Marginal effects for SEM with binary variables in R

I want to interpret the coefficients of a SEM model with binary endogenous variables (estimator = WLSMV).