Repository navigation
OPDS 2.0
From version 1.5.0, OPDS Plus reads OPDS 2.0 catalogs (the JSON format) as well as OPDS 1 catalogs (the Atom format). You add an OPDS 2.0 catalog the same way as any other. There is nothing to set: OPDS Plus tells the two formats apart by the feed the server sends.
| Server | OPDS 2.0 address | Notes |
|---|---|---|
| Komga | https://<server>/opds/v2/catalog |
Log in with your Komga username and password. Komga's OPDS 1 address (/opds/v1.2/catalog) still works too. |
| test.opds.io | https://test.opds.io/2.0/home.json |
A public sample catalog, handy to try the feature. |
| Any other OPDS 2.0 server | the server's start feed | Username and password (Basic) login only. |
Calibre-Web, Calibre-Web-Automated, Kavita, COPS and the default catalogs send OPDS 1. They keep working as before.
- Navigation links open like folders.
- Groups with a link of their own (for example "Latest" or "On deck") open that link. When the catalog says how many books the group holds, the number shows next to it in List View.
- Shelves are groups of books that a catalog lists on its start page without a link of their own, such as a "Staff picks" or "Keep reading" row. They open as their own folder with their book count, so the start page stays short. Search results that come only in such groups, as Komga sends them, are listed straight away.
- Books show their authors, series and number in it, cover, description, publisher, publication date and language when the catalog provides them. Every downloadable format is offered: EPUB, PDF, CBZ, CBR and so on.
- Paging: the next page is fetched as you scroll or tap the next page button, until the end of the feed.
- Facets (filters such as genre or read status) are in the menu at the top left, with their counts, and the active one is ticked.
- Search is in the same menu when the catalog offers it.
Catalog requests carry an Accept header that lists OPDS 2.0 first and OPDS 1 after it, so a server that offers
both sends the newer one. If a server refuses such a request, OPDS Plus asks again without the header until
KOReader restarts. A catalog's own Accept header, set under its custom headers, always wins: set it to
application/atom+xml to keep a server on OPDS 1.
- "This address is a web page, not an OPDS catalog": the address leads to a login page or the server's web reader. Use the OPDS address from the table above or from the server's documentation.
- A shelf is empty: the catalog changed since its start page was opened. Go back and open the start page again.
- A book has no download button: the catalog lists it without a file KOReader can open (an audiobook, or a book that can only be read in the browser).
- Login keeps failing: check the catalog's Basic credentials or custom headers. Versions 1.6.0 and later explain authentication documents as described below, but OAuth login remains unsupported.
An OPDS authentication document is information about a catalog's login methods, not a catalog of books.
From 1.6.0, it is recognized in a 401 response or at its direct address. Provider title/description,
Basic username/password labels and unsupported methods are shown in a scrollable information window.
Edit the catalog's existing username and password fields to use Basic authentication.
An advertised authentication-document link in an HTTP header, Atom feed or OPDS 2 feed becomes an
Authentication information entry without hiding the public catalog. Opening it uses the usual
cancellable background loader and leaves the current listing/history unchanged. A 401 response
advertising only a document link loads that document before explaining the available methods.
Linked discovery requires HTTPS on the exact source origin (scheme, host and port), verifies the certificate chain and hostname with KOReader's CA bundle, and sends no catalog credentials or custom headers. Redirects and cross-origin linked documents are refused rather than implicitly trusted. Missing CA bundles, failed verification and malformed documents are explained. Existing catalog requests, including Basic/self-signed setups, retain their current behavior.
This is discovery, not automatic login. Provider text is untrusted information; logos, registration, help and token endpoints are not automatically opened. OAuth remains unimplemented pending a concrete, verified provider/flow/storage decision. The old OPDS draft's password and implicit grants are not used as compatibility shortcuts; see RFC 9700.
Getting started
Servers
Help