You're correct that to collect an MS/MS spectrum (what PRM is doing), you only need to select a m/z for your precursor selection and the amount of energy that you want to supply. However, to extract a precursor > product ion chromatogram, you need to give it a precursor and product ion pair to extract. For peptides we can usually predict likely product ions but this is harder for non-peptide molecular species you will want to have data from either a standard or some prior measurement where you are confident the spectrum reflects the target molecule of interest.
I will also point out that it is usually best to start your method development in Skyline itself. It is designed to help with the method development and analysis. I would suggest looking through the tutorial on PRM https://skyline.ms/wiki/home/software/Skyline/page.view?name=tutorial_prm. Even though it is for peptides many of the aspects are the same.
You can also look at the small molecule targets tutorial (https://skyline.ms/wiki/home/software/Skyline/page.view?name=tutorial_small_molecule). The method is for a triple quad but the process is the same except in this case you need to tell the QQQ instrument what m/z values to measure for the product ions first. In PRM you collect the entire spectrum so you can collect all m/z values and extract the data after the fact.
There are some great webinars too:
https://skyline.ms/project/home/software/Skyline/events/2017 Webinars/Webinar 16/begin.view?
https://skyline.ms/project/home/software/Skyline/events/2018 Webinars/Webinar 17/begin.view?
Yes, you should be able to use gnps data to help pick product ions. However, just remember that the product ions will be dependent on the instrument platform and collision energy used.
Best of luck in your experiments. Hope this helps,
Mike