What does goodness of fit tell us with skewed data? 
"The goodness of fit test is a statistical hypothesis test to see how well sample data fit a distribution from a population with a normal distribution. Put differently, this test shows if your sample data represents the data you would expect to find in the actual population or if it is somehow skewed."

But if our population data is by itself skewed,is there any meaningful inference we can gather from it?
Is there any practical use to it for such data?
Source: Investopedia
 A: The general sentiment on Cross Validated is that goodness of fit testing is no more helpful than plotting the data; that is, formal inference with the p-value does not help. This is because large sample sizes are going to cause rejections when there are subtle differences that don’t matter to you, and small sample sizes will not give the test enough power to detect much of anything. 
However, sure, you can test goodness of fit for skewed distributions. The ks.test function in R will test for goodness of fit to normal, exponential, uniform...any distribution X for which there is a pX function. The documentation for ks.test discusses this: https://www.rdocumentation.org/packages/dgof/versions/1.2/topics/ks.test.
The quote you gave seems to imply that the test is for normality. Is the context, by any chance, the Shapiro-Wilk test?
A: Statistical software was one of the first computer applications widely available. So for a long time there was an approach to use flow-chart style decision rules like goodness of fit tests in quantitative fields like finance. If the test passed normality, "Then Do",
otherwise "Else Do." In recent years statisticians have questioned these types of approaches.  However I think goodness of fit tests are still of value for teaching about distributions.
