Thank you for sending that Skyline document.
It looks like for each of those peptides, Skyline is thinking that there is no turning point.
That is, the calibration curve is sufficiently straight so that it is best modeled as a single straight line.

The behavior that you are seeing looks like a bug in that it is then setting the Limit of Detection as the highest calibration point.
I forget exactly what Skyline is supposed to be doing in this case (I will have to look at your Python code again) but I suspect that Skyline is not doing the right thing here.
-- Nick

Oops. That was just a bug. Skyline would get confused when the calibration curve had no turning point.

I will try to get this fixed in Skyline-daily soon.
I have made a special build of Skyline-daily which fixes this bug that you can install from here:
https://proteome.gs.washington.edu/~nicksh/SpecialSkylines/FiguresOfMerit/
-- Nick

Ok, glad this is an easy fix! Hopefully Daily can be updated soon, this has caused problems for a few people I know of already.

Thanks!