for each agency, find a scrapped document and read the webserver off it