Readeck

Self-hosted bookmark manager and read-later tool, honestly reviewed. For people who lost their Pocket exports when Omnivore shut down and don’t want to repeat the mistake.

TL;DR

• What it is: Open-source (AGPL-3.0) read-it-later and bookmark manager — save articles, images, and videos from the web, read them offline, sync to a Kobo e-reader, highlight, export to EPUB [3][5].
• Who it’s for: People who relied on Pocket, Instapaper, or Omnivore and want something self-hosted that survives a startup shutdown. Especially useful if you read on a Kobo or other e-ink device [1][2].
• Cost savings: Pocket Premium runs ~$44.99/yr. Instapaper Premium is ~$35/yr. Readeck self-hosted runs on the Raspberry Pi gathering dust in your closet, or a $5/mo VPS — the software itself is free [3].
• Key strength: A single binary that’s genuinely fast. One reviewer imported ~4,000 articles from Pocket and noted the all-articles page loaded near-instantly on a local install [2]. Kobo/KOReader OPDS support is the decisive feature for e-ink readers [1].
• Key weakness: No bulk-edit or bulk-delete as of this writing (on the roadmap), limited data portability on export, and no hosted option yet — a managed service is planned for 2026 but doesn’t exist [2][5].

---

What is Readeck

Readeck is a web application that saves a local copy of web content — articles, images, videos — so you can read or revisit it later, whether or not the original URL still exists. The pitch on the homepage is “Save the story, cut the clutter” — which is accurate and refreshingly honest [homepage].

It’s hosted on Codeberg (not GitHub), maintained by a single developer, distributed as a single compiled binary with no external dependencies, and licensed under AGPL-3.0. There’s no company behind it in the Activepieces or n8n sense — it’s an independent open-source project, with a hosted service listed as “coming in 2026” [5].

The practical feature set: save a URL and Readeck fetches and archives the readable content, images, and metadata. You can label and search your collection, highlight text, export individual articles or entire collections as EPUB e-books, and save video links to get their transcripts [homepage]. A browser extension handles saving while you browse. The Kobo integration works via OPDS — the same catalog protocol used by Calibre — meaning your e-reader can browse and download directly from your Readeck instance without manual file transfers [1].

The codebase is Go, which explains the performance profile. It can run on a Raspberry Pi 2, uses under 512MB RAM in normal operation, and handles a collection of thousands of articles without database tuning [3][2].

---

Why people choose it

The two most detailed independent reviews both come from people migrating away from Wallabag, which tells you something about the audience.

Angela Ambroz [1] laid out her decision criteria explicitly: Kobo sync, Firefox extension, link archiving, and highlights — roughly in that order. She’d been on Wallabag but ran into two problems: a broken CSS/JS situation caused by an awkward environment variable handling PHP issue, and a KOReader sync that required bulk-downloading 50-100 articles at a time to browse what was available. She seriously considered Karakeep (for its LLM integration) before landing on Readeck because the e-ink sync story was better. Setup took about 20 minutes via Docker Compose [1].

The Autodidacts review [2] is more ambivalent and more useful for that reason. The author had originally passed on Readeck — partly because Wallabag’s Kobo integration seemed stronger at the time, partly because the project was newer and newer projects die. What changed their mind was performance: “The pageload times are insanely fast.” They imported a Pocket CSV of approximately 4,000 articles and found the all-articles page nearly instant. As a comparison point, they showed a screenshot of a different (unnamed) read-it-later app loading an empty unread view with 30 articles: slower. The performance gap was striking enough to reconsider [2].

The second thing that moved the Autodidacts reviewer was reading through the entire release history. Their conclusion: the developer “seems to care about performance, privacy & security, and code quality; listen to user feedback, but not unquestioningly; accept contributions; and have a positive, community-minded attitude.” The AI stance got a specific callout: the developer stated he won’t force AI features on users, but would consider LLM integration for tasks where it’s actually the right tool (like auto-tagging) — described as a “thoughtful, non-dogmatic, pragmatic stance” [2].

The Marius Hosting guide [4] takes a narrower angle — it’s a Synology NAS install walkthrough, not a review — but its existence is a useful data point: Readeck is popular enough in the NAS homelab community that step-by-step NAS guides get written for it. The Docker Compose configuration shown is six lines [4].

The Omnivore shutdown in late 2024 is context the provided articles don’t mention but any honest review of this space has to include: it drove a wave of self-hosted read-it-later migrations, and Readeck was one of the primary landing spots. If you self-host, you own the data. That’s the core value proposition.

---

Features

Based on the official website and documentation [3][homepage]:

Core reading and saving:

• Browser extension for saving while browsing, including pages behind a login that Readeck’s server can’t reach directly [homepage][1]
• Full content archiving — article text, images, and metadata saved locally [homepage][3]
• Readable view with adjustable font, text size, and line height — settings persist [homepage]
• Dark/light theme [4]
• Label-based organization with full-text search across your entire collection [homepage]
• Dynamic collections (saved searches) [homepage]

Highlights:

• Highlight text in any saved article
• Browse all highlights across your collection in one place [homepage]

Video:

• Save a video URL and Readeck retrieves the transcript when available
• Transcript is searchable, highlightable, and exportable like an article [homepage]

E-reader integration:

• Export individual articles or entire labeled collections as EPUB e-books [homepage]
• OPDS catalog support for e-readers — Kobo and KOReader can browse and download directly [1]

Deployment:

• Single binary — no dependencies, no runtime installation [3]
• Docker with one command for testing; Docker Compose for production [3][4]
• SQLite (default) or PostgreSQL as the database [2][3]
• Binary runs on Linux, FreeBSD, macOS, or Windows [3]
• Minimum 512MB RAM; runs well on a Raspberry Pi 2 [3]
• OPDS catalog exposed for e-reader clients [1]

API and integrations:

• REST API available; described as pleasant to use [2]
• Wallabag API import supported (pull from an existing Wallabag instance) [2]
• Pocket CSV import [2]
• Export: Readeck-specific ZIP format importable into Readeck, but not into other services without conversion [2]

What’s missing or limited:

• No bulk-edit or bulk-delete as of current release (on roadmap) [2]
• Can’t limit archiving to starred articles only, or disable image saving, or skip content extraction for specific items [2]
• No manual re-fetch trigger for a specific article after failed fetch [2]
• No Wallabag JSON import — only the API import route [2]
• No LLM/AI features (by deliberate choice for now) [2]

---

Pricing: SaaS vs self-hosted math

Readeck has no commercial pricing today. The software is free and self-hosted only. A managed hosted service is listed on the Get Started page as “coming in 2026” with pricing not yet announced [5].

For comparison, the mainstream read-it-later SaaS options:

Pocket (Mozilla):

• Free: unlimited saves, basic search
• Premium: ~$44.99/yr — adds permanent library, full-text search, suggested tags

Instapaper:

• Free: unlimited saves
• Premium: $2.99/mo ($35/yr) — adds text-to-speech, unlimited highlights, speed reading

Raindrop.io (bookmark-manager angle, not pure read-it-later):

• Free: unlimited bookmarks, basic features
• Pro: $3/mo (~$28/yr) — nested collections, full-text search, duplicate finder

Readeck self-hosted:

• Software: $0 (AGPL-3.0)
• VPS to run it: $4–6/mo on Hetzner or Contabo — though it also runs on hardware you already own (Synology NAS, Raspberry Pi, home server) [3][4]
• Storage: roughly 300KB per saved bookmark; a 10,000-article archive is about 3GB [3]

Concrete math:

If you’re a Pocket Premium subscriber paying $44.99/yr, switching to Readeck on a $5/mo Hetzner VPS costs $60/yr — so self-hosting costs more than Pocket Premium if you don’t have existing infrastructure. The calculus flips if you already have a NAS or home server running, in which case the marginal cost is nearly zero.

The real argument for Readeck isn’t price — it’s control. Pocket is owned by Mozilla, which has been under financial pressure. Omnivore, a well-funded read-it-later app, shut down in late 2024. When the service shuts down, your saved content and reading history go with it. Readeck running on your own hardware doesn’t shut down when someone’s runway ends.

If Readeck launches its 2026 hosted service, pricing is genuinely unknown — but the developer’s stated intent is to offer it as a way to support the project for people who don’t want to manage infrastructure [5]. Data not available on what it will cost.

---

Deployment reality check

The documentation makes a promise that’s actually accurate: a single Docker command gives you a working instance.

docker run --rm -ti -p 8000:8000 -v readeck-data:/readeck codeberg.org/readeck/readeck:latest

That’s it for local testing [3]. Production deployment uses Docker Compose and adds a reverse proxy for HTTPS — the docs cover Caddy and nginx configurations.

What you actually need:

• A Linux machine (VPS, NAS, Raspberry Pi, or a spare laptop)
• Docker installed
• A domain name and reverse proxy if you want HTTPS access outside your home network
• About 20 minutes if you’ve touched Docker before [1]

For Synology NAS users: Marius Hosting has a complete step-by-step guide using Portainer. The Docker Compose for Synology adds READECK_USE_X_FORWARDED: true to handle the reverse proxy header correctly [4]. Six environment variables total in the production stack.

Storage estimate: Plan ~300KB per bookmark. A collection of 10,000 articles needs about 3GB of storage [3]. For most people this is noise.

What can go sideways:

The main friction point documented in the reviews is the import experience. The Autodidacts reviewer imported a Pocket CSV of ~4,000 articles, had an internet glitch mid-import, and couldn’t find a way to resume — they deleted the data volume and started over [2]. There’s no partial-import recovery. Budget for a stable internet connection if you’re doing a large migration.

The same reviewer flagged the lack of bulk-delete as a concrete problem during setup: if an import goes wrong, your only option is deleting the Docker volume. There’s no in-app “delete everything” or multi-select delete [2].

The KOReader OPDS integration that’s a major selling point requires getting HTTPS set up correctly — the Marius Hosting guide specifically notes WebSocket must be enabled for HTTPS to work [4]. Not hard, but it’s a step.

If you’re running Readeck on hardware accessible only from your local network, reaching it from your phone or a work browser requires either a VPN or exposing it to the internet with proper HTTPS. Not Readeck-specific, but worth planning for.

Realistic time estimates: 15–30 minutes for a technical user on a VPS with Docker experience. 1–3 hours for someone setting up a domain, reverse proxy, and figuring out HTTPS for the first time. Running on a Synology NAS with the Marius Hosting guide: 45–60 minutes following steps.

---

Pros and cons

Pros

• Genuinely fast. Go-based single binary with near-instant page loads even with thousands of articles [2]. This matters more than it sounds — slow read-it-later apps are annoying to use daily.
• Minimal footprint. 512MB RAM, runs on a Raspberry Pi 2, single binary with no dependencies [3]. You can deploy it on hardware you already own.
• OPDS catalog for Kobo/KOReader. The e-ink sync story is one of the best in the category. OPDS lets your e-reader browse and download directly without manual file transfers [1][2].
• Simple deployment. Single Docker command for testing, Docker Compose for production. No required external databases or services for basic use [3].
• Handles large collections. 4,000+ articles imported without performance degradation [2].
• Developer track record. Reading through every release note gave one reviewer confidence: care about performance, privacy, security, responsiveness to feedback, and a non-dogmatic stance on AI [2].
• Video transcripts. Save a YouTube or video URL and get a searchable, highlightable transcript — genuinely useful [homepage].
• EPUB export. Take a collection with you on an e-reader for a weekend without managing files [homepage].
• SQLite or PostgreSQL. SQLite default is fine for personal use; PostgreSQL available if you need it [2][3].

Cons

• No bulk-edit or bulk-delete. If you need to clean up or reorganize a large collection, you’ll feel this quickly. On the roadmap but not shipped as of this review [2].
• Export portability is limited. Readeck exports to a Readeck-specific ZIP format. If you want to migrate away later, you’ll need to munge the format — unlike the open Wallabag JSON export [2]. For a tool whose main pitch includes “keep it forever,” this is an ironic weakness.
• No hosted instance yet. Self-host or nothing, until sometime in 2026 [5]. Non-technical users have no easy entry point.
• AGPL-3.0, not MIT. If you want to embed this in a commercial product or build a white-label service around it, the license requires you to open-source your modifications and the surrounding application [profile]. Less of an issue for personal or small-team use.
• No manual re-fetch. If Readeck failed to capture content for an article, there’s no “try again” button for individual articles [2].
• Content extraction is all-or-nothing. Can’t tell Readeck “save full text for starred, save just the URL for everything else” [2]. Heavy archivers will feel the storage and bandwidth implications.
• Single developer. The project is healthy and active, but it’s one person. The business model for long-term sustainability is unproven — the 2026 hosted service is the stated plan [2][5].
• Monetization unknown. The question one reviewer raised — what will this cost in the long run, and will all features remain available to self-hosters? — hasn’t been answered yet [2].

---

Who should use this / who shouldn’t

Use Readeck if:

• You read on a Kobo or other KOReader-compatible e-ink device and want seamless article sync.
• You’re migrating away from Pocket, Instapaper, or Omnivore and want to own your data this time.
• You have a NAS, Raspberry Pi, or home server already running — the marginal cost to add Readeck is essentially zero.
• You prioritize simplicity and performance over a large feature surface.
• You want something that runs quietly without maintenance attention.

Skip it if:

• You need a mobile app rather than a web UI (Readeck is web-only; mobile access is through the browser).
• Bulk operations across a large collection are part of your workflow — the missing bulk-edit is a real friction point [2].
• You need a hosted service today, not in 2026 [5].
• You want LLM-assisted organization (auto-tagging, semantic search) — Karakeep is the option to evaluate for that.
• You plan to build a commercial product around it — the AGPL-3.0 license has implications [profile].

Stay on Pocket/Instapaper if:

• You don’t have any existing self-hosted infrastructure and aren’t interested in learning.
• The command line makes you anxious and there’s no technical person to help.
• You have a small collection and a free tier covers you — the cost argument for switching is weak in that scenario.

---

Alternatives worth considering

• Wallabag — the older, more established self-hosted read-it-later app. PHP-based, so heavier to run; KOReader integration is present but more limited than Readeck’s [1][2]. Has bulk operations and Wallabag JSON export. Solid if you need Wallabag’s broader integration ecosystem.
• Karakeep (formerly Hoarder) — self-hosted, adds LLM-based auto-tagging and semantic search. The choice if AI-assisted organization matters to you. The Autodidacts reviewer seriously considered it [2]; Angela Ambroz did too [1].
• Omnivore — shut down in late 2024. Mentioned here because many people searching for Readeck reviews are Omnivore refugees.
• Pocket — the incumbent SaaS. Easiest onboarding, owned by Mozilla, Premium at ~$44.99/yr. Your data lives on their servers.
• Instapaper — simpler SaaS option, cleaner reading experience, Premium ~$35/yr. Has shut down and relaunched before, which is relevant context.
• Raindrop.io — leans more bookmark-manager than read-it-later, but covers both. Polished SaaS with a free tier and Pro at ~$28/yr.
• Shiori — minimal self-hosted bookmark manager, less focused on reading experience, no e-reader integration.

For the audience most likely to read this — someone escaping a SaaS that shut down or got expensive — the realistic shortlist is Readeck vs Wallabag vs Karakeep. Readeck wins on performance and e-ink integration. Wallabag wins on maturity and export portability. Karakeep wins if you want AI features. Pick based on which of those three matters most to you.

---

Bottom line

Readeck is a focused, well-built tool that does what it says: save web content, read it later, sync to your Kobo. It’s not trying to be an AI research assistant or a visual mood board. The Go architecture makes it genuinely fast and trivial to run — a single binary that’s comfortable on a Raspberry Pi is a real achievement in a space full of PHP and Node.js apps that need 2GB of RAM to boot. The OPDS e-reader integration is the best in the self-hosted category.

The meaningful downsides — no bulk-edit, limited export portability, no hosted option yet, single-developer sustainability questions — are real and worth knowing before you migrate 5,000 articles into it. They’re not dealbreakers for most personal use cases, but they should factor into the decision.

For a non-technical founder who currently pays Pocket or Instapaper and reads on a Kobo or tablet, the value is clear: a one-time deployment puts the archive on infrastructure you control, the software costs nothing, and the reading experience is better. The setup is genuinely approachable — 20 minutes with Docker. If that 20 minutes is still a blocker, that’s exactly the kind of deployment upready.dev handles for clients.

---

Sources

1. Angela Ambroz — “Migrating from Wallabag to Readeck” — angelaambroz.com. https://www.angelaambroz.com/posts/readeck/
2. The Autodidacts — “First impressions of Readeck” — autodidacts.io. https://www.autodidacts.io/readeck-open-source-read-it-later-app-with-kobo-support/
3. Readeck Official Documentation — “Documentation — Installation, Requirements, Deployment” — readeck.org. https://readeck.org/en/docs/
4. Marius Hosting — “How to Install Readeck on Your Synology NAS” — mariushosting.com. https://mariushosting.com/how-to-install-readeck-on-your-synology-nas/
5. Readeck Official — “Get Started with Readeck” — readeck.org. https://readeck.org/en/start

Primary sources:

• Official website: https://readeck.org/en/
• Code repository: https://codeberg.org/readeck/readeck
• Browser extension: https://readeck.org/en/extension
• Download page: https://readeck.org/en/download