Readwise Reader

We fixed some bugs where the share sheet would show a login error, and RSS folders were not able to be collapsed/expanded.

Core libraries update — We've updated many of the core libraries that support Reader, which should result in a more native-feeling experience, many fewer bugs, and faster performance!

Previously:

Readwise MCP & CLI — The (new, improved) Readwise MCP server gives Claude, ChatGPT, Cursor, or any MCP-compatible AI direct access to your Readwise highlights, and the full text of all your Reader documents. Once connected, you can ask your AI to search your notes, answer questions using your own reading history as context, or even make changes to your Reader library for you. We also pre-built AI workflows like inbox triage, feed catchup, a custom book review that is aware of related context in your vault, and a quiz mode for testing yourself on recent reads.

GPT-5.4 Support — If you bring your own OpenAI API key for custom Ghostreader prompts, you can now use GPT-5.4.

Improved Reading Progress — Artem and Lior enhanced the reading progress modal for EPUBs to show more detail at a glance. You can now choose between percent, page number, time left in chapter, time left in book, or hidden and each option previews the current value right in the picker.

Restore All from Trash — Mati added a "Restore All" button to the trash view on both web and mobile.

Improved Search Indexing — Piotr fixed an issue where profiles with extremely large documents (like multi-volume epub collections at 50MB+) could cause issues with search indexing. Large document batches are now split by size, so even the biggest libraries should index smoothly.

Fixed Tag Trimming — Adam improved server-side trimming for tag names, so tags created via the API or share extension can no longer end up with trailing spaces.

Modify Action Tags — You can now modify or remove header markers after creating them, and the associated sections and auto-generated tags update to match. As part of this change, Ibai also fixed an issue where .h1/.h2 header action tags and .c1/.c2 concatenation tags could interfere with exports.

Parsing Updates — Mati improved the speed and reliability of our YouTube parsing and fixed a bug where YouTube channel URLs were being treated as individual videos. Krzys improved page fetching on a wide variety of sites. Several engineers worked at fixing a variety of glitches with Twitter/X saving. Piotr fixed an issue where RSS feeds from certain sites were saving broken content instead of the actual feed. Mati fixed an issue where internal marker text was appearing at the bottom of emails in original view mode.

Fixed Highlight Navigation — Tapping highlights in the side panel should now take you to the relevant part of the document.

Podcast Transcripts — share any podcast episode from Apple Podcasts, Spotify, or Overcast into Reader, and it will be saved as a permanent, highlightable transcript.

Sticky Tablet Sidebars — You can now keep sidebars open while using Reader on tablets. In documents, opening the table of contents or notebook sidebar shrinks the reading view, so you can still scroll and highlight text. In your Library and Feed, the sidebar stays open so you can browse documents with the same context you’d see on the web version of the app.

Better Offline Control — Choose which categories of documents to keep available offline from the offline documents section of the Account screen. A new status shows whether a location is currently downloading.

Improved Action Sheets — Action sheets should now float nicely in the center of tablet screens, instead of at the bottom. Using “find in document” and similar options should be easier now.

Our full ePubs v2 feature includes 10x+ better performance on books, a dedicated long-form reading UI, and so much more. Our goal is to have Reader be the best app in the world for reading books (in addition to other formats), and this project is a big step towards that.