This solution works great, thanks Dfay!
A minor problem (which may be unrelated with your Skimmer) is that the resulting PDF-files sometimes may contain wrong information on the page orientation.
I OCRed a large file made with PDFScanner, which recognized the text well but saved the searchable text with a different orientation in a few pages.
I had not experienced this problem with other scans, but I have not tried this with many files)
As a workaround (thanks to people at PDFScanner) you can:
Select the pages and run edit->convert to black and white before performing ocr. That will rerender the pages and therefore correct the orientation information.