Plex MCP. Manage your whole library through conversation.

You connect this MCP Server to manage every damn thing on your Plex Media Server using just plain talk. It takes the nightmare of clicking through menus and turns it into a simple chat session with your media archive. Your AI client talks directly to the server, making library navigation automatic.

To start, you can check the overall health and connection status of the entire Plex system using get_server_info. You'll also list every distinct content section—whether that's 'Movies,' 'TV Shows,' or 'Music'—by calling get_libraries, giving you a map of everything configured.

Need to find something specific? The system lets you search your entire media collection using keywords, titles, or artists with search_library. You can also see what content was recently added to the server by running get_recently_added and pulling up all movies, episodes, or albums that just showed up. Once you know which library you're dealing with, you pull all its contents—filtered by type—using get_library_content, making sure your agent knows exactly what it’s looking at.

When you zero in on a piece of media, you get deep data dives. You run get_metadata to grab full details for any single item; that means getting the cast list, the director, file paths, genres, and summaries all in one shot. If the content is a TV series, you can expand your search using get_children, which lets you jump from the main show title directly into its seasons or even related albums.

For episode-specific details, you've got get_all_episodes, which pulls the complete list for any given TV series.

To keep things organized, you manage playlists with two tools: first, get_playlists counts and names every user-made playlist on the server; second, if you want to see what’s inside one of those lists, get_playlist_items returns a full list of titles and types contained within a specified ID. If your library structure gets messy—like after moving or adding a bunch of files manually—you gotta run refresh_library on a specific section so the server actually sees it.

Tracking who’s watching what is straightforward. You pull detailed information on every single streaming session using get_sessions, knowing exactly which user, device, and amount of progress they've hit. If you want to know what episode is next for any show currently being watched by users, get_on_deck identifies that specific unwatched episode.

When the viewing status needs changing—maybe someone paused it or finished it early—you programmatically update records: run mark_watched to confirm an item was viewed and take it off 'On Deck,' or use mark_unwatched to reset its status, making it appear available again.

This whole process keeps your AI client talking only in function calls. It manages everything from listing content libraries using get_libraries, to checking if the server is even up with get_server_info. You’re not limited to just searching; you can get a full inventory of playlists, pull specific items, and maintain detailed records on every single piece of media in your archive.