01.VeriHarm-Bench: Measuring Harmful Success in LLM Coding Agents Verifiably

86% worth reading · novelty 1.9/3 ·benchmark/eval ·cs.CL

02.MCP Tool Poisoning in Agentic AI Frameworks: A Cross-Framework Measurement Study

88% worth reading · novelty 1.8/3 ·benchmark/eval ·cs.AI

03.Strategist: Evolving Reasoning with Agents

85% worth reading · novelty 2.0/3 ·new method ·cs.AI

04.Cloudra: Geometric Reasoning and Scientific Validation for Autonomous Reaction Mechanism Exploration

86% worth reading · novelty 1.8/3 ·new method ·cs.MA

05.Coati: Host-Grounded Execution for Installable Mobile GUI Agents

84% worth reading · novelty 1.9/3 ·applied system ·cs.MA

06.WebHarbor: Docking Real Websites for Evolving GUI Agent Environments

80% worth reading · novelty 1.8/3 ·benchmark/eval ·cs.AI

07.FedReady: Bounded Agents for Federation Formation

78% worth reading · novelty 1.9/3 ·applied system ·cs.AI

08.Multi-Modal Multi-Agent System for Bridge Operation and Maintenance based on Large Language Model

80% worth reading · novelty 1.7/3 ·new method ·cs.MA

09.Can LLM Agents Be Forward-Deployed Engineers? Requirement Discovery and Delivery for Enterprise Text-to-SQL

81% worth reading · novelty 1.6/3 ·benchmark/eval ·cs.CL

10.From Natural Language to Exact Cover: A Neuro-Symbolic Approach to Zebra Puzzles

86% worth reading · novelty 1.1/3 ·new method ·cs.AI

11.Optimize the Harness, Not the Model: Fitting a Minimal Agent to a Self-Hostable LLM for Investigative Analysis

80% worth reading · novelty 1.5/3 ·new method ·cs.CL

12.DiabAgent: Toward Agentic Physiological Control

79% worth reading · novelty 1.6/3 ·applied system ·cs.AI

13.NitiBench+: A Temporal Thai Legal Question Answering Benchmark Across 17 Domains

77% worth reading · novelty 1.6/3 ·benchmark/eval ·cs.CL

14.Four Bridges: Measuring Deception Propensity in a Multi-Agent Game

75% worth reading · novelty 1.8/3 ·benchmark/eval ·cs.AI

15.SelfImprove-Bench: Evaluating Generalization in Self-Improving Agents

76% worth reading · novelty 1.5/3 ·benchmark/eval ·cs.MA

16.IdeaTaste: Do LLMs Share the Research Taste of Human Researchers?

61% worth reading · novelty 1.9/3 ·benchmark/eval ·cs.CL

17.D&A: Decoupling Description and Assessment Enables VLMs as Zero-Shot Robotic Reward Models

63% worth reading · novelty 1.8/3 ·new method ·cs.AI

18.Not All Reasoning Generalizes: A Case Study on Learning Generalizable Reasoning Strategies in RLVR

58% worth reading · novelty 1.8/3 ·benchmark/eval ·cs.LG

19.VIAS: Value-Isolating Activation Steering for Multi-Value Alignment and Compliance for Organisations

61% worth reading · novelty 1.3/3 ·new method ·cs.CL

20.Library of Layers (LoL): Adaptively Ordering Transformer Layers at Runtime

50% worth reading · novelty 2.0/3 ·new method ·cs.CL