Hi Tilman,
Generally in the presence of heavy standards, Skyline scores the heavy standard and then just quantifies the signal on the light channel. Skyline has background subtraction, which should reduce the noise contribution considerably, but it can also be advisable to perform some assay development ahead of time to determine which transitions may have interference and therefore should be marked as not quantitative, which would keep them from contributing to the TotalArea or light:heavy ratio.
It is generally enough to prove you are integrating the right retention time range based on consistency of your detection of the across replicates, based on intensity and retention time.
If you really want a q value, you can look into using mProphet in Skyline, described in the Advanced Peak Picking tutorial, but this is now considered most appropriate for larger-scale methods like DIA, and usually without injected standards.
Hope this helps. Sorry for not getting back to you sooner.
--Brendan