicts before generating responses, and the system outperforms senior specialists by 17.3 percentage points on multi‑label diagnosis across four benchmarks.
By Zijie Meng, Xiwei Dai, Yixuan Tang, Jin Hao, Yang Feng, Fudong Zhu, Xiaoqiang Liu, Shaosheng Cao, Zuozhu Liu
The paper introduces an adaptive memory and reflection (AMR) multi‑agent system for medical question answering. Each agent has dedicated memory and uses reflection‑based feedback to retrieve relevant prior cases, improving reasoning. The system routes questions through solo, collaborative, or escalated workflows and includes consensus and ethical overseer modules, achieving strong performance on MedQA and MedMCQA datasets.
By Pradeep Murugesan, Luoxiao Yang, Xueli Chen, Xinqi Fan
The paper "Jailbreaking in the Haystack" introduces NINJA, a jailbreak technique that exploits long-context language models by appending benign, model-generated content to harmful user goals. It demonstrates that the position of harmful goals within the context is crucial for safety, and shows that NINJA significantly boosts attack success rates on models such as LLaMA, Qwen, Mistral, and Gemini. Unlike previous methods, NINJA is low-resource, transferable, less detectable, and compute‑optimal, revealing that carefully crafted benign long contexts can expose fundamental vulnerabilities in modern LMs.
By Rishi Rajesh Shah, Chen Henry Wu, Shashwat Saxena, Ziqian Zhong, Alexander Robey, Aditi Raghunathan
The paper introduces a post‑training framework that teaches a 4B‑parameter language model to exercise task‑conditioned authority in executable terminal and Model Context Protocol (MCP) environments. By auditing each action across six risk dimensions with deterministic verifiers and optimizing for task‑specific excess‑privilege values, the authors achieve 98.48% safe success and reduce excess‑authority errors from 4.56% to 0.79% on held‑out tasks. The study also demonstrates capability retention, prompt‑directed improvement, and generalization over a 400‑task continuation test.
By Alexander Tu, Michael Tu
TrojanGYM is an LLM‑driven framework that automatically generates diverse hardware Trojan (HT) insertions to expose blind spots in learning‑based detectors. It uses multiple large language models to propose and refine RTL modifications, while an agentic loop with syntactic checks, functional verification, and GNN‑based detectors iteratively improves the HT designs. The authors also present Robust‑GNN4TJ, a more robust detector that improves detection rates on TrojanGYM benchmarks, and demonstrate that TrojanGYM can achieve up to 68.75% evasion against modern GNN detectors on SRAM, AES‑128, UART, and RISC‑V RTL designs.
By Saideep Sreekumar, Zeng Wang, Akashdeep Saha, Weihua Xiao, Minghao Shao, Muhammad Shafique, Ozgur Sinanoglu, Ramesh Karri, Johann Knechtel
DeepTCM1.0 is a multi‑expert AI agent built on the DeepSeek V3.2 large language model, designed to integrate classical traditional Chinese medicine (TCM) theory with modern life sciences for mechanistic analysis of TCM compound formulas. The framework employs a three‑tier collaborative architecture and a three‑round iterative quality‑control workflow, simulating 11 interdisciplinary intelligent agents. It was validated on the Guizhi Decoction, with performance evaluated through double‑blind five‑dimensional scoring, ICC reliability testing, Mann‑Whitney U tests, and effect size analysis across 100 independent assessments by four large language models.
By Wenxin Duan, Hanwei Wang, Zhongying Peng, Zhonghua Lu, Jiayi An, Fan Song, Yong Liang
SeisEvo is a method that uses a large language model (LLM) and multi‑agent search to evolve seismic data reconstruction algorithms rather than optimize a single result. Starting from a classical algorithm, the agents modify only user‑opened components, rejecting candidates that violate physical constraints and scoring the rest by execution. The resulting white‑box algorithms—such as a residual‑gated, phase‑aligned dip‑consistency projection for interpolation and a reliability‑grouped singular‑value shrinkage for simultaneous interpolation and denoising—outperform classic methods by several decibels and generalize to unseen data.
By Yingjie Xu, Siwei Yu, Jianwei Ma
The paper introduces the Environment State-Text Injection (ESTI) attack, a novel method that manipulates the textual representation of environment states in large language model‑driven embodied agents without altering user instructions, model parameters, or executors. ESTI re‑frames adversarial goals as false state evidence that aligns with the current environment, thereby influencing both planning and execution through object properties, spatial relations, affordances, task‑stage rules, and execution feedback. The authors also present ESTI‑Bench, a benchmark that evaluates attack propagation across the planning‑to‑execution closed loop, and demonstrate that ESTI outperforms existing baselines on multiple embodied task datasets, achieving up to 89.32% higher planning‑level and 43.69% higher execution‑level attack success rates.
By Jiawei Liu, Jiacheng Guo, Tian Zhang, Yiwei Xu, Juan Wang, Jinlin Fan, Bowen Xiao, Chi Guo, Keyan Guo, Hongxin Hu