Added

• DOMParser for Content Scripts

Fixed

• Youtube transcript extraction

Added

• Pinboard import adapter, by @jlecour

Improved

• ACLs system

Changed

• updated Go to 1.26.1

Fixed

• Markup error in "about" page
• HTTP cache for zstd responses

Added

• TOTP support
• Forwarded authentication
• mTLS for HTTPS listener
• Bookmark properties editor
• Highlights with annotations
• User's last login date and time
• Content-script for nebula.tv, by @xvello

Improved

• Enabled theme toggle for public bookmarks, by @shom
• Configurable UNIX socket file mode
• WebP probing and processing, by @mislav
• Log warning instead of error when JSON-LD can't be parsed
• Let the PWA use the device orientation, by @mlfh
• Fixed positioning of highlight menu on mobile
• Thumbnail and bookmark image size, by @mislav
• Handling zero-width spaces when stripping heading anchors, by @mislav
• Speed up fetching bookmark outgoing links, by @mislav
• Support application/xml documents when they are HTML
• Keep HTML content from metadata
• Mediapart content-script
• (X)HTML initial scan for charset detection, by @mislav
• Support declarative Shadow DOM, by @mislav

Changed

• updated Go to 1.26.0

Fixed

• Back/forward navigation with gestures
• EPUB mimetype
• Prevent error log in Fediverse content-script
• Preserve anchor links with special characters, by @mislav
• Display of SVGs with percentage dimensions, by @mislav
• Sorting on api/bookmarks/sync
• Typo in API documentation, by @clmt
• Wallabag import encoding and content type issues, by @mislav
• Raise validation errors before the import process, by @mislav
• Font immutable caching
• Restore Safari scroll position on back/forward navigation, by @mislav
• Added email/send permission in bookmarks:read OAuth scope
• Support SVG images via object embeds, by @mislav

Fixed

• Sorting by date with SQLite

Fixed

• Prevent infinite loop during Wallabag and Omnivore import

Improved

• MathML support, by [@mislav]
• Better extracted domain parser

Added

• Mathjax and Katex support, by [@mislav]

Improved

• Website icons visibility in dark mode
• Handle "pt" CSS unit in SVG dimensions, by [@mislav]

Fixed

• Return a 422 with errors on /api/bookmarks when the filters contain errors

Fixed

• User management command would create users without UID
• OAuth token revocation error when called with a session

Changed

• Readability v2.1.0
• Improved internal testing

Fixed

• Regression with Youtube extraction
• Regression with cover image extraction
• Layout bug with reading time on bookmark list

Fixed

• Initial PostgreSQL schema

Improved

• Save animated GIFs and convert single-frame GIFs to PNG
• Github extraction, by by [@mislav] and [@linsui]

Added

• HTTPS listener with certificate/key pair
• log_time_format (and READECK_LOG_TIME_FORMAT) configuration option, by @mislav
• OAuth2 authorization code flow
• OAuth2 device code flow

Changed

• Replaced the ACL policy enforcer for better performances and less dependencies
• Removed sign-out button from main menu (now available in the profile page)
• Go-Readability v2, by @mislav

Improved

• Handle lazy-loading "data-src" pattern for embedded videos, by @mislav
• Support CSS units when parsing SVG element dimensions, by @mislav
• Stop tracking reading progress once it reaches 100%
• ARIA attributes on forms and messages
• Better JSON-LD and microdata support in content extraction
• Split token for password recovery

Fixed

• Reddit content-script, by @linsui
• Cut words on runes, not bytes
• Silent bug in internal/db/scanner
• Disable bfcache after side effect requests (post, put, patch, delete)
• Layout but with modal image button on small images
• Readability: avoid unwrapping meaningful HTML elements, by @mislav
• Sync API date selection with SQLite
• Retrieve label with zero width joiner (ie. compound emojis)
• Content Script for arstechnica (com and co.uk)

Changed

• Updated Go to 1.25.3

Improved

• Aria roles on forms and messages
• Prioritize PNG and JPEG images when fetching images

Fixed

• PostgreSQL bugs on full export
• Always remove remote source URLs that can't be retrieved during the archive process

Changed

• Updated Go to 1.25.1

Fixed

• Restored CSRF protection exemption for token authenticated requests

Changed

• Updated site config files

Fixed

• make CSRF protection compatible with the browser add-on

Added

• autocomplete on title, author, site and label in filter form
• share by link and email API documentation
• extract fediverse:creator metadata as author
• feeds for any bookmark list
• Readwise Reader CSV import, by @mislav
• Linkwarden import adapter
• synchronization API
• listen on unix domain sockets, by @noctux
• http/2 cleartext support
• diagnosis modal with extraction log on each bookmark
• modal images in articles
• Open Graph metadata on publicly shared bookmarks, @mislav
• DOM API in content scripts
• Log format configuration entry
• Download API tokens as text files
• Profile (one user only for now) export and import
• readability structured logging, by @mislav

Fixed

• scoped_admin_r scope was missing the system:read permission
• consider the <base> element when present for page's URL resolution
• YouTube video embeds are now preserved as a thumbnail that links to the original video, by @mislav
• HTTP cache couldn't work with an API token
• timezone related tests would fail with a TZ env var, by @mislav
• archive lossless WebP images as PNG instead of JPEG
• archive inline SVGs without an XML namespace, by @mislav
• prevent AVIF images from interfering with fetching of alternative image sources
• more permissive Open Graph metadata scanning
• correct encoding for imported Wallabag articles that were originally non-UTF8, by @mislav
• layout in mosaic view with very long titles
• readeck.exe crash, serving web assets on Windows, by @mislav
• preserve all matched nodes when applying FiveFilters body selector rules, by @mislav
• duplicate load (and run) of user content scripts
• Amazon regional storefronts inherit amazon.com site rules, by @mislav
• recognize historical, IE-compatible markup for site favicons, by @mislav
• Fediverse content script fetching Activity Streams JSON for certain URLs, by @mislav
• fixed a crash during URL processing in the archive process, by @mislav
• prevent content scripts to timeout on the whole extraction process

Improved

• new archiver process: faster and smaller memory footprint
• documentation extracted to gettext
• check that folders and configuration exist before import or export
• improved settings UI
• removed autocapitalization on label form
• headers sent when fetching
• send a referer header when fetching favicon or a page's image
• hide bookmark description when it exists at the begining of the article
• increase sharpness of archived images on high-resolution ("2x") displays, by @mislav
• strip links from self-linking heading elements, by @mislav
• handling of multiple source images (picture source and srcset)
• cookbook API documentation (POST on extract route), by @mislav
• switch to the readeck/go-readability fork for compatibility with Mozilla Readability.js 0.6.0, by @mislav
• refresh user content scripts when they've changed
• improved flash messages styling
• feedback on "copy to clipboard" buttons
• removed server hostname from the request ID in the log entries
• new internal HTTP client, shared by the extract and archive process

Changed

• updated Go to 1.25.0
• new .ico decoder
• Go 1.25 CSRF protection

Deprecated

• markdown export in multipart is replaced by the sync API and will be removed in the the next version

Fixed

• Default scheme to https when a request comes from a reverse proxy
• More translation updates

Fixed

• pt-PT translation is back

Added

• share a bookmark by email
  • send the bookmark's content as a message
  • send the bookmark as an EPUB attachment
• mosaic view (image wall on bookmark list)
• browser extension calls to action (onboarding and sidebar)
• author to the exported EPUB, by @ruben

Changed

• Reverted the SQLite driver to its CGO version

Removed

• Application passwords. API token can now be used in HTTP Basic Authentication when needed

Fixed

• Bug in Wallabag import adapter that would not load entries with empty headers
• Bug when retrieving SVG images

Improved

• Check Origin header in CSRF protection
• Strings that were misspelled or difficult to translate

Added

• server.base_url setting

Fixed

• Extraction error when a page contains empty link header value

Fixed

• Fatal error on "About Readeck" page with fresh install
• Pocket import adapter
• Bug with contiguous highlights not displaying properly (or at all)
• href value in bookmark list (API)

Improved

• Use current list ordering when exporting EPUBs, by @knightpp
• Removed forced server port in container image, by @martenlienen
• Removed AMD64=v2 build optimisation

Added

• Export command
• Import command
• Import from CSV and Instapaper
• Last used dates on API Tokens and Application Passwords
• Show disk usage in the "about Readeck" page, by @Pontoporeia
• Spanish (español de España) translation
• Galician (galego) translation

Improved

• Mark imported archived bookmarks as read
• Option to archive and/or mark read all imported bookmarks
• Reading time pluralization, by @zull
• Case insensitive label suggestions
• Exhaustive and restrictive Permissions-Policy headers
• Extraction of pages on Wordpress with the activitypub plugin
• Extraction of pages with a Link <url>; rel="original" header

Changed

• Updated Go to 1.24
• New SQLite, CGO-free driver. Important: it won't run with MemoryDenyWriteExecute=yes in the systemd service file.
• New internal form API
• New secure cookie for session and CSRF protection

Fixed

• Safari display bug with "details" HTML elements in bookmarks
• Regression with Safari when creating highlights

Breaking changes

• All bookmark internal URLs have changed
• New API token format. Previous token format won't work anymore.
• New secret key derivation scheme. Application password must be refreshed.
• When using the browser extension, you must logout and login again.
• MemoryDenyWriteExecute=yes must be removed from the systemd service (see above).

• Reading progress tracker on articles
• Filter bookmarks by reading progress
• Filter bookmarks by more than one type
• user sub command to create or update users, by @algernon
• Colored highlights, by @makebit
• Adjustable content's width on bookmark view
• Added word_count and reading_time to bookmark api, by @eleith

Fixed

• Search query with backslash would be parsed incorrectly
• Username validation during onboarding

Improved

• Only load translation that are at least 90% completed
• Resize images by maximum width only
• Bookmark display with wide tables, by @makebit
• Container based responsive layout

Changed

• New, native, logging library
• Bigger card icons on desktop and even bigger on touch screens
• Build with Go 1.23.4

• Reading progress tracker on articles
• Filter bookmarks by reading progress
• Filter bookmarks by more than one type
• user sub command to create or update users, by @algernon
• Colored highlights, by @makebit
• Adjustable content's width on bookmark view
• Added word_count and reading_time to bookmark api, by @eleith

Fixed

• Search query with backslash would be parsed incorrectly

Improved

• Only load translation that are at least 90% completed
• Resize images by maximum width only
• Bookmark display with wide tables, by @makebit
• Container based responsive layout

Changed

• New, native, logging library
• Bigger card icons on desktop and even bigger on touch screens

• Content Script for 404media.co
• Permit removal of a loading bookmark
• Filter bookmarks with labels, with errors and by loading state
• Show bookmark creation date on the sidebar
• Alignment and hyphenation reader settings
• Use alignment and hyphenation settings in EPUB export, by @bramaudi

Fixed

• Prevent the sort menu to go offscreen on mobile
• Identify a photo type by using the DCMI type value

Improved

• Hide top navigation when scrolling down
• Sliding menu on mobile
• Bookmark sidebar can be hidden on all screen size
• Normal size of label input field, by @lilymara
• Better User-Agent for nytimes.com retrieval

Changed

• trusted_proxies setting. This list the proxy that are authorized to set X-Forwarded-... headers. The default list provides sensible defaults so it's not a major breaking change.
• Updated site-config rules

• Added Literata, by @diginaut

Fixed

• Support of new Pocket export format

• Import from Omnivore
• Improved Youtube video descriptions
• Support for text/plain extraction

Fixed

• Fixed a bug with video bookmarks without an actual video content
• Fixed a typo on the onboarding page, by @ggtr1138

• Documentation in German, by @FrankDex
• Convert sections from browser bookmarks to labels
• Support for "tag" attribute in browser bookmarks import files (pinboard format)
• Support for "toread" attribute in browser bookmarks import files (pinboard format)
• Import from Goodlinks

Improved

• Custom error page when a bookmark doesn't exist
• Improved selected text color in dark mode

Fixed

• Support for t.co links, by @joachimesque
• Fixed extraction error with fastcompany.com
• Stop emptying the resource cache when closing its entry io.Reader

• Spanish translation, by @vsc55
• Hugarian translation, by @User240803
• Chinese translation, by @shichen437

Changed

• Updated dependencies
• Updated site-config rules

Fixed

• German translation errors
• Only reset the user seed upon username or email address change.

Added

• Bookmark list ordering options
• Polish translation, by @anarion
• Czech translation, by @marapa
• German translation, by @FrankDex

Improved

• Can now change an API token name
• Extract inline SVG as remote images

Changed

• Updated readability
• Updated site-config rules

Fixed

• Fixed a bug with PostgreSQL installation on a different schema, by @winston0410
• Fixed a bug with the Wallabag import tool. It would fail when the provided URL is missing a trailing slash.

Added

• Import from other services
  • Text file. A simple file with one URL per line.
  • Browser Bookmarks. The file you can export from your browser menu. The title and date are retrieved from the file.
  • Pocket. From Pocket, you can export a file and upload it in Readeck import tool. Title, labels and date are retrieved from the file.
  • Wallabag. This uses the Wallabag API to retrieve everything, including the content the way it was initially saved.
• Breadcrumbs on profile, admin, documentation and import
• Keep the URL as submitted by the user in a new database column
• cleanup command to remove all pending (loading state) bookmarks
• Optional compact layout for bookmark lists
• Unified bookmark list header with layout switcher and export menu
• Publicly shared link expiration can be configured (instance wide)
• Direct link to the original URL on every bookmark card

Changed

• Major internal/bookmarks refactor
• Updated site-config rules
• Metrics and workers configuration
  • keep only one metrics configuration (in [metrics]) with environment
    overriding when needed.
  • added environment configuration for the worker configuration
  • adjust the default number of workers
  • unset every READECK_ environment variable upon startup so they won't
    leak in case of a major security breach.
• SQLite connection optimizations
• DELETE /api/bookmarks/{id} performs an immediate deletion (unlike the web view that schedules a task)
• Button appearance (better active state, slight change of layout)
• HTTP client configuration (TLS 1.2 and 1.3 only), chrome like cipher list

Fixed

• Keep namespaces during SVG files cleanup
• Data race condition with the in-memory store

• Healthcheck command (only connect to server, don't perform any request)

Added

• Internationalization
• French translation
• Bluesky content script. Thanks @joachimesque
• Healthcheck command

Improved

• Bookmark search with wildcards and exclusions
• Preserve titles in content during extraction
• Embed video privacy
• Embed player for invidious videos
• Disable readability when saving selection from the browser addon
• More natural label ordering
• White border around QR codes so scanning works in dark mode too

Added

• More font choices. The reader interface offers now 8 font faces
• Dark mode. A new color palette for dark mode and a theme switcher
  between auto / dark / light.
• Progressive Web App support
• Extraction of Fediverse content
• Directly save images
• HLS video player

Improved

• Accept TLS 1.0 and negotiate more ciphers in the extractor HTTP client
• Selected font is applied on bookmark's title and description
• Regrouped user preferences and admin under the same menu item
• Mark required form fields
• Better password reveal widget

• Better newyorker.com configuration
• Improved reddit.com content extraction (pictures, galleries, self text, links)
• Enabled HTTP/2, compression and keep-alive on the extractor HTTP client
• Ensure images from articles are properly requested to avoid edge cases

• Always keep not empty inline nodes during cleanup. Some can be used for punctuation purposes.

Added

• Sharing saved bookmarks. Thanks @bacardi55

Improved

• Better display of images in lists (when used as bullets by some contents)
• Improved display of inner links in titles
• Save more images from articles
• Apply Wikipedia extraction rules to Wikinews
• Added robots rules on all the app's pages (noindex, nofollow, noarchive)
• Restrict CSP frame-src rule to video provider when a bookmark is a video
• Force youtube videos to play through the "nocookie" variant
• Escape key on menu-like UI elements
• Reduced title sizes. Thanks @joachim
• More tests
• Container for arm64 architecture

Fixed

• Removed the "summary" element marker on Safari. Thanks @joachim
• Fixed an extraction bug with code formatting in articles

• Content Script for BNF websites going through oclc.org. This is mainly for the extension where a user can save content on mediapart and arretsurimage while using the BNF portal.
• When readability is disabled, keep performing the pre and post cleaning process
• Much better Wikipedia extraction. Thanks @joachimesque
• Remove some absolute image sources that were not playing well when behind a reverse proxy. Thanks @JerryWham and @franckpaul
• Added a global READECK_USE_X_FORWARDED environment variable that should be set when running Readeck in a container behind a reverse proxy.

Fixed

• Presentation bug with very long numbered lists
• Typo in the main menu, by @denis_defreyne
• Fixed a bug preventing loading all the site configuration rules

• Added more API documentation and some examples

Fixed

• Fixed API documentation "try" mode not working
• Fixed a bug when creating a bookmark with cached resources

• Fixed a bug preventing the execution of the main content script
• Fixed the newyorker.com site configuration

• Updated site config files
• Updated dependencies

Fixed

• Fixed a CSS print display bug with QR Codes

• Bookmark print stylesheet
• Display QR codes in print and ebook
• Better youtube transcript retrieval

• Compress HTTP responses for some content types

Fixed

• Fixed metric name for bookmark_resources_total

• Content Script system. This includes the five filters rulesets and improves the previous rules system with a documented and much safer JS API to override predefined rules or create complex workflows
• Video transcripts for youtube, vimeo and TED
• Content script for substack
• Store the video duration when available
• Extract text direction, when available and apply it when showing an article's content
• Added a scroll indicator on articles, by @joachimesque
• Added a quick access menu for keyboard navigation, by @joachimesque
• API documentation (not full yet)
• Prometheus metrics

Improved

• Shorten the bookmark card's title to avoid disgraceful overflows
• Prefixed every id and name (on a tags) attribute on extracted articles
• Lots of CSS improvement, by @joachimesque
• Improved the collection creation workflow
• Refactored and improved the bookmark storage system (non breaking)

Fixed

• Better initial content cleanup
• Fixed Safari Mobile layout, by @joachimesque

Added

• New, web based, onboarding process
• Code of Conduct

Fixed

• Default listening port is now 8000
• EPUB filenames length limit

• Improved bookmark creation form
• Links are extracted in the background so a bookmark is visible earlier
• Bookmark creation from the API can now receive a label list
• Improved performances of log-in with an application password
• Configuration/env/flags priority for command line
• Bookmark order in collection's ebook

Fixed

• Send a CSRF secure cookie only if scheme is https
• Search string must keep invalid fields and treat them as quoted values
• Major improvements in Print Stylesheet (by @joachimesque)
• Instagram picture title and description