A Plex library that started at a couple of terabytes usually crosses ten in a year or two, and most of that growth is not new media. It is duplicate rips, wrong matches, artwork churn, and files no one ever plays. Reclaiming disk and fixing metadata by hand takes days. These seven desktop tools do it in an evening, and one of them will save you the terabytes you were about to spend on a new drive.
What to look for in a Plex cleanup tool
Pick based on what the library actually needs:
- Do you need dedup across identical rips, or just prune-by-play-count?
- Does the tool talk to Plex directly, or does it just walk the filesystem?
- Can it fix wrong matches without breaking watch history?
- Does it write to the media files themselves (tags, subtitles) or only to Plex?
- Is it a one-shot or something you run on a schedule?
Quick comparison
| App | Best for | Free plan | Paid | Platforms | Standout |
|---|---|---|---|---|---|
| Tautulli | Play statistics and watch history | Yes | Free | Windows, macOS, Linux, Docker | Answers “what should I delete” with real data |
| Plex Meta Manager | Fixing wrong matches and building collections | Yes | Free | Windows, macOS, Linux, Docker | Config-file-driven metadata rewrites |
| Sonarr | Automated TV cleanup and upgrades | Yes | Free | Windows, macOS, Linux, Docker | Replaces low-quality rips with better ones |
| Radarr | Same for movies | Yes | Free | Windows, macOS, Linux, Docker | Cross-references TMDB and IMDb |
| dupeGuru | Filesystem-level duplicate finder | Yes | Free | Windows, macOS, Linux | Fuzzy audio and video match on hash |
| PlexCleaner | Codec, subtitle, and metadata normalization | Yes | Free | Windows, macOS, Linux | Repacks files without re-encoding |
| Claude Code | AI-assisted analysis and scripted fixes | Free with API | Anthropic subscription tiers | Windows, macOS, Linux | Writes the pruning script for your specific library |
The apps
1. Tautulli — best for deciding what to prune first
Tautulli monitors a Plex server and reports on every play, every user, and every unwatched file. Sort by “never played” and you find the movies you ripped in 2019 and forgot. Sort by size divided by play count, and you find the four-hour Blu-ray rip of a movie you watched once.
Where it falls short: Reports only. You still make the delete decision. Setup requires giving Tautulli the Plex token, which some setups guard tightly.
Pricing: Free, open source.
Platforms: Windows, macOS, Linux, Docker.
Download: Tautulli
Bottom line: Run this before any cleanup so decisions rest on real play data.
2. Plex Meta Manager — best for wrong matches and metadata
Plex Meta Manager applies a YAML config to a Plex library: rename items, fix wrong matches, build smart collections, and set consistent artwork. When Plex has stubbornly matched Ford v Ferrari to the 1966 documentary for six months, PMM fixes it in one run.
Where it falls short: YAML learning curve. Some fixes require pinning to a specific external ID (TMDB, IMDb), which is not obvious on first read.
Pricing: Free, open source.
Platforms: Windows, macOS, Linux, Docker.
Download: Plex Meta Manager
Bottom line: The tool for anyone tired of clicking “Fix Match” one item at a time.
3. Sonarr — best automated TV cleanup and upgrade
Sonarr tracks TV shows and quietly upgrades to better rips when they become available. It also removes duplicate episode files, honors quality profiles, and can be configured to auto-delete unmonitored seasons. Pair it with Tautulli and you get “auto-remove any show unwatched for a year.”
Where it falls short: Full automation needs an indexer subscription (Torznab or Usenet). Wrong-quality upgrades happen if profiles are loose.
Pricing: Free, open source.
Platforms: Windows, macOS, Linux, Docker.
Download: Sonarr
Bottom line: Set once, and the TV library maintains itself.
4. Radarr — best for the movie side of the same problem
Radarr is Sonarr for movies. Quality profiles auto-swap a 4GB 1080p rip for a 15GB 4K remux when one appears. It also flags duplicates across your library (two Radarr instances pointed at the same folder is the standard trick).
Where it falls short: Same indexer requirement as Sonarr. TMDB integration occasionally lags for niche films.
Pricing: Free, open source.
Platforms: Windows, macOS, Linux, Docker.
Download: Radarr
Bottom line: The other half of an arr stack. Do not run one without the other.
5. dupeGuru — best raw filesystem dedup
dupeGuru walks any folder and finds exact and fuzzy duplicates across audio, video, and pictures. For a Plex library, it catches the same movie stored under two different names in two different folders, which happens more often than anyone likes to admit.
Where it falls short: Filesystem-only. Does not update Plex, so a dedup run needs a follow-up Plex library scan.
Pricing: Free, open source.
Platforms: Windows, macOS, Linux.
Download: dupeGuru
Bottom line: Run once every few months. The first run will surprise you.
6. PlexCleaner — best for codec and subtitle normalization
PlexCleaner normalizes files without re-encoding: repacks to MKV, strips foreign audio tracks, extracts and cleans SRT subtitles, and fixes broken chapter tables. That trims file sizes and stops Plex from transcoding on the fly, which is a real disk and CPU win.
Where it falls short: Configuration is dense. Some rare codecs still need a one-time re-encode. On low-power servers, batch runs take hours.
Pricing: Free, open source.
Platforms: Windows, macOS, Linux (via Docker).
Download: PlexCleaner on GitHub
Bottom line: The right tool to run once before you invest in a beefier transcoder.
7. Claude Code — best for AI-assisted, library-specific pruning scripts
Claude Code is not a Plex tool out of the box, but pointed at a Tautulli export or a filesystem walk it writes bespoke pruning scripts for your specific library. Point it at “here are files, here is play data, here is disk pressure” and it produces a targeted script that trims exactly what you should trim.
Where it falls short: Requires you to run it, review the script, and trust it. Not a one-click experience. Costs API tokens.
Pricing: Free CLI. Anthropic API is metered; subscription tiers cover heavy use.
Platforms: Windows, macOS, Linux.
Download: Claude Code
Bottom line: The right final pass when the standard tools cannot handle your library’s weirdness.
How to pick the right one
- If you want data before you delete anything: Tautulli.
- If your library is full of wrong matches: Plex Meta Manager.
- If you want automated upgrades and cleanup: Sonarr and Radarr as a pair.
- If duplicates are the main issue: dupeGuru.
- If Plex keeps transcoding: PlexCleaner to normalize codecs.
- If your library is weird and needs a bespoke fix: Claude Code, pointed at your data.
FAQ
How much disk can I realistically reclaim? Depending on how long the library has grown unchecked, teams routinely reclaim between 20% and 60%. Multi-terabyte savings are common on libraries older than five years.
Will deleting files break watch history? Plex keeps watch history keyed to metadata, not files. Deleting a file removes it from the library but not from history.
What is the safest first cleanup step? Run Tautulli for a month, then delete anything zero-played and larger than 5 GB. It is boring, but it works.
Do I need Sonarr and Radarr if I already use Plex? Plex only indexes what is there. Sonarr and Radarr add and remove files. Use both if you download.
Are these tools safe to run on a live library? Tautulli, Plex Meta Manager, dupeGuru, and PlexCleaner are safe on live libraries. Sonarr and Radarr take irreversible actions, so start in dry-run.
Can I run all of these in Docker? Every tool except the Claude Code CLI ships an official or community Docker image. The whole stack fits in one docker-compose file.
