Omnivore

Omnivore is an open-source read-it-later library: capture articles, pages and PDFs as clean text, then mark them up and read them offline. The feature set is unusually complete for the category — highlights and labels, full-text search, text-to-speech on iOS, an API, and iOS, Android and browser-extension clients — all under the AGPL-3.0 license. What changed is the hosting. The README states that the cloud app was deprecated in November 2024 and that Omnivore is now 'completely self-hosted': you run the server yourself, a Docker Compose stack with its own database, and point the clients at your instance. The hosted service is gone, the old app domain no longer belongs to the project, and the repository is the project's only home.

That makes Omnivore a sovereignty play, not a convenience play. You get a read-later archive no company can delist, reprice or shut down — but the vendor is now you: you patch the updates, back up the database, and fix sync yourself the day a mobile client stops talking to your build. Check the commit history before betting a decade of highlights on a repo whose maintenance depends on whoever still cares. If your 'later' pile is personal, text-first and you have a NAS to host it on, this is a genuinely rare thing: an archive you own. If you just want the pile to work without any ops at all, pick a hosted reader instead and spend zero minutes on servers.