Codex

Keep agent worktrees out of the Docker build context; Stop a session that moved nothing from rewriting reading progress; Report a concurrently replayed reading session as a duplicate

State scoped imports as coverage, and carry want-to-read across

Let an import actually use external ids, and scope it to a library

Show reading of deleted books, and a confirmed way to delete it; Add a Reading Progress settings page for export and import; Count reading of deleted books, and let a reader purge it

Validate a hand-written filter condition against the field catalog; Add a JSON view of the filter rule, with copy and paste; Show the rule JSON panel under the filter builder

Stop a parked tab from clobbering progress read elsewhere

Honor the sort parameter on unsearched book lists

Codex 2.4.14 released.

Gate progress endpoints on the progress permissions; Require series:read on want-to-read, recommendations, and series exports; Unify the permission wire format and expose plugins-manage in the picker

Move the plugin publish jobs to npm 12

Use the lockfile in the cross-build image too

Install the frontend from the lockfile, not a live re-resolve

Size the webtoon seam overlap to a full device pixel under zoom-out

Measure the away gap from the pause, not the last page turn

Keep 8-bit alpha in the favicon's 16px frame

Switch the app icon to the ice+ember branding

Bucket reading activity by the viewer's day instead of UTC

Overlap webtoon pages 1px to stop sub-pixel seams at zero gap

Let a reader drop a series reading direction back to the inherited one

Stop losing the renumber pass a book was waiting on

Give the sorts with guaranteed ties a defined order; Finish giving every ordered list a defined total order; Order on-deck and the error list past their ties

Store per-user reader settings for a series; Carry per-series reading directions over to the account; [breaking] Sync only reading direction per series, not every reader setting

[breaking] Report live scan progress, and purge in seconds not days; Stop book search multiplying its page offset by the page size; Lock every series metadata field the write endpoints touch

Stop per-book analysis racing on positional book numbers

Serve the archive entry the page row names, not the one at that index; Stop dropping AVIF pages at the extension gate; Stop webtoon pages deadlocking on their own lazy load

Remove the duplicate header left by the 2.2.1 release; Make the auth pages follow the color scheme instead of forcing dark; Stop borrowing the book count as a volume or chapter numerator

Make the reading-session rules a shared file instead of test code; Document the preference keys that exist and stop mocking ones that don't

Make the book file endpoint range-capable and conditional; Describe the full response shapes as their own routes; Describe series pages, library book params, and the filter grammar correctly

Keep OIDC login state in the database so any replica can complete it; Stop plugin OAuth and entity events from being trapped in one process

[breaking] Layer configuration with figment and nest env vars with __; Accept the environment value shapes the override layer allowed; [breaking] Make the ad-hoc environment reads real configuration keys

Make webtoon page requests honour the downscale setting; Stop the webtoon reader mutating the DOM when a scroll settles

Disclose sittings that reported no page count

Let the reading dashboard measure pages and books finished; Add calendar-year and all-time ranges to the reading dashboard; Stop the reading dashboard misnaming and mis-ranking series

Keep the activity calendar populated on the one-year range

Make reading statistics work on PostgreSQL

Derive reading progress from an append-only session log; Add a batched endpoint for recording reading sessions; Measure reading time and pages read in all three readers

Retry the initial connection when running migrations

Stop starting processes from exhausting the connection limit

Scaffold the mangabaka recommendations plugin; Generate mangabaka recommendations from library seeds; Keep franchise spam out of mangabaka recommendations

Show a real label for every active task

Serve collection and read list covers as image bytes

Reorder filter builder conditions by dragging; Honour the sort parameter on a collection's series; Honour the sort parameter on a read list's books

Explain each release action in its tooltip; Add rating filters and multi-library selection; Resolve rule-backed collection membership at read time

Migrate mangabaka to the api.mangabaka.org host

Expose the want-to-read queue as a virtual collection and read list

Make the mobile search sheet scrollable above the iOS keyboard

Add summary as a filterable field for series and books; Add webLinks manifest capability schema; Add plugins/web-links endpoint resolving webLinks templates

Add collection and read-list membership fields to data exports; Show all catalog fields in the export modal field groups

Shared media grid with sort selectors and drag-and-drop reorder; Manual ordering everywhere, collection summaries, persisted queue sort; Honor the ordered flag when listing collection and read list members

Drop unsafe optional chaining flagged by biome in tests

Persist incognito mode across chapter boundaries

Keep book completion sticky in double-page mode

Bound concurrent image decode to prevent serve OOM

Bridge task progress across processes via LISTEN/NOTIFY

Add --progress and default row-count verification; Add --full-verification report for import and copy; Cap insert batches to the destination bind-parameter limit

Read SQLite UUIDs stored as text, not just blobs

Add entity-driven data transfer engine; Add portable export archive with artifact bundling; Add export, import, and copy database commands

Decode mislabeled Windows-1252 ComicInfo.xml

Mark the last book of a series read at the end-of-book overlay

Enable swipe navigation to and on the chapter overlay

Add opt-in page downscaling for smoother turns; Warn that page downscaling is slow and not a flicker fix

Decode preloaded pages to stop swipe-in black flicker; Only decode neighbor pages to prevent mobile OOM crash; Drive swipe track transform imperatively to kill commit flash

Fix swipe page-turn flash and missing snap-back

Swipe down to exit the comic reader; Prevent swipe page-turn from flashing the previous page

Add swipe-paging gesture helpers and setting; Add finger-drag swipe pager and drag input; Enable swipe paging in the comic reader

Make auto-advance countdown configurable and fix webtoon loop; Lock page scroll while the mobile nav menu is open; Keep webtoon scroll restore from drifting backward on reload

Enrich adjacent-book data for chapter-transition panels; Add chapter-transition panel and auto-advance countdown; Show chapter-transition panels in the webtoon reader

Use npm ci for web job and harden cron /n parsing; Use npm ci across web and plugin jobs, harden cron /n parsing

Add detail hover cards to series and book cards

Refresh want-to-read cards and reading feeds without reload

Add multi-select and bulk actions to advanced search; Add Want to Read toggle to card menu; Add bulk add endpoint and toolbar action

Add collections, read lists, and want-to-read schema; Add collection/read-list permissions and repositories; Add want-to-read queue endpoints

Add paginated ReadingFeedSection for keep-reading and on-deck; Add Keep Reading and On Deck tabs with See all links; Show hasUserRating and bookType in active filter chips

Share existing_books_map via Arc to stop scan OOM; Bound image memory to prevent decode OOM

Add IdP bearer token validator and per-provider accepted audiences; Accept IdP bearer tokens on the API and resolve linked users

Anchor webtoon scroll position when images load above the viewport

Add host-configurable plugin log level; Add debug logging across network-backed plugins; Per-plugin log level override, default to server level

Break reverse-RPC deadlock during plugin initialize

Scaffold tsundoku release-source plugin; Add tsundoku feed fetcher and cursor persistence; Add tsundoku external-id matching and candidate mapping

Codex 1.36.1 released.

Add data layer for scheduled user-plugin sync; Fan out scheduled user syncs on a plugin's cron; API for admin sync cron and per-user auto/manual sync

Send library context with sync and recommendation entries; Scope sync push and pull to a plugin's allowed libraries; Scope recommendation generation to a plugin's allowed libraries

Issue refresh tokens for OIDC logins; Store OIDC refresh token to keep SSO sessions alive

Make DB fan-out bound and background pool size configurable; Eliminate list N+1 and SSE-driven refetch storm; Bound per-request DB query fan-out to prevent pool exhaustion

Serve API docs for bare /docs path under active service worker

Delete series external links by ID instead of source name

Add created_at and external ratings to LLM presets

Sortable inbox columns and split observed/released dates

Make sidebar scroll when libraries overflow the navbar

Add slim series external-index endpoint for discovery tools; Update openapi-sync paths for crate-split layout

Repair advanced search and filters on mobile

Add per-group search input for long chip lists

Add schema, entities, and repository for sharing-tag access groups; Merge group grants into ContentFilter; Add admin API for access group management

Decouple browser RUM proxy endpoint from SDK exporter

Point rust-embed at workspace-root web/dist

Add opt-in OTLP traces and metrics scaffolding; Instrument plugin RPCs, repositories, scanner, and task worker with OTel spans; Emit OTel counters, histograms, and gauges alongside in-app metrics

Unwire sccache from repo config to keep CI/Docker green; Enable sccache wrapper and document fast-build tooling

Keep session through transient refresh-token failures

Show global presets, allow saving from draft, and offer replace on duplicate

Add unified filter-preset storage for saved filters and advanced search; Add path, format, year, page count, date, and author filter variants; Compose fuzzy index with /list endpoints and add relevance sort

Trust-list external-ID sources, hydrate members in list endpoint; Align series duplicate mocks with hydrated members schema; Stabilize EPUB pointer test against extra ReactReader renders

Detect series duplicates by external ID or normalized title; Add Series tab to duplicate detection UI; Show book title on duplicate cards and prevent overflow

Batch DTO hydration in fuzzy search path

Add in-memory fuzzy index over series and books; Keep fuzzy index live via event-driven incremental updates; Route global search through the fuzzy index behind a runtime flag

Persist refresh tokens with rotation and theft detection; Add /auth/refresh and wire refresh tokens into login/logout; Refresh access tokens transparently on 401 in the web client

Structured tracing for PDF page render path; Add in-memory open-document handle cache service; Route page renders through in-memory handle cache

Expose Ignore and Reset as per-row actions

Add motion provider, easings, useDelayedFlag, and depth tokens; Shadow-based depth and translucent mobile surfaces; Motion layer with press, stagger, theme toggle, and reduced-motion guard

Foundation pass — xs breakpoint, drawer auto-close, overflow fixes; Full-screen search sheet below the xs breakpoint; Reader polish with overflow menu and bottom navigation bar

Show tracking badge and toggle on series cards

Add async BulkTrackForReleases task type and handler; Wire bulk track/untrack endpoints to async task; Switch bulk track/untrack toolbar to queue-then-await pattern

Support shift+click range selection in release tracker

Configurable RPC request timeout per call and per plugin

Auto-ignore releases on disabled axes

Add tri-state isTracked filter for release-tracking series; Plugin progress reporting + aggregated release toasts

Poll-all + reset-all admin actions, plugin FK cascade

Embed frontend in installer-built binaries

Hide Releases nav entry when no release-source plugin is enabled

Harden release-tracking and library-jobs scenarios; Preserve search title in get() responses

Add series_tracking sidecar and series_aliases schema; Add backfill task, tracking API, and UI surface; Add release ledger schema, repos, and HTTP API

Emit progress events from nine more task handlers; Show target titles and live elapsed time in active-tasks badge; Render progress bar and message for live tasks in active-tasks badge

Add per-library scheduled refresh config storage; Add scheduled per-library refresh task type and handler; Add field-group resolver and dry-run mode to series metadata apply

Split total_book_count into separate volume and chapter counts; Add total_volume_count and total_chapter_count to plugin protocol; Apply total_volume_count and total_chapter_count via MetadataApplier

Add series_exports data model, migration, and repository; Add ExportStorage service for atomic file writes; Add series export data collector and bulk rating helpers

Add Continue/Read button to series detail page; Add metadata fields, plugin-side filters, and frontend filter UI

Fix cross-app progress sync when within-resource progression is zero

Add KOReader sync API for reading progress synchronization; Add R2Progression support for cross-device EPUB reading sync; Add server-side Readium positions for cross-app progression sync

Prevent webtoon progress stuck at 99% when navigating to next volume

Add separate fit mode for webtoon vs comic reader

Make webtoon boundary nav arrow-key-only and hide layout toggle; Mark book as read when navigating to next volume

Add IANA timezone support for cron job scheduling; Add accent-insensitive Unicode search via search_title column; Prevent panic on series list with readStatus condition filter

Add book context support for custom metadata templates; Add inline helpers, improved error feedback, and interactive help section; Add select all and fix selection persistence across navigation

Add application.base_url as unified external URL setting; Add centralized data_dir config and plugin file storage; Add admin plugin storage management page

Handle SQLite blob UUIDs and PostgreSQL INT8 in consolidate_authors

Make consolidate_authors migration idempotent for partial failure recovery

Add many-to-many genre/tag support, custom metadata editor, and title_sort_lock; Enrich BookDetail page with full metadata and generic badge chips; Add externalIds support to PluginBookMetadata for book cross-references

Redesign recommendation cards with float-based grid layout; Add hasUserRating filter and fix rating sort for PostgreSQL; Add user-scoped plugin task endpoint for sync status visibility

Move settings to server-side with per-user seed curation; Remove duplicate "Based on" text in recommendation cards; Only exclude read series and enrich instead of filtering local matches

Add glob pattern support for rate limit exempt paths; Display description preview in metadata search results; Correct series download button endpoint URL

Add searchURITemplate to plugin manifests with external search link in metadata modal; Add dynamic document titles to series, book, reader, and search pages; Add collapsible overflow for tags and alternate titles on series detail page

Rewrite plugin scenarios for official store and update docs; Clamp mangabaka search limit to 50, add sort_by config, handle null JSON-RPC result

Add series metadata reset endpoint with bulk support and UI; Add official plugin store with 3D flip card carousel

Make GET and dismiss endpoints non-blocking with SSE task tracking; Add internal_config with search results limit and external ID lookup in search modal; Add series external ID CRUD endpoints and edit modal for series/books

Add configurable task timeout via DB settings; Add private and hiddenFromStatusLists visibility controls to AniList sync; Renumber series books on delete and restore, clear deleted book numbers

Sort search results by title similarity scoring; Stop recording reading progress on v1 page fetch, move to OPDS-only endpoint; Swap sort by and page size positions in library toolbar

Add user plugin system database schema and core infrastructure; Add OAuth authentication flow and user plugin API endpoints; Add plugin storage system with bidirectional RPC and SDK support

Use COALESCE for title sort in v1 series endpoints to handle NULL title_sort values

Set page to last page on mark-as-read and sort on-deck by recency; Correct progress bar click mapping in RTL mode; Use COALESCE for title sort to handle NULL title_sort values

Auto-detect book completion when current page reaches page count

Add OIDC configuration and data model; Implement OIDC service core; Add OIDC API endpoints

Populate genres, tags, links, authors, and alternate titles in series endpoint; Populate book metadata fields (authors, tags, summary, dates)

Improve double-page spread handling and RTL progress bar; Fix double-page spread initialization and page jump on load

Enhance plugin details with two-column layout, metadata targets, and search config indicators

Codex 1.8.1 released.

Add comprehensive book metadata system with plugin support; Add sidecar metadata support for OPF and Mylar series.json; Correct credential_delivery CHECK constraint values for PostgreSQL

Add sort fields, condition filters, and integration tests for Komic compatibility

Handle rate-limited tasks with rescheduling instead of retries

Add missing series read-progress endpoints; Prevent navigation when triggering library menu actions

Book and series search with library filter

Use package.json version in manifests instead of hardcoded values

Change default books per page to 50 in series detail; Add book count to metadata search results; Add cover lock to preserve user cover selection during auto-fetch

Add GET /api/v1/user endpoint for current user profile; Add ZeroPages error type for books with no pages; Add komga-sync tool for reading progress migration

Add external source ID filter and improve metadata display; Add config schema support and improve plugin logging; Handle claim race condition gracefully instead of logging error

Add SSE events for plugin configuration changes; Add preprocessing rules, search templates, auto-match conditions, and external ID tracking; Support custom user directive in docker-compose

Upgrade Mantine to v8 and refine card styling; Add series completion filter; Update title_sort automatically when title changes via plugin

Add plugin system for external metadata providers; Add plugin metrics tracking and scheduled thumbnail generation; Add bulk operations and info modals for books and series

Add automated screenshot capture system with Playwright; Show focus ring only for keyboard navigation; Fix PDF fit modes not working on initial page load

Multi-format support: CBZ, CBR (with UnRAR), EPUB, and PDF; Dual database support: SQLite for simplicity, PostgreSQL for scale; Library management: Multiple libraries with customizable scanning strategies