Had a quick look at the data that you have posted. I noticed that there are no zeros in Effort so hurdle and zero-inflation won’t “work” - they’ only fit your prior. Maybe I got something wrong here.
More importantly though, your predictor matrix is not full rank. Removing Deleted (hehe) should make things work more smoothly. See below for a simple example on how to spot stuff like that.
Another problem are the weird scales that your predictors are on. The estimated coefficients are really small if you fit something with a log-link. Using a QR transformation should help - not sure if that’s in brms.
This
library(rstanarm)
fit <- stan_glm.nb(Effort ~ Input + Output + Enquiry + File + Interface + Added + Changed, data = data, QR = TRUE)
worked fine.
Hope that helps! :)