Abstract: To investigate the crack evolution mechanism and precursor characteristics of rockburst, granite roadway rockburst simulation experiments were conducted using a true triaxial testing machine. The evolution of tensile and shear cracks was analyzed, showing that tensile cracks dominate in the early stage, whereas shear cracks prevail as failure approaches. Accordingly, a new monitoring index—the Tensile–Shear Ratio (TSR)—was proposed to quantify the dynamic evolution of both crack types. Based on the theory of critical slowing down, the autocorrelation coefficients and variances of acoustic emission (AE) parameters were examined. Results indicate that AE duration, peak frequency, amplitude, ringing count, rise time, RA value, RA/AF, and TSR all increased significantly prior to rockburst, exhibiting typical precursor behavior and underscoring the advantages of this theory for early warning. A precursor response coefficient was further defined to quantitatively compare precursor sensitivity. TSR showed the highest response coefficient, with precursors appearing earlier and demonstrating stronger timeliness in early warning, while traditional AE parameters exhibited comparable effectiveness. These findings provide a novel parameter and theoretical foundation for advancing rockburst monitoring and early warning.