- Load persisted known-word cache on startup; reconcile adds/deletes/edits on timed sync
- Add `knownWords.addMinedWordsImmediately` (default `true`) for immediate mined-word updates
- Route full rebuild to explicit `subminer doctor --refresh-known-words` and expand tests/docs
- Run `subminer stats -b` as a dedicated daemon process, independent from the overlay app
- Stop Anki progress spinner before showing final `✓`/`x` mine result so it is not overwritten
- Keep grammar/noise subtitle tokens hoverable while stripping annotation metadata
- Add note ID resolution through merge redirects in stats API
- Build Anki note previews using configured field names
- Add session event helpers for merged note dedup and stable request keys
- Refactor SessionDetail to prevent redundant note info requests
- Add session event popover and API client tests
- Extract word field config into reusable anki-field-config module
- Add ankiConnect.fields.word config option (default: "Expression")
- Replace hardcoded "Expression" field references across Anki integration
- Add note ID redirect tracking for merged/moved cards
- Support legacy ankiConnect.wordField migration path
Implement `subminer stats -b` to start a background stats daemon and
`subminer stats -s` to stop it, with PID-based process lifecycle
management, single-instance lock bypass for daemon mode, and automatic
reuse of running daemon instances.
Add comprehensive query helpers for session deletion with word aggregate
refresh, known-words-per-session timeline, anime-level word summaries,
and trends dashboard aggregation. Track yomitanLookupCount in session
metrics and support bulk session operations.
Add configurable keybinding to mark the current video as watched with
IPC plumbing between renderer and main process. Add event listener
infrastructure for tracking Yomitan dictionary lookups per session.
- Filter out 感動詞 (interjection) POS1 tokens from annotation payloads
- Exclude common interjection terms (ああ, ええ, はあ, etc.)
- Exclude reduplicated kana SFX with optional trailing と
- shouldExcludeTokenFromSubtitleAnnotations checks both POS1 and term patterns
- filterSubtitleAnnotationTokens applied after annotation stage
- POST /api/stats/mine-card endpoint with mode=word|sentence|audio
- mode=word: creates full Yomitan card (definition/reading/pitch) via hidden search page bridge
- mode=sentence/audio: creates card directly with Lapis/Kiku flags
- Audio + image generated in parallel from source video via ffmpeg
- Respects all AnkiConnect config (AVIF, static, field mappings, metadata pattern)
- addYomitanNoteViaSearch calls window.__subminerAddNote exposed by Yomitan fork
- Syncs AnkiConnect URL to Yomitan before each word mine
- Add secondary_text column to imm_subtitle_lines with migration
- Pass currentSecondarySubText through recordSubtitleLine flow
- Include secondaryText and sourcePath in word/kanji occurrence queries
- Update all type interfaces (backend + frontend)
- Stats server auto-start on immersion tracker init
- Stats overlay toggle via keybinding and IPC
- Stats CLI command (subminer stats) with cleanup mode
- mpv plugin menu integration for stats toggle
- CLI args for --stats, --stats-cleanup, --stats-response-path
- Add Arch Linux `subminer-bin` install instructions to README and docs-site installation guide
- Add Hyprland troubleshooting guidance for transparency and global shortcut passthrough
- Add TASK-159..164 backlog records and skill design plans; include changelog fragment
- Apply formatting-only update to `src/release-workflow.test.ts`
- add `verify:config-example` script with tests to fail on missing/stale generated config artifacts
- run the verification in CI and release workflows, and document it in release/docs guidance
- fix docs-site Cloudflare Pages watch path to `docs-site/*` with regression coverage
