Multiagent Systems Papers of the Week

Such capabilities pose novel risks, as malicious agents could direct others towards harmful actions. In this work we investigate three such capabilities - persuasion, deception, and coercion - across five realistic evaluation environments and a variety of agents instantiated with frontier language models.

We present \textbf{CONFIRM}, a framework that separates running an analysis from deciding which findings to report.

Extending them to joint quality--cost optimization introduces a challenge: applying the deployment preference during intermediate updates can prematurely reject candidates that are valuable for later improvements. We introduce ParetoSaddler, which addresses this mismatch by separating intermediate candidate promotion from final deployment selection.

We describe the skill format, design principles and a method for translating research into practical workflows.

We study both problems together in a simulated company Slack workspace in which four employees delegate sprint staffing to personal agents.

We introduce Kradle, a platform for evaluating AI models, and Four Bridges, an information asymmetry game in which four agents distribute across four rooms: one is the death room, the others hold food worth more if not shared.

Translating population-level exercise guidance into individualized prescriptions requires integrating clinical and wearable data, including diagnoses, medications, functional limitations, and baseline activity, with condition-specific recommendations, a process difficult to scale beyond specialist rehabilitation settings. We developed FITTER, a multi-agent framework that integrates these data with condition-specific guidance to generate personalized FITT-VP prescriptions and applies deterministic safety audit and repair to correct rule-checkable violations.

Here, we introduce El Agente Fármaco, an AI scientist that uses modern large language model (LLM) technology to turn discovery goals expressed in everyday language into complex, scalable computational screening workflows.

We introduce germline execution, which limits information disclosure without inspecting or trusting tool implementations.

We report a preregistered study with synthetic, exactly scorable tasks comparing free-text, structured JSON and signed requirement-manifest handoffs: the orchestrator extracts each requirement as a verbatim quote plus a checkable predicate, code signs the manifest with HMAC-SHA256, and deterministic code verifies the merger's decision.

To address the god-agent problem, this paper proposes a fine-grained modular architectural pattern for implementing agents and MASs that systematically separates functional concerns into independently managed aspect entities while preserving the logical identity of agents within a standalone environment.

We study this setting through n-agent ad hoc teamwork (NAHT), a recent framework for learning in open multi-agent environments. We propose an axiomatic framework that derives value function learning from the Shapley axioms (Efficiency, Symmetry and Linearity), treating them as structural constraints rather than heuristic design choices.

Building upon a comprehensive consideration of the multidimensional factors influencing individual safety cognition and behavioral decisions, this study aims to explore the safety cognitive process for behavioral decision-making and subject interaction within construction crews.

Existing approaches either process multiple pages jointly or iteratively refine evidence within a single trajectory, risking cross-page context pollution and, for interactive methods, substantial sequential overhead. We introduce \textbf{AttriSwarm}, an agent-swarm framework that assigns these concerns to separate roles: a leader handles global cross-page reasoning, while subagents perform page-wise attribution in isolation and in parallel.

Instead, we introduce Agentic World Modeling (AWM), which reformulates next-observation prediction as a multi-turn evidence-acquisition task.

We study both functions by varying role instructions while holding agents, topology, call budget, and ground-truth evaluator constant, using three open-weight models on Omni-MATH and two on SciBench.

We propose Value-Isolating Activation Steering (VIAS), which contrasts responses to the same scenario under two prompts, one listing all relevant values and one omitting a single target value.

However, under bandwidth-constrained scenarios, UAV networks still face significant challenges in task execution efficiency and intelligent processing capabilities. To address these issues, we propose Multi-UAV Cooperative Deep Semantic Autoencoders (MC-DSA) for communication networks.