JAQ — Joint Action Quality Specification v1.0 (A Pre-Action Legitimacy Metric for Human–AI Collaboration)

JAQ — Joint Action Quality Specification v1.0(A Pre-Action Legitimacy Metric for Human–AI Collaboration) This report defines JAQ (Joint Action Quality) as a canonical governance metric for evaluating the legitimacy of human–AI collaboration at the precise moment a consequential action is authorized and executed. Within Pre-Inference Governance architectures, technical correctness alone is insufficient to establish legitimacy. JAQ evaluates whether a decision emerged from a responsible, comprehensible, and controllable joint action in which human agency, accountability, and intervention capacity were preserved. JAQ operates as a pre-action integrity filter and veto trigger inside ASiSO authorization gates. It does not measure model performance; it evaluates the legitimacy quality of the joint human–machine action relationship itself. Architecture position (canonical order):ASOSE → A7SEM → Semantic Routing → JAQ → ASiSO → Action License: CC BY-NC-ND 4.0Conditions of Access: Reading and citation permitted. No operational or commercial rights without explicit written license from the author.