JustAct: A Framework for Auditable Multi-Agent Systems Regulated by Inter-Organisational Policies

Jul 16, 2026

Upload by Logic

Open multi-agent systems operating across organizational boundaries require actions to be regulated by complex, context-dependent policies. The document focuses on distributed implementations where actors must justify permitted actions using policy information drawn from dynamic statements and agreements. It introduces the JustAct framework, proves key auditability properties, and shows that any permission decision cannot be refuted later after new statements or updated agreements. It further specifies and implements a concrete policy language and runtime, evaluating them through a medical data case study inspired by Brane.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages57

Reading Time2 h 24 min

Introduction

Policy-regulated open multi-agent systems

Challenges in incomplete policy knowledge

Key contributions of the JustAct framework

Auditability properties

Implementation and case study

arXiv :2502 .00 138v 3 [ cs .LO] 9 Jul 2026 JUSTACT: A FRAMEWORK FOR AUDITABLE MULTI-AGENT SYSTEMS REGULATED BY INTER-ORGANISATIONAL POLICIES ∗ CHRISTOPHER A. ESTERHUYSE a , TIM MÜLLER a , AND L. THOMAS VAN BINSBERGEN a Informatics Institute, University of Amsterdam, Amsterdam, The Netherlands e-mail address: [c.a.esterhuyse@uva.nl](c.a.esterhuyse@uva.nl), [t.muller@uva.nl](t.muller@uva.nl), [ltvanbinsbergen@acm.org](ltvanbinsbergen@acm.org) Abstract . In open multi-agent agent systems that cross organisational boundaries, agent actions must be regulated by complex policies. Consider medical data processing systems, which must observe generic laws (e.g., EU data protection regulations) and also specific participants’ resource conditions (e.g., Bob consents to sharing his X-Rays with EU hospitals) . Presently, we address the implementation of these systems as distributed software. Solutions to key sub-problems are available: existing policy languages capture the necessary normative concepts and formalise the computational representation and reasoning about policies, and existing distributed algorithms and protocols coordinate agents’ changing actions and policies. But which policies and protocols are useful in application? With the JustAct framework, we characterise a class of multi-agent systems where actors justify their actions with sufficient policy information collected from dynamic policy statements and agreements. We prove key properties of these systems, e.g. , any decision that an action is permitted now cannot be refuted later, regardless of any added statements or updated agreements. We study a particular instance of the framework by specifying (in Rocq) and implementing (in Rust) a particular policy language and runtime system for mediating agent communications. We demonstrate and assess JustAct via a case study of this implementation: we reproduce the usage scenarios of Brane, an existing policy-regulated, inter-domain, medical data processing system. 1. Introduction In this article, we consider multi-agent software systems that are open (i.e., agents may join and leave at runtime) and subject to regulation by policies: specifications of how agents are permitted to act. These systems are needed in practice whenever software systems cross organisational boundaries, as physical distance separates institutional entities from the users or software components that they aim to control. Such users, components, and entities as unified in agents able to act, and their dependencies are captured as policies. For example, to enable medical research on the rare DIPG disease, hospitals must share their medical records. But, to maintain control, participants are required to accept the data Key words and phrases: Decentralised, Framework, Composition, Coordination, Multi-agent System, Policy, Program Refinement, Specification. ∗ This article is a significantly extended version of [EMvB24] . We define the same framework more precisely and extensively. The prior implementation discussion is replaced: we present our new Rust implementation and Rocq formalisation of our framework, policy language, its interpreter, and runtime system. The prior case study is replaced by a new more extensive one: recreating the brane medical data processing system. Preprint submitted to Logical Methods in Computer Science © C. A. Esterhuyse, T. Müller, and L. T. van Binsbergen ⃝CC Creative Commons sharing agreement of the DIPG registry ecosystem [VvZBC+17] . In general, participants rely on the enforcement the policy, which aligns agents’ actions with what is permitted. For example, auditor agents trigger compensatory actions to punish actors of prohibited actions after the fact, or automated monitors prevent prohibited actions just in time. The terms in which policies define permission are dependent on the case. For example, when agents’ access to data is modelled as instantaneous events, policies are formulated in term

FAQs

What does the JustAct framework aim to solve in open multi-agent systems?

It addresses how to implement multi-agent systems where actors must justify their actions using sufficient policy information gathered from dynamic statements and agreements, ensuring auditability under regulation across organizational boundaries.

What core property is proven about permitted actions in JustAct?

Any decision that an action is permitted now cannot be refuted later, regardless of added statements or updated agreements.

How is JustAct evaluated in the document?

The framework is instantiated by specifying a policy language and implementing a runtime system for mediating agent communications, then assessed via a case study that reproduces scenarios of an existing policy-regulated inter-domain medical data processing system (inspired by Brane).

Jul 16, 2026

Upload by Logic