The competitive propagation of hydraulic fractures during multi-cluster stimulations poses a critical challenge to achieving effective reservoir stimulation in deep shale formations. To address this, our study integrates true triaxial physical simulations with an innovative multi-adjacent well collaborative distributed fiber-optic monitoring strategy. Applied to the deep Longmaxi shale, this approach facilitates full three-dimensional, dynamic tracking of strain evolution, thereby elucidating the mechanisms that govern fracture competition under in-situ stress conditions. The results delineate two distinct fracture architectures dictated by the prevailing stress regime. When the maximum horizontal stress dominates, a cross-shaped network of orthogonal transverse and bedding fractures develops. Conversely, under vertical stress dominance, the combined effects of stress shadowing and geological discontinuities promote the formation of step-like main fractures alongside backward-extending inclined branches. The synchronized interpretation of distributed strain-rate signals—characterized by distinctive patterns such as asymmetric heart-shaped and V-shaped fingerprints—and pump pressure curves elucidates the dynamic process of cluster interaction. This transforms fiber-optic sensing from a monitoring tool into a quantitative diagnostic platform for competitive fracturing. The findings establish a theoretical foundation for designing three-dimensional monitoring frameworks, optimizing cluster spacing, and enabling adaptive pumping strategies, directly supporting the efficient development of deep shale gas resources.