In the admin panel, go to the Options > Scriblio tab and click on the Import link for the III Web Importer (in a section labeled Harvest near the bottom of the screen).
The four form fields aren’t labeled (yet), but they do have example inputs. In order, they’re (1) the hostname of your III system, (2) the starting bib number to import (or only bibn if you’re doing just one), (3) the ending bibn, and (4) a prefix to help identify the source of these records from inside Scriblio.
Pick a range and click Go. The records don’t go immediately into Scrib, but into a “harvest table” (importing the catalog to the live DB can really kill performance). So once that’s done, click the Import Harvested Data link in the Commands section. You’ll see some feedback about the process and you’ll then be able to view and search those records. (Note: the import harvest code loads records one-by-one and is ripe for some optimization.)
About that prefix: Scrib can have any number of different catalogs loaded (currently limited only by the need for a alphanumeric prefix, so you can have up to 1,296 datasources/catalogs/etc). But to show current availability, prevent duplicates, and other things, Scrib needs to be able to identify the specific record in the specific system that it came from. The specifics of this will evolve, I’m sure, but for now Scrib tracks the “source_id” of each record, which for a III system is the bibn with that two-letter prefix attached to it.
Example: you’ve got two III systems you’re importing records from, plus the contents of your A to Z list (I don’t have an importer for that written, but it seems useful). There’s nothing to prevent the numbering systems in those three sources from overlapping, and it might be likely on the III systems. So, when importing the records you assign “bb” to the first III system, “gb” to the second, and “az” to the a to z list.
http://about.scriblio.net/download#comment-186
http://about.scriblio.net/download#comment-190