You can’t make a replica. Create a blank database and put your code in it. Make the code reference the names.nsf data and write your data whereever it needs to go. There is no requirement that the code that is doing an extract run from within the database it is extracting from. In fact, if you have a lot of jobs like this, you might want to control them all in a single database (helps with agent sequencing, if they are dependent on each other).