Temporal Bell Inequalities and Global Consistency in Modal Triplet Theory: A Conditional Sequential-Instrument Analysis

We give a typed analysis of Leggett--Garg inequalities in Modal Triplet Theory (MTT). Three distinctions are decisive. First, the laboratory measurement times t1, t2, and t3 are ordered by a supplied physical clock on a causal spacetime; they are not the auxiliary stabilization parameter used to construct a fixed point. Second, pairwise temporal correlations are generated by declared channels and sequential measurement instruments, including their state-update maps. Third, a Leggett--Garg violation excludes a single context-independent joint distribution for the measured pair protocols; it does not identify projection alone as the cause. We prove the standard Leggett--Garg bound from the common-joint-distribution contract. We prove conditional no-retro-signaling for a causal sequential instrument model and distinguish it from noninvasive measurability and no-signaling in time. We also prove a non-selection result: projection or global-history language without dynamics, instruments, and a probability source supports both Leggett--Garg-satisfying and violating realizations. Finally, an explicit qubit, unitary, and Lueders-instrument packet gives a value of three halves. This is a standard quantum operational realization. It becomes an MTT result only if a selected MTT source descends to that same packet. Thus global consistency remains a possible mechanism for temporal contextuality, but the present paper is a conditional reconstruction and completion contract, not a generic MTT prediction of temporal Bell violation.