This book constitutes the refereed proceedings of the First International Workshop, SEA 2025, Held in Conjunction with NeurIPS 2025, in San Diego, CA, USA, during December 7, 2025.

The 20 full papers presented in this book were carefully selected and reviewed from 154 submissions. They focuses on benchmarks and evaluation; learning and fine-tuning through interaction; tool-use and software environments; multi-agent systems and simulation; embodiment and grounding; and sim-to-real deployment.

Inhalt

.- Examining the Vulnerability of Multi-Agent Medical Systems to Human Interventions for Clinical Reasoning.

.- Mobile-Agent-E: Self-Evolving Mobile Assistant for Complex Tasks.

.- Verifiable Chemical Reasoning through Tool-Calling Agentic Workflow.

.- Towards Agents That Know When They Don't Know: Uncertainty as a Control Signal for Structured Reasoning.

.- EVOLVE-MEM: A Self-Adaptive Hierarchical Memory Architecture for Next-Generation Agentic AI Systems.

.- Natural Language Grounded Reinforcement Learning for Clinical Decision-Making in Virtual Patient Simulations.

.- Evaluating Long-Term Memory and State Tracking in Multi-Platform Dynamic Agent Environments.

.- SEA: Stateful Execution Environment for Conversational Big Data Analytics.

.- When Developer Aid Becomes Security Debt: A Systematic Analysis of Insecure Behaviors in LLM Coding Agents.

.- Enabling User-Created Multi-Agent Simulations: Interactive and Customizable 2D Environments to Study Team Dynamics with LLM Agents.

.- SimuGen: Multi-modal Agentic Framework for Constructing Block Diagram-Based Simulation Models.

.- IndusGCC: A Data Benchmark and Evaluation Framework for GUI-Based General Computer Control in Industrial Automation.

.- DefenderBench: A Toolkit for Evaluating Language Agents in Cybersecurity Environments.

.- The Physical Basis of Prediction: World Model Formation in Neural Organoids via an LLM-Generated Curriculum.

.- Communicating Plans, Not Percepts: Scalable Multi-Agent Coordination with Embodied World Models.

.- Co-Evolving Complexity: An Adversarial Framework for Automatic MARL Curricula.

.- Learning to Make Friends: Coaching LLM Agents toward Emergent Social Ties.

.- When Agents go Astray: Course-Correcting SWE Agents with PRMs.

.- Model Context Protocol for Vision Systems: Audit, Security, and Protocol Extensions.

.- Agentic Persona Control and Task State Tracking for Realistic User Simulation in Interactive Scenarios.