Yeah we really just need git object discovery via Nostr/Blossom. That's fairly simple, since all git objects are uniquely identified by a hash. So, on some level, all we need is a table of git object IDs and where to find them.