ApexFeeder Docs

ApexFeeder is a lightweight media file server for local and cloud libraries. Point it at a folder of movies or TV shows: it reads the filenames, looks each title up on The Movie Database, downloads posters and descriptions, and serves the result through a small web UI you can play from — and to ApexPlayer on your Apple devices.

It is deliberately much smaller than Plex, Jellyfin or Emby. No transcoding, no plugin system, no accounts. Find your media, describe it properly, stream it.

Node.js

Version 22.6 or newer.

ffmpeg / ffprobe

Optional. Used for durations, codec details and reading embedded subtitle tracks.

npm install
npm run build
npm start

Then open http://127.0.0.1:8770 and add a folder. There is nothing to configure — released builds ship with ApexFeeder's own TMDB key.

A library is a folder ApexFeeder watches. It can live on the machine or in a connected cloud account, and you can add as many as you like — everything merges into one catalogue.

1. 1 Open the web UI and go to Libraries → Add.
2. 2 Pick a local path with the folder picker, or choose a connected cloud account.
3. 3 Say whether it holds movies or TV.
4. 4 ApexFeeder scans it, identifies what it finds and fetches artwork. Anything uncertain lands in Review.

Dropbox, Google Drive and OneDrive are all optional, and each needs credentials compiled into the build you are running. Connect an account under Settings, then add a library pointing into it.

Day to day, Dropbox and OneDrive are the better experience of the three: both push changes promptly and both hand out temporary links so playback streams straight from the provider.

ApexFeeder parses the filename, then looks the result up on TMDB. It copes with the conventions people actually use — a plain title, a title with a year, a scene release name, Season folders, SxxExx numbering, multi-episode files and dated daily shows.

Movies/
    Pulp Fiction (1994).mkv
    The.Matrix.1999.1080p.BluRay.x264-GROUP.mkv

TV/
    The Wire/
        Season 01/
            S01E03.mkv

When ApexFeeder is confident, it matches silently. When it is not — two films share a title, or the year does not line up — the file goes to the Review tab with its best guess attached, so confirming is one click. Once confirmed, that decision is locked and no future scan will second-guess it, even if you rename the file.

• Every file is keyed on an identity that survives renames: an inode locally, a permanent file id in each cloud provider.
• Reorganising a library updates a path rather than losing the item, so matches, artwork and watch state survive.
• Filesystems that report no usable inode — exFAT, some network mounts — fall back to path-based identity and simply lose rename tracking.
• Nothing is ever written to your media folders.

ApexFeeder does the part only a server can do well and leaves the rest to the player. It never searches subtitle providers and never adjusts timing — that is ApexPlayer's job.

• Sidecar files next to a local video — Movie.en.srt, Movie.fr.forced.srt — are found during a scan.
• Embedded tracks are read with ffprobe and listed.
• A subtitle ApexPlayer finds elsewhere can be pushed back to ApexFeeder and cached, so every other device gets it without a second download.
• Cached subtitles go in ApexFeeder's data directory, never into your media folder or cloud drive.

ApexFeeder binds to loopback by default. To let ApexPlayer reach it from a phone or TV, bind to the network:

APEXFEEDER_HOST=0.0.0.0 npm start

Nothing is exposed by doing that. Every endpoint except the health check and pairing requires a paired device, and the admin interface is served only to the machine ApexFeeder runs on.

1. 1 Open ApexPlayer and pick this server. It shows a four-digit code.
2. 2 In ApexFeeder, go to Settings — the request appears with the device's name and address.
3. 3 Check the code matches, then Approve.
4. 4 The device can now browse and stream. Revoke it any time from the same screen; access stops immediately.

The four-digit code is not a credential — it exists so you can tell one pending request from another. Two things do the real work: the access token is handed only to whoever presents the matching 256-bit pairing secret, held by the requesting device alone, and only the local machine can approve a request. Someone else on your network cannot approve their own, whether or not they know the code.

A paired device can read the catalogue and stream. It cannot browse your filesystem, add libraries, change settings or approve other devices.

• ApexFeeder announces itself over mDNS/Bonjour as _apexfeeder._tcp. The announcement is link-local — routers do not forward it — and carries only a server id, name and version.
• A second path answers a plain UDP broadcast of "who is ApexFeeder?" on port 27359, for clients that speak that convention instead. It deliberately does not answer Jellyfin's or Plex's query strings, so it never appears as a false match in their clients.
• Both are skipped when bound to loopback. APEXFEEDER_ANNOUNCE=false turns them off; =true forces them on.
• Port 8770 is tried first so a bookmark keeps working. If it is taken, ApexFeeder moves to the next free port and logs where it landed.
• The address it settled on is written to runtime.json in the data directory, which is how a launcher finds the UI without assuming a port.
• Only one instance may use a data directory at a time; a record left by a crash is detected and reclaimed.

Everything is optional; the defaults are meant to just work.

ApexFeeder ships with the project's key, so you do not need one. You can still supply your own under Settings → Metadata source, and it takes priority over the shared one. Worth doing if you have a very large library, or if matching starts failing — the project key is shared by every install, so it shares one rate limit. Either a v3 API key or a v4 read access token works; ApexFeeder detects which you pasted.