Viability-localized autopoiesis II: Context-indexed repair transport and empirical diagnostics

AP1-AP4 define viability-localized production closure, produced boundaries, admissibility-respecting realization, and existence of a globally coherent viable regime. They do not determine whether a repair relation calibrated in one biological context remains adequate after the context changes. This study develops that second-stage question without introducing a new definition of life or a canonical list of categorical operators. The central construction is a context-indexed family of adequate repair surrogates that separates three claims: local surrogate adequacy in each context, reuse of one selected reference relation across contexts, and existence of any shared context-blind relation over the tested context family. A standard overlap-gluing criterion is specialized to make the strongest incompatibility claim explicit. Two quantitative realizations are analyzed. In an Escherichia coli core flux-balance model, three environmental or reaction constraints remain globally feasible while their growth-capacity readouts differ strongly. In a literature-based repressilator model, all 17 tested contexts remain locally reconstructable with high held-out skill, whereas the relation calibrated at the reference condition has a finite threshold-dependent reuse region. The analysis uses strictly disjoint temporal training and evaluation windows, blocked model selection, persistence and state-support controls, and ten-seed sensitivity analysis. A separately fitted state-only relation remains accurate across the complete tested context family, demonstrating that failure of the selected relation does not imply absence of a shared relation. These calculations provide empirical diagnostics of context dependence while deliberately stopping short of claiming full AP1-AP4 validation or a boundary-specific early-warning law.