-
RSS and Atom reading joined the library
- Authenticated readers can import OPML, add or remove feeds, refresh them on demand, and track unread items.
- A feed item can be saved directly into the same article and PDF library for annotation and later research.
-
Memory capture became durable
- Saved research memories gained authenticated persistence, search, and a dedicated reading surface.
- Existing browser memories can be imported into the account-backed collection.
-
Safer failure recovery
- Unhandled API and Worker errors now return a controlled response instead of breaking the request path.
- A React error boundary gives the library a recovery surface when a page fails to render.
-
Reader moved to a faster app shell
- The product moved to a Vite and React SPA backed by a focused Hono Worker while preserving its production identity.
- Articles, PDFs, highlights, notes, boards, lists, search, and AI-assisted reading stayed together in one library.