If I'm understanding your architecture correctly, you could follow a process like this:

1. Put those public  domain documents  into a document library on her site.
2. Give them a unique content type. It doesn't need any metadata, just a CT that identifies them.
3. Ensure that everyone has access to that document library.
4. At the place where you want users to access those document, use a core search web  part with a fixed query along the lines of "ContentType:uniqueCT".

That will display those records from anywhere in the farm.

That's a pretty broad outline, but may meet the requirements.