I’ve had success with Shelfmark, but I don’t use .is.
I leave primary mirror to auto then have the following as mirrors:
I’m not sure I follow what you’re problem is regarding the indexers, so I can’t help there.
I’ve said this in a few other contexts too, but Grimmory/Booklore, while imperfect is loads easier and nicer to use than CW or CWA. You aren’t managing a calibre db + a separate web layer, and this meant I could tighten up my organization significantly since I wasn’t required to be on my host machine.
For me the primary benefits above the interface is the metadata control you have from the web UI. You can dump all those files into a single ingest folder and it will attempt to match metadata with the ability to apply all, or pick and choose as you see fit.
From what I’ve understood, the recent drama was mostly about an immature creator rather than rampant AI generated code.
Glad I could help! I also recommend looking into the IRC integration.