Can a prompt replace Everand?

Build a personal library server over books you own, treating metadata as the difference between a folder and a library.

Run Kavita for ebooks and comics, and Audiobookshelf for audiobooks. Both are actively maintained, both have mobile clients, and both handle reading progress server-side.

The work worth doing is the ingest pipeline. Every incoming file gets identified against an open metadata source — Open Library for books — and stored with a stable identifier, series name and number, publication date and cover. A library where half the series have no ordering is one you stop browsing.

Normalise formats on import: convert to EPUB for anything the reader handles poorly, keep the original alongside, and record which is the source of truth. Never modify an original in place.

Build the deduplication pass, because personal libraries accumulate the same book three times from three sources. Match on identifier where present and on normalised title-plus-author where not, and present candidates for a human decision rather than deleting anything automatically.

Add a reading log: what you started, what you finished, what you abandoned and where. Abandonment data is the honest and most interesting part — most subscription reading is abandoned, and knowing your own rate changes what you start.

Wire in your public library. Most library systems expose their catalogue and your loans, and a page showing what you have borrowed alongside what you own is a genuinely better answer than either alone.

Do not strip DRM. Buy from DRM-free sellers, borrow from libraries, and use public-domain sources; the pipeline stays clean and so does the legal position.