I think you have the right idea but I need it to work in reverse. The attribute values need to drive the length and width values. I think you would still use stretch parameters but grips wouldn't even be necessary. Not sure how this would work or even if it's possible.
It seems like we are on the same page but reading from opposite ends.
If you know the given length of the panel then if this remains constant in the formula then when the slope factor (radian) is applied (presumambly something less than one) the overall length would then be correct in plan for the given slope.