Suwayomi-Server

Last reviewed Sep 13, 2026 · 774 words

Suwayomi-Server, the project formerly called Tachidesk, is Tachiyomi's extension engine moved off your phone and onto a server, so that the same sources you would browse in a mobile manga app are fetched once, cached, and read from any browser with progress synced. That is a different job from Komga or Kavita, which serve files you already own. Suwayomi fetches from online sources through community extensions, and the legality of those sources varies from "official publisher site" to "not remotely licensed", which is the thing to be clear-eyed about before you open a port.

Extensions are the product and they are not bundled

Since 2024 no extensions ship with the server. You add an extension repository URL in the settings, the server lists what that repo offers, and you install sources individually. The repositories are maintained by the community, not the Suwayomi project, precisely because of takedown history: the original Tachiyomi project wound down under legal pressure in early 2024, and the ecosystem regrouped around forks and separately hosted extension repos. Suwayomi-Server itself is MPL-2.0, Java, 7,549 stars, and contains no source-specific code. What you install into it is your decision and your liability.

The practical split is this. Publisher-run and legitimate aggregator sources exist among the extensions and are fine to use. Scanlation aggregators are the majority of what people actually install. If that matters to you, or the server sits at a workplace, install only the sources you can name the licensing of, or use Suwayomi purely with its local source folder, where it reads CBZ and folders of images you supply.

Setup is one container on port 4567

services:
  suwayomi:
    image: ghcr.io/suwayomi/tachidesk:stable
    ports:
      - "4567:4567"
    volumes:
      - ./data:/home/suwayomi/.local/share/Tachidesk
    restart: unless-stopped

The JVM idles at a few hundred megabytes and 512 MB is a realistic floor; give it 1 GB if you download aggressively. Everything (database, downloaded chapters, extension jars, thumbnails) lives under that one data path. There is also a plain JAR and desktop bundles with the web UI embedded, which is the simplest way to try it on a laptop before committing a server.

Turn on basic authentication in Server Settings before exposing it beyond the LAN; the default is open. Better still, keep it behind Tailscale and never expose it at all, since it has no rate limiting and no account system beyond that one credential.

Progress sync is the reason to run it instead of a phone app

Reading position, library, categories, and download state live on the server, so the web UI on a tablet, the browser on a desktop, and a mobile client all show the same place in the same chapter. Mihon and other Tachiyomi forks can connect to a Suwayomi server as a source through a dedicated extension, which turns the phone app into a client of your server rather than a fetcher of its own. Downloads happen once, on your connection, and the phone reads cached pages over your VPN.

Automatic library updates check every followed series on a schedule and can download new chapters as they appear. This is where the JVM cost comes from and where you will get rate-limited by sources if you set the interval too aggressive; every 12 hours is plenty for most libraries.

It is manga-only, and that is the limit to plan around

Suwayomi has no concept of a novel, a comic in the Western sense, or a book. If the household also has a CBZ collection, DRM-free ebooks, or webtoons downloaded as files, you end up running Komga or Kavita beside it anyway. Some people wire the two together: Suwayomi downloads to a folder as CBZ, Komga indexes that folder, and Komga becomes the permanent library with proper metadata while Suwayomi is the fetcher. That is more moving parts than most want, but it gives you a library that survives extensions going dead.

For everything that is not manga, the ebooks category is the right place to look, and Kavita is the larger project with the broader format support.

What I'd do

Run the container behind Tailscale with basic auth on, add one extension repository, install only the sources you have a defensible reason to use, and set library updates to twice a day. Point Mihon at the server for the phone. If any of the reading is from files you own rather than online sources, run Kavita as the permanent library and treat Suwayomi as the fetch-and-cache layer, not the archive.