Plex Music Smart Playlists with Ollama LLM (n8n Workflow)

Automatically classify your Plex Music library and rebuild “smart” playlists using an Ollama LLM

This n8n workflow runs on a schedule to scan every track in your Plex Music library, sends batches of track metadata to a local Ollama LLM for playlist classification, then recreates and fills matching Plex playlists—while skipping spoken-word content.

What this workflow does

• Runs weekly and queries your Plex server for library sections and the server machine identifier.
• Selects your Plex Music library section and paginates through it to collect and normalize track metadata.
• Batch-sends tracks to Ollama (via /api/generate) to assign each track to one of your predefined playlist names or SKIP_NON_MUSIC.
• Parses and validates responses, accumulates an ID-to-playlist mapping across all batches, and excludes tracks marked SKIP_NON_MUSIC.
• Rebuilds playlists in Plex: for each playlist group, it fetches existing Plex playlists by name, deletes any matching playlist, creates a fresh playlist, seeds it with the first 50 tracks, then appends the remaining tracks in chunks of 50.

Use cases

• Create and refresh mood/genre-style playlists from your Plex Music library without manually sorting tracks.
• Exclude spoken-word tracks so your playlists stay music-focused.
• Maintain always-current playlist contents with a weekly automated job in n8n.
• Use a local LLM (Ollama) to control classification logic without sending your library to third-party AI APIs.

Technical details

• Plex integration: uses n8n HTTP Request nodes (requires an HTTP header credential including X-Plex-Token).
• LLM integration: calls Ollama at /api/generate to classify track batches.
• Core n8n nodes: if, set, code, sticky note, http request, and split in batches.

Setup notes: replace MY_N8N_HOST_OR_IP in all Plex HTTP Request nodes, ensure n8n can reach your Plex server, and confirm the target library contains a Music section (type artist) with track items.