Librarian vs tinyMediaManager

Two different philosophies about where metadata lives

The interesting difference here is not features. It is what each tool considers authoritative.

tinyMediaManager writes to disk. It scrapes from multiple providers and puts the result in NFO files next to your media, with artwork alongside. The filesystem is the record. Switch from Jellyfin to Plex to Emby and your metadata comes with you, because it was never in the server to begin with.

Librarian treats Jellyfin as the record. It reads what Jellyfin knows, changes files on disk when it needs to, and asks Jellyfin to rescan. There is no second metadata store and no sidecar file, which means nothing to keep in sync — and also nothing to take with you if you leave Jellyfin.

Neither is wrong. They suit different people, and which one you are depends on how likely you think you are to change media servers.

What Librarian does that tinyMediaManager does not

Import. This is the clearest split. tinyMediaManager works on media that is already where it belongs; getting it there is someone else’s job. Librarian watches your downloads root, recognises finished items, and moves them into the media tree with names Jellyfin parses cleanly.

It does that by hardlinking, which is why it is fast. A hardlink is a second name for the same bytes, so importing a 40 GB remux is instant, costs nothing extra on disk, and the torrent keeps seeding from the original path.

The constraint that follows is the thing most people get wrong. A hardlink cannot cross a filesystem, and — less obviously — it cannot cross a mount point either, even when both sides are the same filesystem. So this fails:

    volumes:
      - ./downloads:/downloads    # separate mount
      - ./media:/media            # links across these will not work

and this works:

    volumes:
      - ./library:/library        # one mount containing downloads/ and media/

Librarian decides by attempting the link rather than comparing device IDs, because nothing that inspects statSync().dev predicts mount boundaries correctly. If the attempt fails, the import becomes a verified copy — correct, just slower and twice the space.

A web interface that is always on. tinyMediaManager is a desktop application you open when you remember to. Librarian runs continuously as a container and is reachable from a phone.

What tinyMediaManager does that Librarian does not

Serious scraping. Multiple providers, per-field control over what gets written, manual match correction for the titles that scrape badly. Librarian’s enrichment is shallower.

Renaming patterns. tinyMediaManager gives you fine-grained control over the naming scheme. Librarian queues bulk renames toward Jellyfin-friendly names but does not expose the same depth of configuration.

Portable metadata. NFO files travel. If there is any chance you move to Plex or Emby, that is worth real money in re-scraping time you never have to spend.

Maturity. Many years of releases against every strange filename and every broken metadata provider. Librarian is on v1.1.0.

Neither of these replaces the *arr stack

Worth stating because it comes up constantly.

Radarr and Sonarr solve acquisition: watch for a release, search indexers, hand the result to a download client. Both tools here start after that, at the point where files exist.

If you already run the *arr stack, Librarian is not a replacement and adding it does not mean removing anything. It sits at the organise-and-clean end of the same pipeline.

If you do not run the *arr stack — you acquire by hand and only want the organising half — Tide plus Librarian covers that ground without Radarr, Sonarr or Prowlarr anywhere in it.

A note on safety

Librarian can move and delete files, and it is the only Coral module that defaults to requiring a Jellyfin sign-in for exactly that reason. The read-only modules do not.

The filesystem roots you seed through LIBRARIAN_DOWNLOADS_DIR and LIBRARIAN_MEDIA_DIR arrive switched off. A human has to enable each one in the UI on first run. That is deliberate: mounting a directory into a container is not the same as granting permission to reorganise it.

Full setup is on the Librarian module page, or see the Jellyfin Docker Compose stack guide for the mount layout that makes hardlinking work across Jellyfin, Tide and Librarian.