Readeck

Save and read web content in a personal library

Bookmarks & Read-It-Later ★ 1.1k stars Easy setup AGPL-3.0

Readeck is a self-hosted application that lets you save web articles and pages into a clean, searchable personal reading library. It targets readers who want a simple read-it-later tool. It is deployed via Docker or a binary.

Key features

  • Clean article reading view
  • Highlights and labels
  • E-book export
  • Lightweight single binary

Pros & cons

Strengths

  • Simple and lightweight
  • Nice reading experience
  • Easy to deploy

Trade-offs

  • Younger project
  • Smaller community

Readeck replaces

Last reviewed Sep 13, 2026 · 840 words

Pocket shut down in 2025, and of the self-hosted replacements Readeck is the one I would put in front of a non-technical household. It is a single Go binary that runs on 256 MB of RAM, stores everything in SQLite by default, and does one job: save a page, strip it to a clean article, keep it forever, let you highlight it and export it as an EPUB for the e-reader. Two years old, AGPL-3.0, 1,100 stars, and already more pleasant to read in than tools twice its age.

The reading view is why you pick it

Most bookmark managers save a URL and a screenshot. Readeck fetches the page, extracts the article, and stores the text, images and a snapshot locally, so the copy you saved survives the original going paywalled or dead. The reading view has adjustable type, a progress marker that syncs across devices, and highlights that you select with a drag and revisit later on the bookmark's page. Labels organise; collections are saved searches over labels, sites and read state, which is how you build a "long reads, unread, from the last month" list without maintaining it.

Videos and pictures save too, with the video embedded and the picture kept at full size. The browser extension for Firefox and Chrome sends the current tab to your instance in one click, and the EPUB export turns any bookmark or collection into a file for a Kobo or Kindle, which is the feature that makes Readeck feel like a library rather than a list. There is an OPDS catalogue too, so an e-reader app can browse your collections directly.

Two minutes with Docker

The image lives on Codeberg's registry, which trips up people expecting Docker Hub:

services:
  readeck:
    image: codeberg.org/readeck/readeck:latest
    ports:
      - "8000:8000"
    environment:
      - READECK_SERVER_HOST=0.0.0.0
      - READECK_SERVER_PORT=8000
    volumes:
      - ./readeck-data:/readeck
    restart: unless-stopped

Open port 8000, and the first visit walks you through creating the initial user. Everything, including the SQLite database and every saved article, lives under /readeck, so backing up is copying one directory while the container is stopped or using SQLite's .backup while it runs. PostgreSQL is supported for larger multi-user installs, but for a household SQLite is the right default and I would not change it.

Put it behind a reverse proxy with HTTPS before installing the browser extension, because the extension stores an API token and you do not want that travelling in the clear. For LAN-only use, a Tailscale address works and needs no port forward.

Readeck, Wallabag and Linkding do three different jobs

Wallabag is the veteran in this space, with a decade of history, native mobile apps, and a PHP stack that wants MySQL or PostgreSQL and closer to 512 MB of RAM. Its reading view is fine; Readeck's is better, and Readeck imports a Wallabag export directly. Wallabag wins if you need its Android and iOS apps today; Readeck's mobile story is the responsive web UI plus a share target, which is adequate and not the same thing.

Linkding is not a reader at all. It stores bookmarks with tags and a search box, blazing fast, and never tries to render the article. People who want a searchable index of links they might revisit want Linkding; people who want to read those links later want Readeck. Running both is not silly. Karakeep sits in a third spot, with AI tagging and a screenshot-plus-archive approach that appeals to people hoarding everything; it is heavier and fussier to run.

The rest of the bookmarks category is mostly variations on these three shapes.

Rough edges in a young project

The project is two years old, and it shows in small places. There is no official mobile app, only the web UI and community clients using the API. Full-text search is present and capable but not the fastest on a 10,000-bookmark library. Extraction fails on a minority of sites with aggressive JavaScript rendering, the same minority every read-it-later tool fails on, and there is no "fetch via headless browser" fallback. And the community is small: the Codeberg issue tracker is responsive but you will not find a thousand blog posts about edge cases.

None of that stops me recommending it. The core is stable, the data format is a plain directory you can back up, and the export options mean you can leave without losing anything.

What I'd do

Install Readeck with the compose above, SQLite, behind Caddy with a real certificate, and import your Pocket or Wallabag export on day one. Add the browser extension on every machine, set up one collection called "unread long reads," and export it to your e-reader weekly. If you need native mobile apps this year, run Wallabag instead; if you never read what you save, run Linkding. For everyone in between, this is the one.

Compare Readeck

15 head-to-head comparisons.

Similar bookmarks & read-it-later apps