I believe I understand what you want to do and feel it would be easier to do in a single script and that single script would be less demanding on your system then the two script approach.
In your two script approach all the image in the tree would come together into single document with layer groups for each sub directory which most likely have a name that it the full path of the sub directory for you never stated if the sub directories could have additional sub directories. The second script would then break out the layer groups and save each as a file it the correct folder.
It seems to me it be easier to build and array of the folder names in the tree. Then create the layered file you want in each of those folders by just using the load files into stack logic for the images files in each of the folders in that array.