like this ?
library(truncnorm)
library(fabricatr)
set.seed(42)
(r1 <- sort(round(rtruncnorm(n=100,a=1,b=7,mean=4,sd=2))))
hist(r1)
r2 <- correlate(given = r1, rho = 0.8,
draw_count, mean = 3)
r3 <- pmax.int(pmin.int(r2,7),1)
hist(r3)
cor(r1,r2)
cor(r1,r3)
fixing r2 to r3 in the way I do is crude, and does drop the correlation slightly as would be expected, but is it not close enough to work with ?