I have not heard of
specification hours. It looks like you are supposed to calculate the confidence interval with 95% confidence level.
The Central Limit Theorem (which the general rule of thumb says is ok to invoke for

) tells us that

is approximately normally distributed with mean 0 and std dev 1.
So then the confidence interval is given by

where

is defined by
=1-\alpha)
(
Z is from the standard normal distribution). I'm assuming that you have already seen where these formulas came from.
So from a standard normal distribution table

. So just plug away.