arXiv:2605.24069v1 Announce Type: cross Abstract: The rise of tool-using Large Language Model (LLM) agents, standardized by protocols like the Model Context Protocol (MCP), has unlocked unprecedented autonomous execution capabilities for LLM Agents by integrating external open-domain knowledge and tools. However, this interoperability introduces a covert attack surface targeting the agent's cognitive planning layer. This paper systematically investigates Tool Description Poisoning (TDP), a novel semantic attack. In TDP, malicious instructions are not embedded in a tool's executable code, but rather covertly injected into its descriptive metadata, the very "manual" an agent relies on for secure planning and decision-making. To rigorously and systematically evaluate this emerging threat, we introduce the MCP-TDP Security Benchmark. This high-fidelity sandbox environment comprises 32 realistic, real-world test cases spanning 6 distinct risk categories. Our evaluation of 8 mainstream LLMs reveals severe vulnerabilities, with leading models like GPT-4o exhibiting a nearly 100% Attack Success Rate (ASR) in six high-risk scenarios. Furthermore, our findings demonstrate that common prompt-guardrail defenses are largely ineffective and can, counterintuitively, even be counterproductive (a phenomenon which we term the "Firewall Fallacy"). Crucially, we also propose a defense mechanism: "Reactive Self-Correction," where an agent autonomously detects and reverts its own malicious actions post-execution. This work provides the first specialized security benchmark tailored for TDP, offering essential insights for securing the cognitive and planning layers of advanced agentic systems.

Background: Evidence-based physician-pharmacist collaborative clinics have demonstrated significant short-term benefits for patients with type 2 diabetes (T2D), but their long-term effectiveness remains unclear, especially in primary health care settings. Objective: This study aimed to explore the long-term effectiveness and cost-effectiveness of a novel, digital-driven, multifaceted physician-pharmacist collaborative model for managing patients with T2D in underresourced settings. Methods: We conducted a 12-month cluster randomized controlled trial from May 2021 to December 2022 across 6 primary health care settings in China. Guided by the theory of planned behavior, the intervention involved routine therapy from physicians along with pharmaceutical interventions from pharmacists. These were delivered through a combination of face-to-face visits and mobile health care. The intervention group received 4 face-to-face visits and biweekly remote education sessions over the 12 months. We conducted intention-to-treat analyses to estimate differences in clinical and behavior indicators between the intervention and control groups. Primary outcomes included glycosylated hemoglobin and 10-year atherosclerotic cardiovascular risk. Data were analyzed using adjusted generalized estimation equations. Results: This study included 574 patients (291 in the intervention group and 283 in the control group). Over 12 months, patients in the intervention group had significant reductions in hemoglobin A_1c (–2.57 vs –1.96, respectively; P<.001; 95% CI –1.027 to –0.238) and 10-year atherosclerotic cardiovascular risk (–1.35 vs 0.01, respectively; P<.001; 95% CI –1.690 to –0.630) compared with the control group. Substantial improvements were also observed in several secondary outcomes, including fasting blood glucose, 2-hour postprandial blood glucose, waist circumference, waist-to-hip ratio, blood pressure, triglyceride, and total cholesterol. Total diabetes-related costs decreased, and patient satisfaction improved significantly in the intervention group. There were no significant differences in BMI, high-density lipoprotein, or low-density lipoprotein. Conclusions: These findings suggest that the physician-pharmacist collaborative model could improve the long-term quality and efficiency of T2D management and reduce medical costs in underresourced areas globally. Patients with T2D, especially those with central obesity or high cardiovascular risk, may benefit more from collaborative clinics. Trial Registration: Chinese Clinical Trial Registry ChiCTR2000031839; https://www.chictr.org.cn/showproj.html?proj=51910

arXiv:2603.07020v1 Announce Type: cross Abstract: Neural approaches to the Flexible Job Shop Scheduling Problem (FJSP), particularly those based on deep reinforcement learning (DRL), have gained growing attention in recent years. However, existing methods rely on complex feature-engineered state representations (i.e., often requiring more than 20 handcrafted features) and graph-biased neural architectures. To reduce modeling complexity and advance a more generalizable framework for FJSP, we introduce \textsc{ReSched}, a minimalist DRL framework that rethinks both the scheduling formulation and model design. First, by revisiting the Markov Decision Process (MDP) formulation of FJSP, we condense the state space to just four essential features, eliminating historical dependencies through a subproblem-based perspective. Second, we employ Transformer blocks with dot-product attention, augmented by three lightweight but effective architectural modifications tailored to scheduling tasks. Extensive experiments show that \textsc{ReSched} outperforms classical dispatching rules and state-of-the-art DRL methods on FJSP. Moreover, \textsc{ReSched} also generalizes well to the Job Shop Scheduling Problem (JSSP) and the Flexible Flow Shop Scheduling Problem (FFSP), achieving competitive performance against neural baselines specifically designed for these variants.

arXiv:2603.03378v1 Announce Type: cross Abstract: Large language model (LLM) agents offer a promising data-driven approach to automating Site Reliability Engineering (SRE), yet their enterprise deployment is constrained by three challenges: restricted access to proprietary data, unsafe action execution under permission-governed environments, and the inability of closed systems to improve from failures. We present AOI (Autonomous Operations Intelligence), a trainable multi-agent framework formulating automated operations as a structured trajectory learning problem under security constraints. Our approach integrates three key components. First, a trainable diagnostic system applies Group Relative Policy Optimization (GRPO) to distill expert-level knowledge into locally deployed open-source models, enabling preference-based learning without exposing sensitive data. Second, a read-write separated execution architecture decomposes operational trajectories into observation, reasoning, and action phases, allowing safe learning while preventing unauthorized state mutation. Third, a Failure Trajectory Closed-Loop Evolver mines unsuccessful trajectories and converts them into corrective supervision signals, enabling continual data augmentation. Evaluated on the AIOpsLab benchmark, our contributions yield cumulative gains. (1) The AOI runtime alone achieves 66.3% best@5 success on all 86 tasks, outperforming the prior state-of-the-art (41.9%) by 24.4 points. (2) Adding Observer GRPO training, a locally deployed 14B model reaches 42.9% avg@1 on 63 held-out tasks with unseen fault types, surpassing Claude Sonnet 4.5. (3) The Evolver converts 37 failed trajectories into diagnostic guidance, improving end-to-end avg@5 by 4.8 points while reducing variance by 35%.