Mylar3

Last reviewed Sep 13, 2026 · 790 words

Mylar3 is the missing *arr for comics, and the one setup decision that determines whether it works is getting a ComicVine API key before you add a single series. Everything Mylar3 does, from recognising that "Saga 063" is the issue you are missing to naming the file so Komga shelves it correctly, flows through ComicVine metadata. With the key in place, the workflow is the one Sonarr users already know: add a series, mark wanted issues, let indexers and a download client do the rest, and read the result in a proper reader. Without it, you have a folder scanner with opinions.

Where it sits in the stack

Mylar3 is the middle box. In front of it sit indexers, which you should manage through Prowlarr so the newznab and torznab feeds sync automatically. Behind it sits a download client, with SABnzbd the smoothest pairing because comic releases are mostly a Usenet phenomenon; qBittorrent works for torrents. After Mylar3 has renamed and tagged a file, a reader takes over: Komga or Kavita point at the same library folder and pick up new issues on their next scan. Mylar3 never displays a page of a comic itself.

Docker is the sane route. The lscr.io/linuxserver/mylar3 image listens on port 8090, keeps config.ini and the SQLite database under /config, and needs the same download and comics folders mounted at the same paths as the download client sees them, or completed downloads will never be found. That path-mismatch mistake accounts for most first-week support threads.

ComicVine is the source of truth, and it rate-limits

Comic Vine issues free keys, but the API allows roughly 200 requests per 15 minutes and Mylar3 is chatty when it first pulls a long-running series. Add a 300-issue title and expect the initial refresh to take a while, with a wait between series if you are importing a whole existing collection. The metadata quality is excellent for mainstream US publishers and inconsistent for manga, European albums and small-press work, where volume naming fights the issue model. If your library is mostly manga, Mylar3 is the wrong shape; use Kavita's own metadata and a manual download flow.

Searching by issue number is harder than by episode

Television has season and episode numbers baked into every release name. Comics have reboots, volume numbers, annuals, "#1 of 6" mini-series and publishers who restart at 1 every few years. Mylar3 handles this with a set of matching rules and per-series overrides, and you will use the overrides. The typical fix is telling a series its start year so the 2018 volume stops matching the 1994 one. Expect an hour of this per 50 series when you first import, and a few minutes a month after.

The weekly pull list is the feature that makes it feel like Sonarr: it fetches the upcoming releases, shows the ones from series you follow, and grabs them on release day. Mark a series "watched" and it fills a folder without you touching it.

Tagging makes the reader work properly

Turn on the ComicTagger integration so every file gets a ComicInfo.xml embedded in the CBZ. Komga and Kavita read that file for series, number, writer, artist, summary and cover date, which is the difference between a shelf that looks like a shop and a folder of filenames. Convert CBR to CBZ at the same time; RAR archives are a legacy format that nothing needs any more. The folder and file naming patterns are configurable in the settings page, and the default of one folder per series with Series (Year) #Issue filenames suits both readers.

Honest limits

Sources come and go. A comic-specific tracker or a direct-download site that Mylar3 supports this year can vanish next year, and the project's "depends on external sources" con is a permanent condition rather than a bug. The interface is functional and older-looking, with a configuration page that runs to several screens. And with a one-developer core team, breaking changes arrive without much ceremony; pin the image tag and read the changelog before pulling.

What I'd do

Prowlarr feeding indexers, SABnzbd for downloads, Mylar3 in the LinuxServer container with identical volume paths across all three, ComicTagger enabled and CBZ conversion on. Import your existing collection 20 series at a time to stay inside the ComicVine limits, fix start years as they come up, then subscribe to the pull list for the 10 or so titles you actually follow. Point Komga at the library and stop thinking about it. If most of what you read is manga, skip Mylar3 and let Kavita do the organising.