Nacre-inspired brick-and-mortar composites achieve high stiffness, toughness, and damage tolerance through microstructural design. In these architectures, mechanical performance is governed not only by macroscopic failure but also by the initiation and evolution of micro-damage, which precedes crack formation and strongly influences fatigue reliability. Because such micro-damage develops locally, it is difficult to detect and predict using conventional stress–strain (S-S) responses alone. Recent advances in structural health monitoring (SHM) have demonstrated that micro-damage is accompanied by coupled multi-physics signatures, including localized temperature rise, electrical resistance variation, and stress redistribution [1]. These thermo-electro-mechanical field responses often emerge well before macroscopic stiffness degradation or ultimate failure, highlighting the importance of multi-physics information for early damage assessment. Despite this complexity, nacre-inspired composite materials are widely employed across diverse applications. Their designs are tailored to application-specific mechanical requirements, which are effectively reflected in different target shapes of the S–S curve. Accordingly, this study adopts the S–S curve as a concise and physically meaningful design objective and proposes a conditional diffusion-based inverse design framework that simultaneously enforces consistency across relevant multi-physics domains [2]. To further ensure physical plausibility, a condition-based filtering metric is introduced by comparing the generated effective stress with the target S-S curve. Results demonstrate low reconstruction error with respect to the target S–S curve, strong agreement across all generated multi-physics fields, and accurate crack-path prediction. Overall, the proposed approach provides a scalable computational route for reliable multi-physics-aware inverse design of nacre-inspired composite materials.