Left: analytic and simulated XSB radial profiles in MW-mass halos. The analytic XSB profile for a P_th-supported halo (purple solid line) truncates steeply relative to the Z24a observations (black/grey shaded and dotted line), similarly to the P_th-supported FIRE run (purple dotted–dashed line), whereas fiducial C-TNG (pink dashed line) can match the outer XSB shape and C-SIMBA (gold dashed line) vastly underpredicts the XSB. P_CR-supported analytic predictions (green solid, with variations in [Ė₄₀, v₁₀₀] labeled) can agree with the observed constraints, as also shown by a P_CR-dominated simulation (green dot-dashed). Right: total outer CGM soft L_X, for ∼L^* galaxies, here integrated between 0.5R_vir and 1R_vir. P_CR-dominated MW-mass simulations (green crosses) and empirical CR-IC models with varied AGN leptonic injection ϵ₋₃ (green diamonds; error bars showing 0.3 dex scatter on ﹩\dot{{M}_{* }}﹩) agree with the observed best-fit (black squares; bands showing stacked range in M_*), while in agreement with ﹩{\tilde{Y}}_{{\rm{200}}}﹩. P_th-supported analytic predictions (purple circles) underpredict/overpredict L_X at low/high M_*, C-TNG meets MW and M31 mass constraints but underpredicts at 2M31 mass, whereas C-SIMBA universally underpredicts L_X. Unilaterally, all P_th-supported models overpredict ﹩{\tilde{Y}}_{{\rm{200}}}﹩, irrespective of L_X predictions (see Figure 2) particularly at low/high masses, but less so at intermediate M_halo.