- Remember Hide Known/Hide Kana filter state in localStorage across sessions
- Detect PNG and WebP cover art MIME types instead of hardcoding image/jpeg
- Use configured AnkiConnect URL for the browse action
- Deduplicate concurrent in-flight alass retime calls via promise caching
- Prefer request-provided secondary subtitle text over retimed sidecar fallback
- Fix cover image record key types from string to number
- Retime local English sidecars against the Japanese sidecar via alass before populating sentence card translation fields; cache retimed copies for the process lifetime
- Reject reversed or non-positive subtitle timings in immersion tracker and before media generation
- Fix word card mining so sentence audio goes to SentenceAudio (not ExpressionAudio) and English subtitle text is not written to SelectionText
- Consolidate 10 individual change fragments into stats-updates.md
- Search by headword enabled by default; finds inflected variants (e.g. 知らない → 知らねえ)
- Add "Search by headword" toggle to switch back to exact text/title matching
- Rename "Similar Words" → "Related Seen Words" with tighter matching (same reading/shared kanji)
- ankiConnect.deck falls back to Yomitan mining deck when empty
- Remove overlay mining image toast; OSD card notifications no longer flash a frame screenshot
- Add Hide Kana toggle to frequency rank table to filter kana-only headwords
- Fix vocab exclusions to deduplicate and match token variants (e.g. ない / 無い) under one exclusion entry
- Skip cover image fetching when the overview tab is inactive
- Remove UTC-based UNION branch from getRollupGroupsForSessions; keep only localtime rollup
- Add resolveSentenceSearchLimit to clamp/sanitize infinity and negative limit values
- Switch onMiningImageEvent from queued to latest-value IPC listener (drop stale frames)
- Add regression tests for rollup key locality and sentence search limit edge cases
- Add Search tab with realtime subtitle sentence search, media context, and mining actions
- Persist library card size selection across reloads
- Fix selection text written only for sentence cards, not word/audio cards
- Fix word mining audio routed to SentenceAudio field when present
- Fix sentence card created immediately before slow media generation completes
- Fix Anki config resolved at request time for sentence mining
- Prefer non-Signs/Songs tracks when auto-selecting secondary subtitle language
sudacode
