Terminal-reward rate tilting for record-conditioned hybrid classical-quantum filtering with openness diagnostics

We formulate a record-conditioned hybrid classical-quantum model in which the Q-state is the filtered quantum state conditioned on an accumulating classical record, and the C-state is the realized record path together with its consistent filter. The baseline is standard quantum measurement theory: quantum instruments in discrete time and stochastic master equations (SMEs) in continuous time; refolding is coarse-graining (discarding records), not an inverse map. On top of this conservative baseline, we introduce (i) a goal-scene macro layer that separates coarse outcomes from short-horizon execution segments, (ii) openness diagnostics that split uncertainty into goal entropy and conditional scene entropy with hysteretic Normal/Open regimes, and (iii) an innovation-anomaly jump channel driven by a finite-resolution innovation statistic. Our main methodological contribution is terminal-reward rate tilting (TPST-RDP): a generator-level exponential tilt (Doob h-transform / entropic tilting) of the SME-conditioned macro jump process that biases the macro-trajectory law toward a terminal objective while remaining absolutely continuous with respect to the baseline law. We derive the tilted intensities, give a monotonicity corollary, and provide record-level signatures and numerical demonstrations. TPST-RDP introduces no new quantum postulates and implies no retrocausal influence; it is a causal change-of-measure/control construction on macro records.