The Judgement Boundary Framework (JBF) and Associated Technical Layers for AI Agency Control

System Overview: This archival record contains the core technical specifications for the Judgement Boundary Framework (JBF), the Material Mark & Signature Interpretation Layer (MMSIL), and their supporting operational layers. Together, these frameworks define a constrained, diagnostic-only role for artificial intelligence within large-scale human systems. Supplementary Material This version includes an executable Prolog reference implementation of the Judgement Boundary Framework (JBF) diagnostic core. The implementation operationalises the framework’s permitted diagnostic functions (stress gradients, accelerations, and audit signals) and explicitly prohibits automated judgement or intervention, consistent with the Responsibility Boundary defined in the paper. Core Objective: To provide a structural "stress sensor" that detects when the scale and complexity of a system are displacing human judgement. This architecture is designed to prevent "legitimacy collapse" by explicitly forbidding AI from crossing into governance, enforcement, or automated decision-making. Contents: JBF Boundary Core: Defines the five indicators of structural stress (Rule-to-discretion ratio, Explanation density, Enforcement-trust divergence, Optimisation bias, and Comprehension span collapse). MMSIL: A technical layer governing the validation of material marks and signatures through physical convergence rather than simple pattern recognition. Universal Base Profile (Silent Mode): Establishes the default operating stance of neutral, non-persuasive, and non-authoritative output. Adjacent Interaction Layer (AIL): Preserves "thinking-in-progress" by distinguishing between orienting thoughts and committed actions. Conditions of Use: Non-Governing Constraint: This framework must not be used to automate decisions, initiate enforcement, or optimize human behavior. Preservation of Agency: AI is limited to detecting structural pressure; interpretation and final authority must remain human. Refusal Logic: If diagnostic convergence is insufficient, the system must remain silent. Refusal to identify is treated as the correct expert outcome. License: This work is shared under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International (CC BY-NC-ND 4.0) license to protect against commercial exploitation and unauthorized derivative governance layers.