fix(tokenizer): stop partial furigana readings from marking words known (#137)

.gitignore

@@ -1,5 +1,6 @@
 # Dependencies
 node_modules/
+/package-lock.json
 
 # Superpowers brainstorming
 .superpowers/
@@ -16,6 +17,8 @@ coverage/
 
 # Launcher build artifact (produced by make build-launcher)
 /subminer
+/main-entry.js
+/main-entry.js.map
 
 # Logs
 *.log

CHANGELOG.md

@@ -1,5 +1,55 @@
 # Changelog
 
+## v0.17.2 (2026-06-28)
+
+### Fixed
+- YouTube Background Cache: Fixed Windows YouTube background media cache startup for YouTube URLs opened directly in mpv, including resolved stream URLs when mpv still exposes the original YouTube playlist entry, so queued Anki media updates can append audio and images after the cache finishes.
+- YouTube Subtitle Picker: Manual subtitle picker requests now show an immediate configured notification while SubMiner probes tracks and opens the modal. Subtitle download progress is replaced with a transient success notification after tracks load.
+
+## v0.17.1 (2026-06-27)
+
+### Added
+- YouTube Media Cache Mode: Adds `youtube.mediaCache.mode` with `direct` and `background` options. Background mode uses a yt-dlp cache download when direct stream extraction is unreliable — creates a text-only card immediately, queues media updates for mined notes, and fills audio/image fields once the download finishes. Progress is announced via overlay/OSD notifications. Downloads are capped at 720p by default (`youtube.mediaCache.maxHeight`). Switching back to direct mode cancels any in-flight background download.
+
+### Fixed
+- Log Export: Fixed log filenames to use the local date so exports around UTC midnight include the current day's logs rather than stale prior-day files. Expanded export redaction to mask IPs, emails, auth and cookie headers, yt-dlp cookie arguments, URL credentials, token/key/password fields, and signed YouTube media URL parameters.
+- YouTube Card Media: Improved media generation reliability by sending safer ffmpeg options for resolved streams and skipping stale stream maps (including cached YouTube files). Hardened background cache downloads with IPv4 and extractor retry flags; failed downloads now notify the user and clear queued media updates instead of leaving them silently pending. Stale background cache files are cleaned on startup and before each new download.
+
+## v0.17.0 (2026-06-15)
+
+### Changed
+
+- **Subtitle Delay Keybindings**: Updated default overlay subtitle delay and step bindings to match mpv conventions: `z`, `Z`, and `x` adjust `sub-delay`; `Ctrl+Shift+Left/Right` run native `sub-step` with OSD feedback. The previous SubMiner-only adjacent-cue delay action has been removed.
+- **Update Notifications**: New installs now default update notifications to overlay-only instead of overlay + system notifications.
+
+### Fixed
+
+- **Anki – Highlight Word**: Fixed bolding of the mined word in Kiku sentence and sentence-furigana fields when the source Yomitan sentence did not already contain bold markup.
+- **Anki – Lapis/Kiku Word Cards**: Fixed word-and-sentence marker missing from Lapis/Kiku word cards enriched through SubMiner, which could hide sentence context on the card front.
+- **Anki – Windows Media Generation**: Fixed known-word cache refreshes when no deck is configured, and fixed audio/image generation after background launches by recreating missing FFmpeg temp directories before clipping.
+- **Character Dictionary – Windows**: Fixed the Windows "SubMiner mpv" shortcut so character dictionary auto-sync can fall back to mpv's current video path when app media state is not yet ready.
+- **Notifications**: Restored the SubMiner app icon on system notifications that do not supply a custom notification image.
+- **Overlay – macOS Yomitan Popup**: Fixed Yomitan popup focus after card mining or popup reload; fixed popups staying open when clicking transparent overlay space — click-away now closes the popup and returns click passthrough to mpv without a hide/reappear cycle.
+- **Overlay – Linux Auto-Pause Startup**: Fixed the visible overlay on Linux auto-paused startup to stay interactive during the initial measurement gap; startup subtitle cache misses now paint raw text before tokenization finishes, and temporarily empty `sub-text` refreshes are resolved before warm readiness resumes playback.
+- **Overlay – Playlist Advance**: Fixed the visible overlay being dismissed when mpv advances to the next playlist item, including when the next episode loads after the warm transition delay.
+- **Overlay – Windows Subtitle Bar**: Fixed shaky hover and click interaction on the Windows subtitle bar when a video attaches to an already-running background SubMiner instance.
+- **Stats – AniList Linking**: Fixed manual AniList linking from the stats anime page so auto-searches strip the generated "Season N" suffix and query only the anime title.
+- **Updates – Linux Support Assets**: Fixed Linux updates to correctly install and refresh the launcher runtime plugin copy and rofi theme alongside AppImage and launcher updates; unrelated SubMiner data directories are now left untouched and plugin copies are staged before replacing the live runtime. Fixed first-launch playback on fresh installs by auto-installing missing managed support assets from the bundled app before mpv starts.
+
+### Docs
+
+- **Linux Update Flows**: Documented that Linux update flows manage the launcher runtime plugin copy and rofi theme from `subminer-assets.tar.gz`, and that normal launcher playback auto-installs those managed support assets on a fresh install if either is missing.
+
+<details>
+<summary>Internal changes</summary>
+
+### Internal
+
+- **Runtime Modules**: Split main-process runtime wiring into focused modules without changing user-facing behavior; hardened helpers against stale background stats daemon PIDs, stalled subtitle extraction, and dropped async errors.
+- **Release CI**: Fixed GitHub release notes to preserve the `What's Changed` and `New Contributors` attribution sections when CI regenerates from the committed changelog; scoped prerelease note reuse to the same base version so a new beta line starts from current fragments.
+
+</details>
+
 ## v0.16.0 (2026-06-10)
 
 ### Breaking Changes

README.md

@@ -213,6 +213,7 @@ On **Windows**, just run `SubMiner.exe` and the setup will open automatically on
 subminer video.mkv          # launch mpv with SubMiner
 subminer /path/to/dir       # pick a file with fzf
 subminer -R /path/to/dir    # pick a file with rofi (Linux only)
+subminer -H                 # browse local watch history (replay / next episode / browse)
 ```
 
 On **Windows**, use the **SubMiner mpv** shortcut created during setup. Double-click it or drag a video file onto it.

changes/audio-normalization.md

@@ -0,0 +1,4 @@
+type: fixed
+area: mining
+
+- Normalized generated card audio by default during media extraction, with `ankiConnect.media.normalizeAudio` available to keep raw source loudness when needed.

changes/desktop-notification-icons.md

@@ -1,4 +0,0 @@
-type: fixed
-area: notifications
-
-- Restored the SubMiner app icon on system notifications that do not provide a custom notification image.

changes/history-rofi-cover-art.md

@@ -0,0 +1,4 @@
+type: added
+area: launcher
+
+- Show cover art icons in the rofi watch-history picker, reusing AniList covers already stored in the stats database (extracted to `~/.cache/subminer/covers`).

changes/known-word-partial-reading-highlight.md

@@ -0,0 +1,4 @@
+type: fixed
+area: overlay
+
+- Fixed never-mined compound words (e.g. 待ち合わせてる) being highlighted green as known: subtitle tokens now carry complete readings instead of kanji-only furigana joins, and the known-word reading fallback rejects readings that don't cover the token surface. Stored word readings in the stats database are no longer truncated for new lines.

changes/launcher-history-command.md

@@ -0,0 +1,4 @@
+type: added
+area: launcher
+
+- Added `subminer -H` / `--history` to browse local watch history, replay the last watched episode, continue to the next episode, or browse episodes with fzf/rofi.

changes/linux-support-assets-updates-docs.md

@@ -1,4 +0,0 @@
-type: docs
-area: updates
-
-- Documented that Linux update flows manage the launcher runtime plugin copy and rofi theme from `subminer-assets.tar.gz`, and that normal launcher playback auto-installs those managed support assets if either one is missing.

changes/linux-support-assets-updates.md

@@ -1,6 +0,0 @@
-type: fixed
-area: updates
-
-- Fixed Linux updates so the managed support-asset install now creates and refreshes both the launcher runtime plugin copy and the rofi theme alongside AppImage and launcher updates.
-- Fixed Linux support-asset refreshes so unrelated SubMiner data directories are left alone and plugin copies are staged before replacing the live runtime plugin.
-- Fixed first Linux launcher playback on fresh installs by auto-installing the managed runtime plugin copy and rofi theme from the bundled app before mpv starts when either support asset is missing.

changes/macos-yomitan-popup-focus.md

@@ -1,5 +0,0 @@
-type: fixed
-area: overlay
-
-- Fixed macOS Yomitan popup focus after card mining or popup reload while still allowing click-away to close the popup without a hide/reappear cycle.
-- Fixed macOS Yomitan popups staying open when clicking transparent overlay space; click-away is captured for popup close, then passthrough returns to mpv.

changes/overlay-autopause-startup-subtitle.md

@@ -1,4 +0,0 @@
-type: fixed
-area: overlay
-
-- Fixed auto-paused Linux visible overlay startup so the overlay stays interactive during the first measurement gap, startup subtitle cache misses paint raw text before tokenization finishes, and temporarily empty mpv `sub-text` refreshes parsed cues before synthetic warm readiness can resume playback.

changes/overlay-mpv-subtitle-keybindings.md

@@ -1,4 +0,0 @@
-type: changed
-area: overlay
-
-- Updated default overlay subtitle delay/step bindings to match mpv: `z`, `Z`, and `x` adjust `sub-delay`; `Ctrl+Shift+Left/Right` run native `sub-step` and show subtitle delay on the OSD. Removed the old SubMiner-only adjacent-cue delay action.

changes/overlay-playlist-visible-state.md

@@ -1,4 +0,0 @@
-type: fixed
-area: overlay
-
-- Kept the visible overlay active while mpv advances to the next playlist item, even when the next episode loads after the warm transition delay.

changes/release-notes-attribution-placement.md

@@ -1,4 +0,0 @@
-type: fixed
-area: release
-
-- Kept the GitHub release `What's Changed` and `New Contributors` attribution sections when CI regenerates release notes from the committed changelog.

changes/split-main-runtime-modules.md

@@ -1,5 +0,0 @@
-type: internal
-area: runtime
-
-- Split main-process runtime wiring into focused modules without changing user-facing behavior.
-- Hardened split runtime helpers against stale background stats daemon PIDs, stalled subtitle extraction, and dropped async errors.

changes/stats-anilist-title-search.md

@@ -1,4 +0,0 @@
-type: fixed
-area: stats
-
-- Fixed manual AniList linking from the stats anime page so automatic searches drop the generated `Season N` suffix and search only the anime title.

changes/update-notification-default-overlay.md

@@ -1,4 +0,0 @@
-type: changed
-area: updates
-
-- New installs now default update notifications to overlay-only instead of overlay + system notifications.

changes/youtube-stream-edl-options.md

@@ -0,0 +1,4 @@
+type: fixed
+area: youtube
+
+- Fixed direct YouTube stream media extraction by parsing mpv EDL stream URLs with their byte-length guards, preventing trailing EDL segment options from corrupting signed googlevideo URLs and causing ffmpeg 403 errors.

config.example.jsonc

@@ -559,6 +559,7 @@
       "animatedMaxHeight": 0, // Maximum height for animated AVIF captures, in pixels. Set to 0 to preserve aspect ratio.
       "animatedCrf": 35, // Animated AVIF CRF quality target. Lower values produce larger, higher-quality files.
       "syncAnimatedImageToWordAudio": true, // For animated AVIF images, prepend a frozen first frame matching the existing word-audio duration so motion starts with sentence audio. Values: true | false
+      "normalizeAudio": true, // Normalize generated sentence audio loudness during media extraction. Values: true | false
       "audioPadding": 0, // Seconds of padding appended to both ends of generated sentence audio and animated AVIF clips.
       "fallbackDuration": 3, // Fallback clip duration in seconds when subtitle timing data is unavailable.
       "maxMediaDuration": 30 // Maximum allowed media clip duration in seconds.
@@ -618,7 +619,11 @@
     "primarySubLanguages": [
       "ja",
       "jpn"
-    ] // Comma-separated primary subtitle language priority for managed subtitle auto-selection.
+    ], // Comma-separated primary subtitle language priority for managed subtitle auto-selection.
+    "mediaCache": {
+      "mode": "direct", // How YouTube card audio/images are extracted. Values: direct | background
+      "maxHeight": 720 // Maximum video height downloaded for the YouTube background media cache. Set to 0 for unlimited.
+    } // Media cache setting.
   }, // Defaults for managed subtitle language preferences and YouTube subtitle loading.
 
   // ==========================================

docs-site/anki-integration.md

@@ -161,13 +161,14 @@ Audio is extracted from the video file using the subtitle's start and end timest
 "ankiConnect": {
   "media": {
     "generateAudio": true,
+    "normalizeAudio": true,      // normalize generated clip loudness
     "audioPadding": 0,           // optional seconds before and after subtitle timing
     "maxMediaDuration": 30       // cap total duration in seconds
   }
 }
 ```
 
-Output format: MP3 at 44100 Hz. If the video has multiple audio streams, SubMiner uses the active stream.
+Output format: MP3 at 44100 Hz. If the video has multiple audio streams, SubMiner uses the active stream. Generated sentence audio is loudness-normalized by default during extraction; set `normalizeAudio` to `false` to keep raw source loudness.
 
 The audio is uploaded to Anki's media folder and inserted as `[sound:audio_<timestamp>.mp3]`.
 
@@ -347,6 +348,7 @@ When you mine the same word multiple times, SubMiner can merge the cards instead
       "imageType": "static",
       "imageFormat": "jpg",
       "imageQuality": 92,
+      "normalizeAudio": true,
       "audioPadding": 0,
       "maxMediaDuration": 30,
     },

docs-site/changelog.md

@@ -1,6 +1,61 @@
 # Changelog
 
-## v0.16.0 (2026-06-10)
+## v0.17.2 (2026-06-28)
+
+**Fixed**
+- YouTube Background Cache: Fixed Windows YouTube background media cache startup for YouTube URLs opened directly in mpv, including resolved stream URLs when mpv still exposes the original YouTube playlist entry, so queued Anki media updates can append audio and images after the cache finishes.
+- YouTube Subtitle Picker: Manual subtitle picker requests now show an immediate configured notification while SubMiner probes tracks and opens the modal. Subtitle download progress is replaced with a transient success notification after tracks load.
+
+## v0.17.1 (2026-06-27)
+
+**Added**
+- YouTube Media Cache Mode: Adds `youtube.mediaCache.mode` with `direct` and `background` options. Background mode uses a yt-dlp cache download when direct stream extraction is unreliable — creates a text-only card immediately, queues media updates for mined notes, and fills audio/image fields once the download finishes. Progress is announced via overlay/OSD notifications. Downloads are capped at 720p by default (`youtube.mediaCache.maxHeight`). Switching back to direct mode cancels any in-flight background download.
+
+**Fixed**
+- Log Export: Fixed log filenames to use the local date so exports around UTC midnight include the current day's logs rather than stale prior-day files. Expanded export redaction to mask IPs, emails, auth and cookie headers, yt-dlp cookie arguments, URL credentials, token/key/password fields, and signed YouTube media URL parameters.
+- YouTube Card Media: Improved media generation reliability by sending safer ffmpeg options for resolved streams and skipping stale stream maps (including cached YouTube files). Hardened background cache downloads with IPv4 and extractor retry flags; failed downloads now notify the user and clear queued media updates instead of leaving them silently pending. Stale background cache files are cleaned on startup and before each new download.
+
+## v0.17.0 (2026-06-15)
+
+**Changed**
+
+- **Subtitle Delay Keybindings**: Updated default overlay subtitle delay and step bindings to match mpv conventions: `z`, `Z`, and `x` adjust `sub-delay`; `Ctrl+Shift+Left/Right` run native `sub-step` with OSD feedback. The previous SubMiner-only adjacent-cue delay action has been removed.
+- **Update Notifications**: New installs now default update notifications to overlay-only instead of overlay + system notifications.
+
+**Fixed**
+
+- **Anki – Highlight Word**: Fixed bolding of the mined word in Kiku sentence and sentence-furigana fields when the source Yomitan sentence did not already contain bold markup.
+- **Anki – Lapis/Kiku Word Cards**: Fixed word-and-sentence marker missing from Lapis/Kiku word cards enriched through SubMiner, which could hide sentence context on the card front.
+- **Anki – Windows Media Generation**: Fixed known-word cache refreshes when no deck is configured, and fixed audio/image generation after background launches by recreating missing FFmpeg temp directories before clipping.
+- **Character Dictionary – Windows**: Fixed the Windows "SubMiner mpv" shortcut so character dictionary auto-sync can fall back to mpv's current video path when app media state is not yet ready.
+- **Notifications**: Restored the SubMiner app icon on system notifications that do not supply a custom notification image.
+- **Overlay – macOS Yomitan Popup**: Fixed Yomitan popup focus after card mining or popup reload; fixed popups staying open when clicking transparent overlay space — click-away now closes the popup and returns click passthrough to mpv without a hide/reappear cycle.
+- **Overlay – Linux Auto-Pause Startup**: Fixed the visible overlay on Linux auto-paused startup to stay interactive during the initial measurement gap; startup subtitle cache misses now paint raw text before tokenization finishes, and temporarily empty `sub-text` refreshes are resolved before warm readiness resumes playback.
+- **Overlay – Playlist Advance**: Fixed the visible overlay being dismissed when mpv advances to the next playlist item, including when the next episode loads after the warm transition delay.
+- **Overlay – Windows Subtitle Bar**: Fixed shaky hover and click interaction on the Windows subtitle bar when a video attaches to an already-running background SubMiner instance.
+- **Stats – AniList Linking**: Fixed manual AniList linking from the stats anime page so auto-searches strip the generated "Season N" suffix and query only the anime title.
+- **Updates – Linux Support Assets**: Fixed Linux updates to correctly install and refresh the launcher runtime plugin copy and rofi theme alongside AppImage and launcher updates; unrelated SubMiner data directories are now left untouched and plugin copies are staged before replacing the live runtime. Fixed first-launch playback on fresh installs by auto-installing missing managed support assets from the bundled app before mpv starts.
+
+**Docs**
+
+- **Linux Update Flows**: Documented that Linux update flows manage the launcher runtime plugin copy and rofi theme from `subminer-assets.tar.gz`, and that normal launcher playback auto-installs those managed support assets on a fresh install if either is missing.
+
+<details>
+<summary>Internal changes</summary>
+
+**Internal**
+
+- **Runtime Modules**: Split main-process runtime wiring into focused modules without changing user-facing behavior; hardened helpers against stale background stats daemon PIDs, stalled subtitle extraction, and dropped async errors.
+- **Release CI**: Fixed GitHub release notes to preserve the `What's Changed` and `New Contributors` attribution sections when CI regenerates from the committed changelog; scoped prerelease note reuse to the same base version so a new beta line starts from current fragments.
+
+</details>
+
+## Previous Versions
+
+<details>
+<summary>v0.16.x</summary>
+
+<h2>v0.16.0 (2026-06-10)</h2>
 
 **Breaking Changes**
 
@@ -24,7 +79,7 @@
 - **Stats Browsing**: Remembers library card size; retries stored cover art without extra AniList lookups; preserves PNG/WebP MIME types; honors custom AnkiConnect URLs for Browse; shows progress during session deletes.
 - **Startup Notifications**: Tokenization, subtitle annotation, and character dictionary status now route through queued overlay notifications in `overlay`/`both` mode instead of falling back to mpv OSD while the overlay loads.
 - **Notification Deduplication**: Cycling subtitle modes updates the active overlay card in place rather than stacking duplicates; repeated progress updates (e.g. subsync) tick in place without flickering.
-- **Update Notification Default**: New installs default `notificationType` to `overlay`, while `both` remains available for overlay + system notifications.
+- **Update Notification Default**: New installs default `notificationType` to `both` so update alerts appear in both overlay and system notifications.
 
 **Fixed**
 
@@ -48,7 +103,7 @@
 
 </details>
 
-## Previous Versions
+</details>
 
 <details>
 <summary>v0.15.x</summary>

docs-site/configuration.md

@@ -188,6 +188,9 @@ Control the minimum log level for runtime output:
 | `files.launcher` | boolean                                  | Write launcher command logs (default: `true`)                        |
 | `files.mpv`      | boolean                                  | Write mpv player logs. Enable temporarily for mpv/plugin debugging.  |
 
+Log filenames use the local calendar date, for example `app-YYYY-MM-DD.log`, `launcher-YYYY-MM-DD.log`, and `mpv-YYYY-MM-DD.log`.
+Log export creates a sanitized copy of those files; it does not rewrite the original log files on disk.
+
 ### Updates
 
 Configure automatic update checks and update notifications:
@@ -948,6 +951,7 @@ Enable automatic Anki card creation and updates with media generation:
       "animatedMaxWidth": 640,
       "animatedMaxHeight": 0,
       "animatedCrf": 35,
+      "normalizeAudio": true,
       "audioPadding": 0,
       "fallbackDuration": 3,
       "maxMediaDuration": 30
@@ -998,6 +1002,7 @@ This example is intentionally compact. The option table below documents availabl
 | `ankiConnect.ai.model`                            | string                                      | Optional model override for Anki AI translation/enrichment flows.                                                                                                                                                               |
 | `ankiConnect.ai.systemPrompt`                     | string                                      | Optional system prompt override for Anki AI translation/enrichment flows.                                                                                                                                                       |
 | `media.generateAudio`                             | `true`, `false`                             | Generate audio clips from video (default: `true`)                                                                                                                                                                               |
+| `media.normalizeAudio`                            | `true`, `false`                             | Normalize generated sentence-audio loudness during media extraction (default: `true`). Set to `false` to keep raw source loudness.                                                                                              |
 | `media.generateImage`                             | `true`, `false`                             | Generate image/animation screenshots (default: `true`)                                                                                                                                                                          |
 | `media.imageType`                                 | `"static"`, `"avif"`                        | Image type: static screenshot or animated AVIF (default: `"static"`)                                                                                                                                                            |
 | `media.imageFormat`                               | `"jpg"`, `"png"`, `"webp"`                  | Image format (default: `"jpg"`)                                                                                                                                                                                                 |
@@ -1520,14 +1525,24 @@ Set defaults used by managed subtitle auto-selection and the `subminer` launcher
 ```json
 {
   "youtube": {
-    "primarySubLanguages": ["ja", "jpn"]
+    "primarySubLanguages": ["ja", "jpn"],
+    "mediaCache": {
+      "mode": "direct",
+      "maxHeight": 720
+    }
   }
 }
 ```
 
-| Option                | Values   | Description                                                                                      |
-| --------------------- | -------- | ------------------------------------------------------------------------------------------------ |
-| `primarySubLanguages` | string[] | Primary subtitle language priority for managed subtitle auto-selection (default `["ja", "jpn"]`) |
+| Option                 | Values                   | Description                                                                                      |
+| ---------------------- | ------------------------ | ------------------------------------------------------------------------------------------------ |
+| `primarySubLanguages`  | string[]                 | Primary subtitle language priority for managed subtitle auto-selection (default `["ja", "jpn"]`) |
+| `mediaCache.mode`      | `direct` \| `background` | YouTube card audio/image extraction mode (default `direct`)                                      |
+| `mediaCache.maxHeight` | number                   | Maximum background cache download height. Set `0` for unlimited (default `720`)                  |
+
+`mediaCache.mode: "direct"` extracts card media from the active YouTube stream URL. `mediaCache.mode: "background"` starts a separate yt-dlp media download after YouTube playback has loaded, including YouTube URLs opened directly in mpv and resolved stream URLs when mpv still exposes the original YouTube playlist entry. Playback and subtitle loading do not wait for that download. Use background mode if direct card media generation hits YouTube `403` errors from expiring stream URLs.
+
+Background cache downloads are capped by `mediaCache.maxHeight`, which defaults to 720p; set it to `0` to let yt-dlp choose the best available height. Downloads use IPv4 and yt-dlp retry flags to reduce YouTube throttling failures. SubMiner announces when the background cache download starts and when the cache is ready, using the configured notification surface; overlay and OSD messages queue until the overlay or mpv is ready. If you mine cards before the cache is ready, SubMiner creates the text fields immediately, queues the audio/image work for those note IDs, shows a status notification, and fills the media fields once the cached file is ready. If the cache download fails, SubMiner shows a failure notification, shows queued-card failure notifications, and clears the pending updates.
 
 Current launcher behavior:
 

docs-site/launcher-script.md

@@ -61,6 +61,23 @@ Override with the `SUBMINER_ROFI_THEME` environment variable:
 SUBMINER_ROFI_THEME=/path/to/custom-theme.rasi subminer -R
 ```
 
+## Watch History
+
+`subminer -H` (or `--history`) browses your local watch history, sourced from the immersion tracker database. It works with both pickers: fzf by default, rofi with `-R -H`.
+
+```bash
+subminer -H                            # fzf history browser
+subminer -R -H                         # rofi history browser
+```
+
+The first menu lists every locally watched series, most recently watched first, using the parsed media title (e.g. the anime title) when available and the directory name otherwise. Selecting a series opens an action menu:
+
+- **Replay last watched** — replays the most recently watched episode
+- **Next episode** — plays the episode after the last watched one (continues into the next season directory when the season ends)
+- **Browse episodes** — lists the video files in the series directory in episode order, using the same fzf/rofi episode picker as directory browsing; if the series has multiple season directories, a season menu is shown first
+
+Series whose directories are not currently accessible (e.g. an unmounted network share) are hidden from the list. Watch history requires the immersion tracker database (`immersionTracking.dbPath`, default `<config dir>/immersion.sqlite`), which SubMiner populates during playback.
+
 ## Common Commands
 
 ```bash
@@ -83,7 +100,7 @@ subminer stats -b                       # start background stats daemon
 | `subminer stats cleanup`                   | Backfill vocabulary metadata and prune stale rows                  |
 | `subminer doctor`                          | Dependency + config + socket diagnostics                           |
 | `subminer settings`                        | Open the SubMiner settings window                                  |
-| `subminer logs -e`                         | Export a sanitized log ZIP and print its path                      |
+| `subminer logs -e`                         | Export a sanitized local-date log ZIP and print its path           |
 | `subminer config path`                     | Print active config file path                                      |
 | `subminer config show`                     | Print active config contents                                       |
 | `subminer mpv status`                      | Check mpv socket readiness                                         |
@@ -105,6 +122,7 @@ Use `subminer <subcommand> -h` for command-specific help.
 | `-d, --directory`     | Video search directory (default: cwd)                                |
 | `-r, --recursive`     | Search directories recursively                                       |
 | `-R, --rofi`          | Use rofi instead of fzf                                              |
+| `-H, --history`       | Browse local watch history (see [Watch History](#watch-history))     |
 | `--setup`             | Open first-run setup popup manually                                  |
 | `-v, --version`       | Print installed SubMiner version                                     |
 | `-u, --update`        | Check for SubMiner updates and update the app/launcher when possible |

docs-site/public/config.example.jsonc

@@ -559,6 +559,7 @@
       "animatedMaxHeight": 0, // Maximum height for animated AVIF captures, in pixels. Set to 0 to preserve aspect ratio.
       "animatedCrf": 35, // Animated AVIF CRF quality target. Lower values produce larger, higher-quality files.
       "syncAnimatedImageToWordAudio": true, // For animated AVIF images, prepend a frozen first frame matching the existing word-audio duration so motion starts with sentence audio. Values: true | false
+      "normalizeAudio": true, // Normalize generated sentence audio loudness during media extraction. Values: true | false
       "audioPadding": 0, // Seconds of padding appended to both ends of generated sentence audio and animated AVIF clips.
       "fallbackDuration": 3, // Fallback clip duration in seconds when subtitle timing data is unavailable.
       "maxMediaDuration": 30 // Maximum allowed media clip duration in seconds.
@@ -618,7 +619,11 @@
     "primarySubLanguages": [
       "ja",
       "jpn"
-    ] // Comma-separated primary subtitle language priority for managed subtitle auto-selection.
+    ], // Comma-separated primary subtitle language priority for managed subtitle auto-selection.
+    "mediaCache": {
+      "mode": "direct", // How YouTube card audio/images are extracted. Values: direct | background
+      "maxHeight": 720 // Maximum video height downloaded for the YouTube background media cache. Set to 0 for unlimited.
+    } // Media cache setting.
   }, // Defaults for managed subtitle language preferences and YouTube subtitle loading.
 
   // ==========================================

docs-site/usage.md

@@ -148,7 +148,7 @@ SubMiner.AppImage --dictionary-select --dictionary-anilist-id 21355  # Pin corre
 SubMiner.AppImage --help                  # Show all options
 ```
 
-The tray menu includes `Export Logs`, which creates the same sanitized log ZIP as `subminer logs -e` and shows the archive path when complete.
+The tray menu includes `Export Logs`, which creates the same sanitized local-date log ZIP as `subminer logs -e` and shows the archive path when complete. Export sanitization masks common PII and secrets, including home-directory usernames, IP addresses, emails, auth/cookie headers, yt-dlp cookie arguments, URL credentials, token/key/password fields, and signed YouTube media URL query strings. The exported copy is sanitized; source log files remain unredacted on disk.
 
 Once Jellyfin is configured, the tray menu includes `Jellyfin Discovery` for starting or stopping cast discovery in the current app session without changing config.
 
@@ -191,7 +191,7 @@ This flow requires `mpv.exe` to be discoverable. Leave `mpv.executablePath` blan
 - `subminer jellyfin` / `subminer jf`: Jellyfin-focused workflow aliases.
 - `subminer doctor`: health checks for core dependencies and runtime paths.
 - `subminer settings`: open the SubMiner settings window (also `subminer --settings`).
-- `subminer logs -e`: export a sanitized ZIP of today's logs, or the most recent logs when no current-day log exists.
+- `subminer logs -e`: export a sanitized ZIP of today's local-date logs, or the most recent logs when no current-day log exists. The exported copy masks common PII and secrets; on-disk logs are unchanged.
 - `subminer config`: config file helpers (`path`, `show`).
 - `subminer mpv`: mpv helpers (`status`, `socket`, `idle`).
 - `subminer dictionary <path>`: generates a Yomitan-importable character dictionary ZIP from a file/directory target.

docs-site/youtube-integration.md

@@ -74,20 +74,30 @@ Press **Ctrl+Alt+C** during YouTube playback to open the subtitle picker overlay
 - Select different primary and secondary tracks
 - Retry track loading if the auto-load failed or picked the wrong track
 
+SubMiner shows an "Opening YouTube subtitle picker..." status through your configured notification
+surface while it probes tracks and prepares the modal, then updates the subtitle download progress
+card to a success notification after the selected tracks load.
+
 The picker displays each track with its language, kind (manual/auto), and title when available.
 
 ## Subtitle Format Handling
 
 SubMiner handles several YouTube subtitle formats transparently:
 
-| Format | Handling |
-| ------ | -------- |
-| `srt`, `vtt` | Used directly (preferred for manual tracks) |
+| Format                 | Handling                                                 |
+| ---------------------- | -------------------------------------------------------- |
+| `srt`, `vtt`           | Used directly (preferred for manual tracks)              |
 | `srv1`, `srv2`, `srv3` | YouTube TimedText XML --- converted to VTT automatically |
-| Auto-generated VTT | Normalized to remove rolling-caption text duplication |
+| Auto-generated VTT     | Normalized to remove rolling-caption text duplication    |
 
 For auto-generated tracks, SubMiner prefers `srv3` > `srv2` > `srv1` > `vtt` (TimedText XML produces cleaner output). For manual tracks, `srt` > `vtt` is preferred.
 
+## Card Media Cache
+
+By default, YouTube card audio and screenshots are extracted directly from mpv's active stream URLs. If generated card media fails with YouTube `403` errors, set `youtube.mediaCache.mode` to `"background"`. Background mode starts a separate `yt-dlp` media download after playback loads, including YouTube URLs opened directly in mpv and resolved stream URLs when mpv still exposes the original YouTube playlist entry. It creates text fields immediately, queues audio/image work for mined notes, and fills those fields once the local cache file is ready.
+
+Background cache downloads use IPv4 and retry flags to reduce YouTube throttling failures. If the background download still fails, SubMiner shows a cache failure notification, shows queued-card failure notifications, and clears those pending updates so cards are not left waiting silently.
+
 ## Configuration Reference
 
 ### Primary Subtitle Languages
@@ -95,13 +105,13 @@ For auto-generated tracks, SubMiner prefers `srv3` > `srv2` > `srv1` > `vtt` (Ti
 ```jsonc
 {
   "youtube": {
-    "primarySubLanguages": ["ja", "jpn"]
-  }
+    "primarySubLanguages": ["ja", "jpn"],
+  },
 }
 ```
 
-| Option | Type | Description |
-| ------ | ---- | ----------- |
+| Option                | Type       | Description                                                                           |
+| --------------------- | ---------- | ------------------------------------------------------------------------------------- |
 | `primarySubLanguages` | `string[]` | Language priority for YouTube primary subtitle auto-loading (default `["ja", "jpn"]`) |
 
 ### Secondary Subtitle Languages
@@ -113,16 +123,16 @@ Secondary track selection uses the shared `secondarySub` config:
   "secondarySub": {
     "secondarySubLanguages": [],
     "autoLoadSecondarySub": false,
-    "defaultMode": "hover"
-  }
+    "defaultMode": "hover",
+  },
 }
 ```
 
-| Option | Type | Description |
-| ------ | ---- | ----------- |
-| `secondarySubLanguages` | `string[]` | Extra language codes (e.g. `["eng", "en"]`) used when auto-selecting a secondary track. Default is empty (`[]`). For YouTube, SubMiner always tries an English track first regardless of this list. |
-| `autoLoadSecondarySub` | `boolean` | Auto-detect and load a matching secondary track (default: `false`) |
-| `defaultMode` | `"hidden"` / `"visible"` / `"hover"` | Initial display mode for secondary subtitles (default: `"hover"`) |
+| Option                  | Type                                 | Description                                                                                                                                                                                         |
+| ----------------------- | ------------------------------------ | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| `secondarySubLanguages` | `string[]`                           | Extra language codes (e.g. `["eng", "en"]`) used when auto-selecting a secondary track. Default is empty (`[]`). For YouTube, SubMiner always tries an English track first regardless of this list. |
+| `autoLoadSecondarySub`  | `boolean`                            | Auto-detect and load a matching secondary track (default: `false`)                                                                                                                                  |
+| `defaultMode`           | `"hidden"` / `"visible"` / `"hover"` | Initial display mode for secondary subtitles (default: `"hover"`)                                                                                                                                   |
 
 Precedence: CLI flag > environment variable > `config.jsonc` > built-in default.
 
@@ -131,6 +141,7 @@ Precedence: CLI flag > environment variable > `config.jsonc` > built-in default.
 - **No subtitles found**: The video may not have Japanese subtitles. Open the picker with `Ctrl+Alt+C` to see all available tracks.
 - **yt-dlp not found**: Install `yt-dlp` and ensure it is on `PATH`, or set `SUBMINER_YTDLP_BIN` to the binary path.
 - **Probe timeout**: `yt-dlp` has a 15-second timeout per operation. Slow connections or rate-limited IPs may hit this. Retry or update `yt-dlp`.
+- **Card media `403` errors**: Switch `youtube.mediaCache.mode` from `"direct"` to `"background"` so card media is generated from a local `yt-dlp` cache instead of ffmpeg reading an expiring YouTube stream URL.
 - **Auto-caption quality**: YouTube auto-generated captions vary in quality. Manual subtitles (when available) are always preferred.
 - **`ytsearch:` targets**: `subminer ytsearch:"keyword"` plays the first search result. Subtitle availability depends on the matched video.
 - **Secondary subtitle fails**: Secondary track failures never block playback. The primary subtitle loads independently.

docs/RELEASING.md

@@ -61,8 +61,10 @@
    committed file — so review it before committing. If you add more
    `changes/*.md` fragments for a later beta/RC, rerun
    `bun run changelog:prerelease-notes --version <version>`; the generator uses
-   the existing prerelease notes as the baseline and asks Claude to merge only
-   the new fragment material. Do not run `bun run changelog:build`.
+   the existing prerelease notes as the baseline only when their hidden
+   `prerelease-base-version` marker matches the current base version, and asks
+   Claude to merge only the new fragment material. Do not run
+   `bun run changelog:build`.
 6. Tag the commit: `git tag v<version>`.
 7. Push commit + tag.
 

launcher/bun-sqlite.d.ts

@@ -0,0 +1,30 @@
+// Minimal ambient typing for bun:sqlite. The launcher always runs under bun
+// (see the build banner in package.json), but the repo typechecks with plain
+// tsc which has no bun type definitions.
+declare module 'bun:sqlite' {
+  export interface RunResult {
+    changes: number;
+    lastInsertRowid: number | bigint;
+  }
+
+  export interface Statement<ReturnType = unknown, ParamsType extends unknown[] = unknown[]> {
+    all(...params: ParamsType): ReturnType[];
+    get(...params: ParamsType): ReturnType | undefined;
+    run(...params: ParamsType): RunResult;
+  }
+
+  export class Database {
+    constructor(
+      filename: string,
+      options?: { readonly?: boolean; readwrite?: boolean; create?: boolean },
+    );
+    query<ReturnType = unknown, ParamsType extends unknown[] = unknown[]>(
+      sql: string,
+    ): Statement<ReturnType, ParamsType>;
+    prepare<ReturnType = unknown, ParamsType extends unknown[] = unknown[]>(
+      sql: string,
+    ): Statement<ReturnType, ParamsType>;
+    run(sql: string, ...params: unknown[]): RunResult;
+    close(throwOnError?: boolean): void;
+  }
+}

launcher/commands/history-command.ts

@@ -0,0 +1,228 @@
+import fs from 'node:fs';
+import path from 'node:path';
+import { spawnSync } from 'node:child_process';
+import { fail, log } from '../log.js';
+import { commandExists } from '../util.js';
+import {
+  collectVideos,
+  findRofiTheme,
+  formatPickerLaunchError,
+  showFzfMenu,
+  showRofiMenu,
+} from '../picker.js';
+import {
+  findNextEpisode,
+  groupHistoryBySeries,
+  listSeasonDirs,
+  materializeCoverArt,
+  queryLocalWatchHistory,
+  resolveImmersionDbPath,
+  sortVideosByEpisode,
+  type HistorySeriesEntry,
+} from '../history.js';
+import type { Args } from '../types.js';
+import type { LauncherCommandContext } from './context.js';
+
+function checkPickerDependencies(args: Args): void {
+  if (args.useRofi) {
+    if (!commandExists('rofi')) fail('Missing dependency: rofi');
+    return;
+  }
+  if (!commandExists('fzf')) fail('Missing dependency: fzf');
+}
+
+function showRofiIndexMenu(
+  labels: string[],
+  prompt: string,
+  themePath: string | null,
+  icons: Array<string | null> = [],
+): number {
+  const rofiArgs = ['-dmenu', '-i', '-matching', 'fuzzy', '-format', 'i', '-p', prompt];
+  const hasIcons = icons.some(Boolean);
+  if (hasIcons) rofiArgs.push('-show-icons');
+  if (themePath) {
+    rofiArgs.push('-theme', themePath);
+  } else {
+    rofiArgs.push('-theme-str', 'configuration { font: "Noto Sans CJK JP Regular 8";}');
+  }
+  if (hasIcons) {
+    rofiArgs.push('-theme-str', 'configuration { show-icons: true; }');
+    rofiArgs.push('-theme-str', 'element-icon { enabled: true; size: 3em; }');
+  }
+  const lines = labels.map((label, index) =>
+    icons[index] ? `${label}\u0000icon\u001f${icons[index]}` : label,
+  );
+  const result = spawnSync('rofi', rofiArgs, {
+    input: `${lines.join('\n')}\n`,
+    encoding: 'utf8',
+    stdio: ['pipe', 'pipe', 'ignore'],
+  });
+  if (result.error) {
+    fail(formatPickerLaunchError('rofi', result.error as NodeJS.ErrnoException));
+  }
+  const out = (result.stdout || '').trim();
+  if (!out) return -1;
+  const idx = Number.parseInt(out, 10);
+  return Number.isInteger(idx) && idx >= 0 && idx < labels.length ? idx : -1;
+}
+
+function showFzfIndexMenu(labels: string[], prompt: string): number {
+  const lines = labels.map((label, index) => `${index}\t${label}`);
+  const result = spawnSync(
+    'fzf',
+    [
+      '--ansi',
+      '--reverse',
+      '--ignore-case',
+      `--prompt=${prompt}: `,
+      '--delimiter=\t',
+      '--with-nth=2..',
+    ],
+    {
+      input: `${lines.join('\n')}\n`,
+      encoding: 'utf8',
+      stdio: ['pipe', 'pipe', 'inherit'],
+    },
+  );
+  if (result.error) {
+    fail(formatPickerLaunchError('fzf', result.error as NodeJS.ErrnoException));
+  }
+  const picked = (result.stdout || '').trim();
+  const tab = picked.indexOf('\t');
+  if (tab === -1) return -1;
+  const idx = Number.parseInt(picked.slice(0, tab), 10);
+  return Number.isInteger(idx) && idx >= 0 && idx < labels.length ? idx : -1;
+}
+
+function pickIndex(
+  labels: string[],
+  prompt: string,
+  useRofi: boolean,
+  themePath: string | null,
+  icons: Array<string | null> = [],
+): number {
+  if (labels.length === 0) return -1;
+  return useRofi
+    ? showRofiIndexMenu(labels, prompt, themePath, icons)
+    : showFzfIndexMenu(labels, prompt);
+}
+
+function formatEpisodeLabel(entry: HistorySeriesEntry): string {
+  const { parsedSeason, parsedEpisode } = entry.lastWatched;
+  if (parsedEpisode === null) return '';
+  return parsedSeason !== null ? `S${parsedSeason}E${parsedEpisode}` : `E${parsedEpisode}`;
+}
+
+function formatSeriesLabel(entry: HistorySeriesEntry): string {
+  const episodeLabel = formatEpisodeLabel(entry);
+  return episodeLabel ? `${entry.displayName}  [last: ${episodeLabel}]` : entry.displayName;
+}
+
+function pickEpisodeFromDir(dir: string, context: LauncherCommandContext): string | null {
+  const { args, scriptPath } = context;
+  const videos = sortVideosByEpisode(collectVideos(dir, false));
+  if (videos.length === 0) {
+    fail(`No video files found in: ${dir}`);
+  }
+  const selected = args.useRofi
+    ? showRofiMenu(videos, dir, false, scriptPath, args.logLevel)
+    : showFzfMenu(videos);
+  return selected || null;
+}
+
+function browseEpisodes(
+  entry: HistorySeriesEntry,
+  context: LauncherCommandContext,
+  themePath: string | null,
+): string | null {
+  const { args } = context;
+  const seasons = listSeasonDirs(entry.seriesRoot);
+  let dir = entry.seriesRoot;
+
+  if (seasons.length > 1) {
+    const idx = pickIndex(
+      seasons.map((season) => season.name),
+      `${entry.displayName} — Season`,
+      args.useRofi,
+      themePath,
+    );
+    if (idx < 0) return null;
+    dir = seasons[idx]!.path;
+  } else if (seasons.length === 1 && collectVideos(dir, false).length === 0) {
+    dir = seasons[0]!.path;
+  }
+
+  return pickEpisodeFromDir(dir, context);
+}
+
+export async function runHistoryCommand(context: LauncherCommandContext): Promise<string | null> {
+  const { args, scriptPath } = context;
+
+  checkPickerDependencies(args);
+  const themePath = args.useRofi ? findRofiTheme(scriptPath) : null;
+
+  const dbPath = resolveImmersionDbPath();
+  if (!fs.existsSync(dbPath)) {
+    fail(`Watch history database not found: ${dbPath}`);
+  }
+
+  const rows = queryLocalWatchHistory(dbPath);
+  const series = groupHistoryBySeries(rows);
+  if (series.length === 0) {
+    fail('No local watch history found (or watched directories are not accessible).');
+  }
+
+  log('info', args.logLevel, `Watch history: ${series.length} series found in ${dbPath}`);
+
+  const coverPaths = args.useRofi
+    ? materializeCoverArt(
+        dbPath,
+        series.map((seriesEntry) => seriesEntry.coverBlobHash),
+      )
+    : new Map<string, string>();
+  const seriesIcons = series.map((seriesEntry) =>
+    seriesEntry.coverBlobHash ? (coverPaths.get(seriesEntry.coverBlobHash) ?? null) : null,
+  );
+
+  const seriesIdx = pickIndex(
+    series.map(formatSeriesLabel),
+    'Watch History',
+    args.useRofi,
+    themePath,
+    seriesIcons,
+  );
+  if (seriesIdx < 0) return null;
+  const entry = series[seriesIdx]!;
+
+  const lastPath = path.resolve(entry.lastWatched.sourcePath);
+  const lastExists = fs.existsSync(lastPath);
+  const nextEpisode = findNextEpisode(lastPath);
+
+  const actions: Array<{ kind: 'replay' | 'next' | 'browse'; label: string }> = [];
+  if (lastExists) {
+    actions.push({ kind: 'replay', label: `Replay last watched — ${path.basename(lastPath)}` });
+  }
+  if (nextEpisode) {
+    actions.push({ kind: 'next', label: `Next episode — ${path.basename(nextEpisode)}` });
+  }
+  actions.push({ kind: 'browse', label: 'Browse episodes' });
+
+  const entryIcon = seriesIcons[seriesIdx] ?? null;
+  const actionIdx = pickIndex(
+    actions.map((action) => action.label),
+    entry.displayName,
+    args.useRofi,
+    themePath,
+    actions.map(() => entryIcon),
+  );
+  if (actionIdx < 0) return null;
+
+  switch (actions[actionIdx]!.kind) {
+    case 'replay':
+      return lastPath;
+    case 'next':
+      return nextEpisode;
+    case 'browse':
+      return browseEpisodes(entry, context, themePath);
+  }
+}

launcher/commands/playback-command.test.ts

@@ -35,6 +35,7 @@ function createContext(): LauncherCommandContext {
       texthookerOnly: false,
       texthookerOpenBrowser: false,
       useRofi: false,
+      history: false,
       logLevel: 'info',
       logRotation: 7,
       passwordStore: '',

launcher/config/args-normalizer.ts

@@ -198,6 +198,7 @@ export function createDefaultArgs(
     texthookerOnly: false,
     texthookerOpenBrowser: false,
     useRofi: false,
+    history: false,
     logLevel: loggingConfig.level ?? 'warn',
     logRotation: loggingConfig.rotation ?? 7,
     passwordStore: '',
@@ -231,6 +232,7 @@ export function applyRootOptionsToArgs(
   if (typeof options.logLevel === 'string') parsed.logLevel = parseLogLevel(options.logLevel);
   if (typeof options.passwordStore === 'string') parsed.passwordStore = options.passwordStore;
   if (options.rofi === true) parsed.useRofi = true;
+  if (options.history === true) parsed.history = true;
   if (options.update === true) parsed.update = true;
   if (options.version === true) parsed.version = true;
   if (options.settings === true) parsed.settings = true;

launcher/config/cli-parser-builder.ts

@@ -64,6 +64,7 @@ function applyRootOptions(program: Command): void {
     .option('--settings', 'Open settings window')
     .option('-u, --update', 'Check for updates')
     .option('-R, --rofi', 'Use rofi picker')
+    .option('-H, --history', 'Browse local watch history')
     .option('-S, --start-overlay', 'Auto-start overlay')
     .option('-T, --no-texthooker', 'Disable texthooker-ui server');
 }

launcher/history-art.ts

@@ -0,0 +1,126 @@
+import fs from 'node:fs';
+import os from 'node:os';
+import path from 'node:path';
+import { Database } from 'bun:sqlite';
+import { withReadonlyWalRetry } from './history-db.js';
+
+const COVER_EXTENSIONS = ['.jpg', '.png', '.webp', '.gif'] as const;
+const SAFE_COVER_HASH_PATTERN = /^[a-z0-9_-]+$/i;
+
+export function getDefaultCoverCacheDir(): string {
+  return path.join(os.homedir(), '.cache', 'subminer', 'covers');
+}
+
+export function detectImageExtension(blob: Buffer): string {
+  if (blob.length >= 8 && blob.subarray(0, 8).equals(Buffer.from('89504e470d0a1a0a', 'hex'))) {
+    return '.png';
+  }
+  if (blob.length >= 3 && blob[0] === 0xff && blob[1] === 0xd8 && blob[2] === 0xff) {
+    return '.jpg';
+  }
+  if (
+    blob.length >= 12 &&
+    blob.subarray(0, 4).toString('ascii') === 'RIFF' &&
+    blob.subarray(8, 12).toString('ascii') === 'WEBP'
+  ) {
+    return '.webp';
+  }
+  if (blob.length >= 4 && blob.subarray(0, 3).toString('ascii') === 'GIF') {
+    return '.gif';
+  }
+  return '.jpg';
+}
+
+function findCachedCover(cacheDir: string, hash: string): string | null {
+  for (const ext of COVER_EXTENSIONS) {
+    const candidate = path.join(cacheDir, `${hash}${ext}`);
+    try {
+      if (fs.statSync(candidate).size > 0) return candidate;
+    } catch {
+      // not cached with this extension
+    }
+  }
+  return null;
+}
+
+function queryCoverBlobs(
+  dbPath: string,
+  hashes: string[],
+  options: { readonly?: boolean; readwrite?: boolean; create?: boolean },
+): Map<string, Buffer> {
+  const blobs = new Map<string, Buffer>();
+  const db = new Database(dbPath, options);
+  try {
+    const hasBlobTable = db
+      .query(`SELECT 1 FROM sqlite_master WHERE type = 'table' AND name = 'imm_cover_art_blobs'`)
+      .get();
+    if (!hasBlobTable) return blobs;
+
+    const stmt = db.query<{ cover_blob: Uint8Array | null }>(
+      'SELECT cover_blob FROM imm_cover_art_blobs WHERE blob_hash = ?',
+    );
+    for (const hash of hashes) {
+      const row = stmt.get(hash);
+      if (row?.cover_blob && row.cover_blob.length > 0) {
+        blobs.set(hash, Buffer.from(row.cover_blob));
+      }
+    }
+    return blobs;
+  } finally {
+    db.close();
+  }
+}
+
+function isSafeCoverHash(hash: string | null | undefined): hash is string {
+  return typeof hash === 'string' && SAFE_COVER_HASH_PATTERN.test(hash);
+}
+
+/**
+ * Ensures cover art blobs referenced by hash exist as image files in the cache
+ * directory, extracting missing ones from the stats database. Returns a map of
+ * blob hash to on-disk image path for every cover that could be materialized.
+ */
+export function materializeCoverArt(
+  dbPath: string,
+  hashes: Array<string | null | undefined>,
+  cacheDir: string = getDefaultCoverCacheDir(),
+): Map<string, string> {
+  const wanted = Array.from(new Set(hashes.filter(isSafeCoverHash)));
+  const resolved = new Map<string, string>();
+  if (wanted.length === 0) return resolved;
+
+  const missing: string[] = [];
+  for (const hash of wanted) {
+    const cached = findCachedCover(cacheDir, hash);
+    if (cached) {
+      resolved.set(hash, cached);
+    } else {
+      missing.push(hash);
+    }
+  }
+  if (missing.length === 0) return resolved;
+
+  let blobs: Map<string, Buffer>;
+  try {
+    blobs = withReadonlyWalRetry(dbPath, (options) => queryCoverBlobs(dbPath, missing, options));
+  } catch {
+    return resolved;
+  }
+  if (blobs.size === 0) return resolved;
+
+  try {
+    fs.mkdirSync(cacheDir, { recursive: true });
+  } catch {
+    return resolved;
+  }
+  for (const [hash, blob] of blobs) {
+    const target = path.join(cacheDir, `${hash}${detectImageExtension(blob)}`);
+    try {
+      fs.writeFileSync(target, blob);
+      resolved.set(hash, target);
+    } catch {
+      // cache write failure just means no icon for this entry
+    }
+  }
+  return resolved;
+}

launcher/history-db.ts

@@ -0,0 +1,151 @@
+import fs from 'node:fs';
+import os from 'node:os';
+import path from 'node:path';
+import { Database } from 'bun:sqlite';
+import { resolveConfigDir } from '../src/config/path-resolution.js';
+import { readLauncherMainConfigObject } from './config/shared-config-reader.js';
+import type { HistoryVideoRow } from './history-types.js';
+import { resolvePathMaybe } from './util.js';
+
+export function resolveImmersionDbPath(): string {
+  const root = readLauncherMainConfigObject();
+  const tracking =
+    root?.immersionTracking &&
+    typeof root.immersionTracking === 'object' &&
+    !Array.isArray(root.immersionTracking)
+      ? (root.immersionTracking as Record<string, unknown>)
+      : null;
+  const configured = typeof tracking?.dbPath === 'string' ? tracking.dbPath.trim() : '';
+  if (configured) return resolvePathMaybe(configured);
+
+  const configDir = resolveConfigDir({
+    platform: process.platform,
+    appDataDir: process.env.APPDATA,
+    xdgConfigHome: process.env.XDG_CONFIG_HOME,
+    homeDir: os.homedir(),
+    existsSync: fs.existsSync,
+  });
+  return path.join(configDir, 'immersion.sqlite');
+}
+
+interface RawHistoryRow {
+  video_id: number;
+  source_path: string | null;
+  parsed_title: string | null;
+  parsed_season: number | null;
+  parsed_episode: number | null;
+  anime_title: string | null;
+  last_watched_ms: number | bigint | null;
+  cover_blob_hash: string | null;
+}
+
+export function queryLocalWatchHistory(dbPath: string): HistoryVideoRow[] {
+  return withReadonlyWalRetry(dbPath, (options) => readHistoryRows(dbPath, options));
+}
+
+export function withReadonlyWalRetry<T>(
+  dbPath: string,
+  query: (options: { readonly?: boolean; readwrite?: boolean; create?: boolean }) => T,
+): T {
+  try {
+    return query({ readonly: true });
+  } catch (error) {
+    if (!isReadonlyWalRetryError(error, dbPath)) throw error;
+    return query({ readwrite: true, create: false });
+  }
+}
+
+export function isReadonlyWalRetryError(error: unknown, dbPath: string): boolean {
+  if (!isWalModeSqliteDatabase(dbPath)) return false;
+  const code =
+    typeof error === 'object' && error !== null && 'code' in error
+      ? String((error as { code?: unknown }).code ?? '')
+      : '';
+  const message = error instanceof Error ? error.message : String(error);
+  const text = `${code} ${message}`.toLowerCase();
+  return (
+    text.includes('readonly') ||
+    text.includes('read-only') ||
+    text.includes('attempt to write a readonly database') ||
+    text.includes('sqlite_cantopen') ||
+    text.includes('unable to open database file')
+  );
+}
+
+function isWalModeSqliteDatabase(dbPath: string): boolean {
+  const header = Buffer.alloc(20);
+  let fd: number | null = null;
+  try {
+    fd = fs.openSync(dbPath, 'r');
+    if (fs.readSync(fd, header, 0, header.length, 0) < header.length) return false;
+  } catch {
+    return false;
+  } finally {
+    if (fd !== null) fs.closeSync(fd);
+  }
+  return header.subarray(0, 16).toString('ascii') === 'SQLite format 3\0' && header[18] === 2;
+}
+
+function tableExists(db: Database, tableName: string): boolean {
+  return Boolean(
+    db.query(`SELECT 1 FROM sqlite_master WHERE type = 'table' AND name = ?`).get(tableName),
+  );
+}
+
+function readHistoryRows(
+  dbPath: string,
+  options: { readonly?: boolean; readwrite?: boolean; create?: boolean },
+): HistoryVideoRow[] {
+  const db = new Database(dbPath, options);
+  try {
+    const hasMediaArt = tableExists(db, 'imm_media_art');
+    const coverSelect = hasMediaArt
+      ? `COALESCE(
+          ma.cover_blob_hash,
+          (SELECT ma2.cover_blob_hash
+           FROM imm_media_art ma2
+           JOIN imm_videos v2 ON v2.video_id = ma2.video_id
+           WHERE v2.anime_id = v.anime_id AND ma2.cover_blob_hash IS NOT NULL
+           LIMIT 1)
+        ) AS cover_blob_hash`
+      : 'NULL AS cover_blob_hash';
+    const coverJoin = hasMediaArt ? 'LEFT JOIN imm_media_art ma ON ma.video_id = v.video_id' : '';
+    const rows = db
+      .query<RawHistoryRow>(
+        `
+        SELECT
+          v.video_id,
+          v.source_path,
+          v.parsed_title,
+          v.parsed_season,
+          v.parsed_episode,
+          COALESCE(a.title_romaji, a.canonical_title) AS anime_title,
+          MAX(CAST(s.started_at_ms AS INTEGER)) AS last_watched_ms,
+          ${coverSelect}
+        FROM imm_sessions s
+        JOIN imm_videos v ON v.video_id = s.video_id
+        LEFT JOIN imm_anime a ON a.anime_id = v.anime_id
+        ${coverJoin}
+        WHERE v.source_type = 1 AND v.source_path IS NOT NULL AND v.source_path != ''
+        GROUP BY v.video_id
+        ORDER BY last_watched_ms DESC
+        `,
+      )
+      .all();
+
+    return rows
+      .filter((row) => typeof row.source_path === 'string' && row.source_path.length > 0)
+      .map((row) => ({
+        videoId: row.video_id,
+        sourcePath: row.source_path!,
+        parsedTitle: row.parsed_title,
+        parsedSeason: row.parsed_season,
+        parsedEpisode: row.parsed_episode,
+        animeTitle: row.anime_title,
+        lastWatchedMs: Number(row.last_watched_ms ?? 0),
+        coverBlobHash: row.cover_blob_hash,
+      }));
+  } finally {
+    db.close();
+  }
+}

launcher/history-navigation.ts

@@ -0,0 +1,132 @@
+import fs from 'node:fs';
+import path from 'node:path';
+import { parseMediaInfo } from '../src/jimaku/utils.js';
+import { collectVideos } from './picker.js';
+import type { HistorySeriesEntry, HistoryVideoRow, SeasonDirEntry } from './history-types.js';
+
+const SEASON_DIR_PATTERN = /^(?:season|s)[\s._-]*(\d{1,3})\b/i;
+
+export function seasonNumberFromDirName(name: string): number | null {
+  const match = name.trim().match(SEASON_DIR_PATTERN);
+  if (!match) return null;
+  const parsed = Number.parseInt(match[1]!, 10);
+  return Number.isFinite(parsed) ? parsed : null;
+}
+
+export function resolveSeriesRoot(filePath: string): string {
+  const parent = path.dirname(filePath);
+  if (seasonNumberFromDirName(path.basename(parent)) !== null) {
+    return path.dirname(parent);
+  }
+  return parent;
+}
+
+export function groupHistoryBySeries(
+  rows: HistoryVideoRow[],
+  existsFn: (candidate: string) => boolean = fs.existsSync,
+): HistorySeriesEntry[] {
+  const byRoot = new Map<string, HistorySeriesEntry>();
+  const sorted = [...rows].sort((a, b) => b.lastWatchedMs - a.lastWatchedMs);
+
+  for (const row of sorted) {
+    const seriesRoot = resolveSeriesRoot(row.sourcePath);
+    const existing = byRoot.get(seriesRoot);
+    if (existing) {
+      if (existing.coverBlobHash === null && row.coverBlobHash !== null) {
+        existing.coverBlobHash = row.coverBlobHash;
+      }
+      continue;
+    }
+    if (!existsFn(seriesRoot)) continue;
+    const displayName =
+      row.parsedTitle?.trim() || row.animeTitle?.trim() || path.basename(seriesRoot);
+    byRoot.set(seriesRoot, {
+      seriesRoot,
+      displayName,
+      lastWatched: row,
+      coverBlobHash: row.coverBlobHash,
+    });
+  }
+
+  return Array.from(byRoot.values());
+}
+
+function compareNatural(a: string, b: string): number {
+  return a.localeCompare(b, undefined, { numeric: true, sensitivity: 'base' });
+}
+
+export function sortVideosByEpisode(videos: string[]): string[] {
+  const parsed = videos.map((video) => ({ video, info: parseMediaInfo(video) }));
+  parsed.sort((a, b) => {
+    if (a.info.episode !== null && b.info.episode !== null) {
+      const seasonA = a.info.season ?? 0;
+      const seasonB = b.info.season ?? 0;
+      if (seasonA !== seasonB) return seasonA - seasonB;
+      if (a.info.episode !== b.info.episode) return a.info.episode - b.info.episode;
+    }
+    return compareNatural(a.video, b.video);
+  });
+  return parsed.map((entry) => entry.video);
+}
+
+function dirContainsVideo(dir: string): boolean {
+  return collectVideos(dir, true).length > 0;
+}
+
+export function listSeasonDirs(seriesRoot: string): SeasonDirEntry[] {
+  let entries: fs.Dirent[];
+  try {
+    entries = fs.readdirSync(seriesRoot, { withFileTypes: true });
+  } catch {
+    return [];
+  }
+
+  const dirs = entries
+    .filter((entry) => entry.isDirectory())
+    .map((entry) => ({
+      name: entry.name,
+      path: path.join(seriesRoot, entry.name),
+      season: seasonNumberFromDirName(entry.name),
+    }))
+    .filter((entry) => dirContainsVideo(entry.path));
+
+  dirs.sort((a, b) => {
+    if (a.season !== null && b.season !== null && a.season !== b.season) {
+      return a.season - b.season;
+    }
+    return compareNatural(a.name, b.name);
+  });
+  return dirs;
+}
+
+function findFirstEpisodeInNextSeason(resolvedLast: string, dir: string): string | null {
+  const seriesRoot = resolveSeriesRoot(resolvedLast);
+  if (seriesRoot === dir) return null;
+  const seasons = listSeasonDirs(seriesRoot);
+  const currentIdx = seasons.findIndex((season) => path.resolve(season.path) === dir);
+  if (currentIdx < 0 || currentIdx + 1 >= seasons.length) return null;
+  const nextSeason = sortVideosByEpisode(collectVideos(seasons[currentIdx + 1]!.path, false));
+  return nextSeason[0] ?? null;
+}
+
+export function findNextEpisode(lastPath: string): string | null {
+  const resolvedLast = path.resolve(lastPath);
+  const dir = path.dirname(resolvedLast);
+  const episodes = sortVideosByEpisode(collectVideos(dir, false));
+  const idx = episodes.indexOf(resolvedLast);
+
+  if (idx >= 0) {
+    if (idx + 1 < episodes.length) return episodes[idx + 1]!;
+  } else {
+    const lastInfo = parseMediaInfo(resolvedLast);
+    if (lastInfo.episode !== null) {
+      const candidate = episodes.find((episode) => {
+        const info = parseMediaInfo(episode);
+        return info.episode !== null && info.episode > lastInfo.episode!;
+      });
+      if (candidate) return candidate;
+    }
+  }
+
+  return findFirstEpisodeInNextSeason(resolvedLast, dir);
+}

launcher/history-types.ts

@@ -0,0 +1,23 @@
+export interface HistoryVideoRow {
+  videoId: number;
+  sourcePath: string;
+  parsedTitle: string | null;
+  parsedSeason: number | null;
+  parsedEpisode: number | null;
+  animeTitle: string | null;
+  lastWatchedMs: number;
+  coverBlobHash: string | null;
+}
+
+export interface HistorySeriesEntry {
+  seriesRoot: string;
+  displayName: string;
+  lastWatched: HistoryVideoRow;
+  coverBlobHash: string | null;
+}
+
+export interface SeasonDirEntry {
+  name: string;
+  path: string;
+  season: number | null;
+}

launcher/history.test.ts

@@ -0,0 +1,441 @@
+import test from 'node:test';
+import assert from 'node:assert/strict';
+import fs from 'node:fs';
+import os from 'node:os';
+import path from 'node:path';
+import { Database } from 'bun:sqlite';
+import {
+  detectImageExtension,
+  findNextEpisode,
+  groupHistoryBySeries,
+  isReadonlyWalRetryError,
+  listSeasonDirs,
+  materializeCoverArt,
+  queryLocalWatchHistory,
+  resolveSeriesRoot,
+  seasonNumberFromDirName,
+  sortVideosByEpisode,
+  type HistoryVideoRow,
+} from './history.js';
+
+function makeRow(overrides: Partial<HistoryVideoRow> = {}): HistoryVideoRow {
+  return {
+    videoId: 1,
+    sourcePath: '/media/anime/Show/Season-1/Show - S01E01.mkv',
+    parsedTitle: 'Show',
+    parsedSeason: 1,
+    parsedEpisode: 1,
+    animeTitle: null,
+    lastWatchedMs: 1000,
+    coverBlobHash: null,
+    ...overrides,
+  };
+}
+
+test('seasonNumberFromDirName detects common season directory names', () => {
+  assert.equal(seasonNumberFromDirName('Season-1'), 1);
+  assert.equal(seasonNumberFromDirName('Season 2'), 2);
+  assert.equal(seasonNumberFromDirName('S03'), 3);
+  assert.equal(seasonNumberFromDirName('season_04'), 4);
+  assert.equal(seasonNumberFromDirName('Specials'), null);
+  assert.equal(seasonNumberFromDirName('Show Name'), null);
+});
+
+test('resolveSeriesRoot skips season directories', () => {
+  assert.equal(
+    resolveSeriesRoot('/media/anime/Show/Season-1/Show - S01E01.mkv'),
+    '/media/anime/Show',
+  );
+  assert.equal(resolveSeriesRoot('/media/anime/Show/Show - 01.mkv'), '/media/anime/Show');
+});
+
+test('groupHistoryBySeries keeps most recent entry per series root', () => {
+  const rows = [
+    makeRow({ videoId: 1, parsedEpisode: 1, lastWatchedMs: 1000 }),
+    makeRow({
+      videoId: 2,
+      sourcePath: '/media/anime/Show/Season-1/Show - S01E02.mkv',
+      parsedEpisode: 2,
+      lastWatchedMs: 3000,
+    }),
+    makeRow({
+      videoId: 3,
+      sourcePath: '/media/anime/Other/Other - 05.mkv',
+      parsedTitle: 'Other',
+      parsedSeason: null,
+      parsedEpisode: 5,
+      lastWatchedMs: 2000,
+    }),
+  ];
+
+  const series = groupHistoryBySeries(rows, () => true);
+
+  assert.equal(series.length, 2);
+  assert.equal(series[0]?.displayName, 'Show');
+  assert.equal(series[0]?.seriesRoot, '/media/anime/Show');
+  assert.equal(series[0]?.lastWatched.parsedEpisode, 2);
+  assert.equal(series[1]?.displayName, 'Other');
+});
+
+test('groupHistoryBySeries filters series roots that no longer exist', () => {
+  const rows = [
+    makeRow({ videoId: 1 }),
+    makeRow({
+      videoId: 2,
+      sourcePath: '/gone/anime/Missing/Season-1/Missing - S01E01.mkv',
+      parsedTitle: 'Missing',
+      lastWatchedMs: 5000,
+    }),
+  ];
+
+  const series = groupHistoryBySeries(rows, (candidate) => !candidate.startsWith('/gone/'));
+
+  assert.equal(series.length, 1);
+  assert.equal(series[0]?.displayName, 'Show');
+});
+
+test('groupHistoryBySeries falls back to directory name for display', () => {
+  const rows = [
+    makeRow({
+      sourcePath: '/media/anime/Some Show Dir/video.mkv',
+      parsedTitle: null,
+      animeTitle: null,
+    }),
+  ];
+
+  const series = groupHistoryBySeries(rows, () => true);
+
+  assert.equal(series[0]?.displayName, 'Some Show Dir');
+});
+
+test('sortVideosByEpisode orders by parsed episode with natural fallback', () => {
+  const videos = [
+    '/media/Show/Show - S01E10 - Ten.mkv',
+    '/media/Show/Show - S01E02 - Two.mkv',
+    '/media/Show/Show - S01E01 - One.mkv',
+  ];
+
+  assert.deepEqual(sortVideosByEpisode(videos), [
+    '/media/Show/Show - S01E01 - One.mkv',
+    '/media/Show/Show - S01E02 - Two.mkv',
+    '/media/Show/Show - S01E10 - Ten.mkv',
+  ]);
+});
+
+function createSeriesTree(): string {
+  const root = fs.mkdtempSync(path.join(os.tmpdir(), 'subminer-history-'));
+  const seriesRoot = path.join(root, 'Show');
+  const season1 = path.join(seriesRoot, 'Season-1');
+  const season2 = path.join(seriesRoot, 'Season-2');
+  fs.mkdirSync(season1, { recursive: true });
+  fs.mkdirSync(season2, { recursive: true });
+  fs.mkdirSync(path.join(seriesRoot, 'extras-empty'), { recursive: true });
+  for (const name of ['Show - S01E01.mkv', 'Show - S01E02.mkv', 'Show - S01E03.mkv']) {
+    fs.writeFileSync(path.join(season1, name), '');
+  }
+  fs.writeFileSync(path.join(season2, 'Show - S02E01.mkv'), '');
+  fs.writeFileSync(path.join(season1, 'notes.txt'), '');
+  return seriesRoot;
+}
+
+test('listSeasonDirs returns only video-bearing directories in season order', () => {
+  const seriesRoot = createSeriesTree();
+  try {
+    const seasons = listSeasonDirs(seriesRoot);
+    assert.deepEqual(
+      seasons.map((entry) => entry.name),
+      ['Season-1', 'Season-2'],
+    );
+    assert.deepEqual(
+      seasons.map((entry) => entry.season),
+      [1, 2],
+    );
+  } finally {
+    fs.rmSync(path.dirname(seriesRoot), { recursive: true, force: true });
+  }
+});
+
+test('findNextEpisode advances within a season and across seasons', () => {
+  const seriesRoot = createSeriesTree();
+  try {
+    const season1 = path.join(seriesRoot, 'Season-1');
+    const season2 = path.join(seriesRoot, 'Season-2');
+
+    assert.equal(
+      findNextEpisode(path.join(season1, 'Show - S01E02.mkv')),
+      path.join(season1, 'Show - S01E03.mkv'),
+    );
+    assert.equal(
+      findNextEpisode(path.join(season1, 'Show - S01E03.mkv')),
+      path.join(season2, 'Show - S02E01.mkv'),
+    );
+    assert.equal(findNextEpisode(path.join(season2, 'Show - S02E01.mkv')), null);
+  } finally {
+    fs.rmSync(path.dirname(seriesRoot), { recursive: true, force: true });
+  }
+});
+
+test('findNextEpisode falls back to episode numbers when file was removed', () => {
+  const seriesRoot = createSeriesTree();
+  try {
+    const season1 = path.join(seriesRoot, 'Season-1');
+    const missing = path.join(season1, 'Show - S01E02 - Deleted Cut.mkv');
+    assert.equal(findNextEpisode(missing), path.join(season1, 'Show - S01E03.mkv'));
+  } finally {
+    fs.rmSync(path.dirname(seriesRoot), { recursive: true, force: true });
+  }
+});
+
+test('findNextEpisode advances seasons when a deleted file was the last episode', () => {
+  const seriesRoot = createSeriesTree();
+  try {
+    const season1 = path.join(seriesRoot, 'Season-1');
+    const season2 = path.join(seriesRoot, 'Season-2');
+    const missing = path.join(season1, 'Show - S01E03 - Deleted Cut.mkv');
+    assert.equal(findNextEpisode(missing), path.join(season2, 'Show - S02E01.mkv'));
+  } finally {
+    fs.rmSync(path.dirname(seriesRoot), { recursive: true, force: true });
+  }
+});
+
+const PNG_MAGIC = Buffer.from('89504e470d0a1a0a0000000d49484452', 'hex');
+
+function createHistoryDb(
+  dbPath: string,
+  options: { wal?: boolean; coverArt?: boolean } = {},
+): void {
+  const db = new Database(dbPath);
+  try {
+    if (options.wal) db.run('PRAGMA journal_mode = WAL;');
+    db.run(`
+      CREATE TABLE imm_anime(
+        anime_id INTEGER PRIMARY KEY,
+        canonical_title TEXT,
+        title_romaji TEXT
+      );
+    `);
+    db.run(`
+      CREATE TABLE imm_videos(
+        video_id INTEGER PRIMARY KEY,
+        anime_id INTEGER,
+        source_type INTEGER,
+        source_path TEXT,
+        parsed_title TEXT,
+        parsed_season INTEGER,
+        parsed_episode INTEGER
+      );
+    `);
+    db.run(`
+      CREATE TABLE imm_sessions(
+        session_id INTEGER PRIMARY KEY,
+        video_id INTEGER,
+        started_at_ms TEXT
+      );
+    `);
+    db.run(`INSERT INTO imm_anime VALUES (1, 'Show Season 1', 'Show Romaji');`);
+    db.run(`
+      INSERT INTO imm_videos VALUES
+        (1, 1, 1, '/media/Show/Season-1/Show - S01E01.mkv', 'Show', 1, 1),
+        (2, 1, 1, '/media/Show/Season-1/Show - S01E02.mkv', 'Show', 1, 2),
+        (3, NULL, 2, NULL, 'Remote Show', NULL, NULL),
+        (4, NULL, 1, '', 'Empty Path', NULL, NULL);
+    `);
+    db.run(`
+      INSERT INTO imm_sessions VALUES
+        (1, 1, '1000'),
+        (2, 1, '5000'),
+        (3, 2, '3000'),
+        (4, 3, '9000');
+    `);
+    if (options.coverArt) {
+      db.run(`
+        CREATE TABLE imm_media_art(
+          video_id INTEGER PRIMARY KEY,
+          cover_blob_hash TEXT
+        );
+      `);
+      db.run(`
+        CREATE TABLE imm_cover_art_blobs(
+          blob_hash TEXT PRIMARY KEY,
+          cover_blob BLOB NOT NULL
+        );
+      `);
+      // Art only on video 1; video 2 resolves it through the shared anime_id.
+      db.run(`INSERT INTO imm_media_art VALUES (1, 'hash-1');`);
+      db.query('INSERT INTO imm_cover_art_blobs VALUES (?, ?)').run('hash-1', PNG_MAGIC);
+    }
+    if (options.wal) db.run('PRAGMA wal_checkpoint(TRUNCATE);');
+  } finally {
+    db.close();
+  }
+}
+
+function assertHistoryRows(dbPath: string): void {
+  const rows = queryLocalWatchHistory(dbPath);
+
+  assert.equal(rows.length, 2);
+  assert.equal(rows[0]?.videoId, 1);
+  assert.equal(rows[0]?.lastWatchedMs, 5000);
+  assert.equal(rows[0]?.animeTitle, 'Show Romaji');
+  assert.equal(rows[1]?.videoId, 2);
+  assert.equal(rows[1]?.lastWatchedMs, 3000);
+}
+
+test('queryLocalWatchHistory returns local files ordered by most recent session', () => {
+  const dir = fs.mkdtempSync(path.join(os.tmpdir(), 'subminer-history-db-'));
+  const dbPath = path.join(dir, 'immersion.sqlite');
+  try {
+    createHistoryDb(dbPath);
+    assertHistoryRows(dbPath);
+    const rows = queryLocalWatchHistory(dbPath);
+    assert.equal(rows[0]?.coverBlobHash, null);
+  } finally {
+    fs.rmSync(dir, { recursive: true, force: true });
+  }
+});
+
+test('queryLocalWatchHistory resolves cover hashes directly and via shared anime', () => {
+  const dir = fs.mkdtempSync(path.join(os.tmpdir(), 'subminer-history-art-'));
+  const dbPath = path.join(dir, 'immersion.sqlite');
+  try {
+    createHistoryDb(dbPath, { coverArt: true });
+    const rows = queryLocalWatchHistory(dbPath);
+    assert.equal(rows[0]?.videoId, 1);
+    assert.equal(rows[0]?.coverBlobHash, 'hash-1');
+    assert.equal(rows[1]?.videoId, 2);
+    assert.equal(rows[1]?.coverBlobHash, 'hash-1');
+  } finally {
+    fs.rmSync(dir, { recursive: true, force: true });
+  }
+});
+
+test('materializeCoverArt extracts blobs to the cache dir and reuses cached files', () => {
+  const dir = fs.mkdtempSync(path.join(os.tmpdir(), 'subminer-history-covers-'));
+  const dbPath = path.join(dir, 'immersion.sqlite');
+  const cacheDir = path.join(dir, 'covers');
+  try {
+    createHistoryDb(dbPath, { coverArt: true });
+
+    const covers = materializeCoverArt(
+      dbPath,
+      ['hash-1', 'hash-1', null, 'hash-missing'],
+      cacheDir,
+    );
+    const coverPath = covers.get('hash-1');
+    assert.ok(coverPath);
+    assert.equal(path.extname(coverPath!), '.png');
+    assert.ok(fs.statSync(coverPath!).size > 0);
+    assert.equal(covers.has('hash-missing'), false);
+
+    // Cached file is reused even when the database has disappeared.
+    fs.rmSync(dbPath);
+    const cachedCovers = materializeCoverArt(dbPath, ['hash-1'], cacheDir);
+    assert.equal(cachedCovers.get('hash-1'), coverPath);
+  } finally {
+    fs.rmSync(dir, { recursive: true, force: true });
+  }
+});
+
+test('materializeCoverArt rejects cover hashes that escape the cache dir', () => {
+  const dir = fs.mkdtempSync(path.join(os.tmpdir(), 'subminer-history-cover-safety-'));
+  const dbPath = path.join(dir, 'immersion.sqlite');
+  const cacheDir = path.join(dir, 'covers');
+  const unsafeHash = '../escape';
+  try {
+    createHistoryDb(dbPath, { coverArt: true });
+    const db = new Database(dbPath);
+    try {
+      db.query('INSERT INTO imm_cover_art_blobs VALUES (?, ?)').run(unsafeHash, PNG_MAGIC);
+    } finally {
+      db.close();
+    }
+
+    const covers = materializeCoverArt(dbPath, [unsafeHash], cacheDir);
+
+    assert.equal(covers.has(unsafeHash), false);
+    assert.equal(fs.existsSync(path.join(dir, 'escape.png')), false);
+  } finally {
+    fs.rmSync(dir, { recursive: true, force: true });
+  }
+});
+
+test('detectImageExtension identifies common cover formats', () => {
+  assert.equal(detectImageExtension(PNG_MAGIC), '.png');
+  assert.equal(detectImageExtension(Buffer.from([0xff, 0xd8, 0xff, 0xe0])), '.jpg');
+  assert.equal(detectImageExtension(Buffer.from('RIFF0000WEBPVP8 ', 'ascii')), '.webp');
+  assert.equal(detectImageExtension(Buffer.from('GIF89a', 'ascii')), '.gif');
+  assert.equal(detectImageExtension(Buffer.from('unknown', 'ascii')), '.jpg');
+});
+
+test('groupHistoryBySeries backfills cover hash from older rows of the same series', () => {
+  const rows = [
+    makeRow({ videoId: 2, parsedEpisode: 2, lastWatchedMs: 3000, coverBlobHash: null }),
+    makeRow({ videoId: 1, parsedEpisode: 1, lastWatchedMs: 1000, coverBlobHash: 'hash-1' }),
+  ];
+
+  const series = groupHistoryBySeries(rows, () => true);
+
+  assert.equal(series.length, 1);
+  assert.equal(series[0]?.lastWatched.videoId, 2);
+  assert.equal(series[0]?.coverBlobHash, 'hash-1');
+});
+
+test('queryLocalWatchHistory reads a cleanly-closed WAL database', () => {
+  const dir = fs.mkdtempSync(path.join(os.tmpdir(), 'subminer-history-wal-'));
+  const dbPath = path.join(dir, 'immersion.sqlite');
+  try {
+    createHistoryDb(dbPath, { wal: true });
+    // Reproduce the state after the app shuts down cleanly: WAL journal mode
+    // with no -wal/-shm sidecar files on disk. A read-only connection then
+    // fails at query time because it cannot recreate them.
+    fs.rmSync(`${dbPath}-wal`, { force: true });
+    fs.rmSync(`${dbPath}-shm`, { force: true });
+    assertHistoryRows(dbPath);
+  } finally {
+    fs.rmSync(dir, { recursive: true, force: true });
+  }
+});
+
+test('isReadonlyWalRetryError only accepts readonly errors from WAL-mode databases', () => {
+  const dir = fs.mkdtempSync(path.join(os.tmpdir(), 'subminer-history-retry-'));
+  const walDbPath = path.join(dir, 'wal.sqlite');
+  const rollbackDbPath = path.join(dir, 'rollback.sqlite');
+  try {
+    createHistoryDb(walDbPath, { wal: true });
+    createHistoryDb(rollbackDbPath);
+
+    assert.equal(
+      isReadonlyWalRetryError(
+        Object.assign(new Error('attempt to write a readonly database'), {
+          code: 'SQLITE_READONLY',
+        }),
+        walDbPath,
+      ),
+      true,
+    );
+    assert.equal(
+      isReadonlyWalRetryError(
+        Object.assign(new Error('unable to open database file'), {
+          code: 'SQLITE_CANTOPEN',
+        }),
+        walDbPath,
+      ),
+      true,
+    );
+    assert.equal(
+      isReadonlyWalRetryError(new Error('no such table: imm_sessions'), walDbPath),
+      false,
+    );
+    assert.equal(
+      isReadonlyWalRetryError(
+        Object.assign(new Error('attempt to write a readonly database'), {
+          code: 'SQLITE_READONLY',
+        }),
+        rollbackDbPath,
+      ),
+      false,
+    );
+  } finally {
+    fs.rmSync(dir, { recursive: true, force: true });
+  }
+});

launcher/history.ts

@@ -0,0 +1,4 @@
+export * from './history-art.js';
+export * from './history-db.js';
+export * from './history-navigation.js';
+export type { HistorySeriesEntry, HistoryVideoRow, SeasonDirEntry } from './history-types.js';

launcher/jellyfin.test.ts

@@ -29,6 +29,7 @@ function createArgs(): Args {
     texthookerOnly: false,
     texthookerOpenBrowser: false,
     useRofi: false,
+    history: false,
     logLevel: 'info',
     logRotation: 7,
     passwordStore: '',

launcher/log.test.ts

@@ -2,9 +2,10 @@ import test from 'node:test';
 import assert from 'node:assert/strict';
 import path from 'node:path';
 import { getDefaultLauncherLogFile, getDefaultMpvLogFile } from './types.js';
+import { localDateKey } from '../src/shared/log-files.js';
 
 test('getDefaultMpvLogFile uses APPDATA on windows', () => {
-  const today = new Date().toISOString().slice(0, 10);
+  const today = localDateKey(new Date());
   const resolved = getDefaultMpvLogFile({
     platform: 'win32',
     homeDir: 'C:\\Users\\tester',
@@ -20,7 +21,7 @@ test('getDefaultMpvLogFile uses APPDATA on windows', () => {
 });
 
 test('getDefaultLauncherLogFile uses launcher prefix', () => {
-  const today = new Date().toISOString().slice(0, 10);
+  const today = localDateKey(new Date());
   const resolved = getDefaultLauncherLogFile({
     platform: 'linux',
     homeDir: '/home/tester',

launcher/main.ts

@@ -21,6 +21,7 @@ import { runDictionaryCommand } from './commands/dictionary-command.js';
 import { runLogsCommand } from './commands/logs-command.js';
 import { runStatsCommand } from './commands/stats-command.js';
 import { runJellyfinCommand } from './commands/jellyfin-command.js';
+import { runHistoryCommand } from './commands/history-command.js';
 import { runPlaybackCommand } from './commands/playback-command.js';
 import { runUpdateCommand } from './commands/update-command.js';
 
@@ -142,6 +143,16 @@ async function main(): Promise<void> {
     return;
   }
 
+  if (appContext.args.history) {
+    const selected = await runHistoryCommand(appContext);
+    if (!selected) {
+      log('info', args.logLevel, 'No watch history selection made, exiting');
+      return;
+    }
+    appContext.args.target = selected;
+    appContext.args.targetKind = 'file';
+  }
+
   await runPlaybackCommand(appContext);
 }
 

launcher/mpv.test.ts

@@ -570,6 +570,7 @@ function makeArgs(overrides: Partial<Args> = {}): Args {
     texthookerOnly: false,
     texthookerOpenBrowser: false,
     useRofi: false,
+    history: false,
     logLevel: 'error',
     logRotation: 7,
     passwordStore: '',

launcher/parse-args.test.ts

@@ -42,6 +42,19 @@ test('parseArgs maps root settings window option', () => {
   assert.equal(parsed.settings, true);
 });
 
+test('parseArgs maps root watch history flags', () => {
+  const shortParsed = parseArgs(['-H'], 'subminer', {});
+  const longParsed = parseArgs(['--history'], 'subminer', {});
+  const rofiParsed = parseArgs(['-R', '-H'], 'subminer', {});
+  const defaultParsed = parseArgs([], 'subminer', {});
+
+  assert.equal(shortParsed.history, true);
+  assert.equal(longParsed.history, true);
+  assert.equal(rofiParsed.history, true);
+  assert.equal(rofiParsed.useRofi, true);
+  assert.equal(defaultParsed.history, false);
+});
+
 test('parseArgs maps root update flags without conflicting with jellyfin username', () => {
   const shortParsed = parseArgs(['-u'], 'subminer', {});
   const longParsed = parseArgs(['--update'], 'subminer', {});

launcher/smoke.e2e.test.ts

@@ -554,13 +554,15 @@ test(
           typeof mpvError === 'string' && /eperm|operation not permitted/i.test(mpvError);
 
         assert.equal(result.status, unixSocketDenied ? 3 : 0);
-        assert.equal(appEntries.length > 0, true);
-        assert.equal(
-          appEntries.every((entry) =>
-            (entry.argv as string[]).includes('--ensure-linux-runtime-plugin-assets'),
-          ),
-          true,
-        );
+        if (process.platform === 'linux') {
+          assert.equal(appEntries.length > 0, true);
+          assert.equal(
+            appEntries.every((entry) =>
+              (entry.argv as string[]).includes('--ensure-linux-runtime-plugin-assets'),
+            ),
+            true,
+          );
+        }
         assert.equal(appStartEntries.length, 0);
         assert.equal(appStopEntries.length, 0);
         assert.equal(controlEntries.length, 1);
@@ -617,11 +619,13 @@ test(
         /subminer-auto_start_pause_until_ready_owns_initial_pause=yes/,
       );
       assert.match(result.stdout, /pause mpv until overlay and tokenization are ready/i);
-      assert.match(result.stdout, /managed plugin\/theme assets/i);
-      assert.equal(
-        fs.existsSync(path.join(smokeCase.xdgDataHome, 'SubMiner', 'themes', 'subminer.rasi')),
-        true,
-      );
+      if (process.platform === 'linux') {
+        assert.match(result.stdout, /managed plugin\/theme assets/i);
+        assert.equal(
+          fs.existsSync(path.join(smokeCase.xdgDataHome, 'SubMiner', 'themes', 'subminer.rasi')),
+          true,
+        );
+      }
     });
   },
 );

launcher/types.ts

@@ -112,6 +112,7 @@ export interface Args {
   texthookerOnly: boolean;
   texthookerOpenBrowser: boolean;
   useRofi: boolean;
+  history: boolean;
   logLevel: LogLevel;
   logRotation: LogRotation;
   passwordStore: string;

release/prerelease-notes.md

@@ -1,168 +1,24 @@
 > This is a prerelease build for testing. Stable changelog and docs-site updates remain pending until the final stable release.
 
+<!-- prerelease-base-version: 0.17.1 -->
+
 ## Highlights
 ### Added
 
-- **Settings Window:** A dedicated Settings window is now available via `subminer --settings` or `subminer settings`, organized into Appearance, Behavior, Anki, Input, and Integration sections.
-  - Includes click-to-learn keybinding controls, an AnkiConnect deck dropdown that auto-fills from Yomitan's current mining deck, and AnkiConnect-backed deck, field, and note-type pickers.
-  - Live-saves changes for subtitle CSS declarations, stats keys, logging level, Anki field mappings, sentence card model, and other annotation and runtime options; search narrows across all categories including multi-word terms. AI and translation settings remain config-file only.
-
-- **Auto-Updater:** SubMiner can now check for and apply updates from the system tray or by running `subminer -u`, with checksum verification and configurable update notifications.
-  - The `subminer` launcher and Linux rofi theme update automatically alongside the app.
-  - Set `updates.channel` to `"prerelease"` to receive beta and RC builds.
-
-- **First-Run Setup:** A new optional setup flow installs Bun and the `subminer` command-line launcher on Linux, macOS, and Windows.
-  - Windows users get a `subminer.cmd` PATH shim so `subminer` works in any terminal without manually adding `SubMiner.exe` to PATH.
-  - Setup recognizes existing `subminer` installs in Homebrew or user PATH directories and avoids writing into Homebrew-owned paths. An Open SubMiner Settings button is included on completion; the standalone setup app quits after finishing.
-
-- **Character Portraits:** Character-name subtitle matches can now show optional inline AniList character portraits.
-  - Manual AniList title overrides are scoped per media directory so separate season folders keep independent character dictionary selections.
-
-- **Log Export:** Sanitized log ZIP archives can be exported from the tray menu or by running `subminer logs -e`, with home-directory usernames redacted from the exported contents.
-
-- **Logging Configuration:** SubMiner's logging level is now forwarded into launcher-started and Windows shortcut-started mpv sessions, controlling mpv log verbosity and plugin script logging.
-  - The new `logging.rotation` config sets daily log retention (default 7 days). `logging.files` toggles let you enable or disable per-component log files; mpv logs are off by default unless explicitly enabled.
-
-- **Yomitan Popup Visibility:** The new `subtitleStyle.primaryVisibleOnYomitanPopup` option keeps hover-mode primary subtitles visible while a Yomitan lookup popup is open.
-
-- **Launcher:** `subminer --version` / `subminer -v` now prints the installed app version. The new `mpv.profile` config option passes an mpv profile to SubMiner-managed mpv launches, and bundled mpv plugin startup options are now configurable from SubMiner config.
-
-### Changed
-
-- **Subtitle Appearance:** Primary and secondary subtitle appearance now use color controls plus CSS declaration editors, saved as `subtitleStyle.css` and `subtitleStyle.secondary.css`; sidebar appearance uses `subtitleSidebar.css`.
-  - Default font stack updated to `Hiragino Sans, M PLUS 1, Source Han Sans JP, Noto Sans CJK JP`; default text shadow is stronger, JLPT underlines are thicker, and the frequency `topX` threshold defaults to `10000`.
-  - Existing configs are migrated automatically: legacy appearance options and hover token colors fold into `subtitleStyle.css`, and user config files are preserved.
-
-- **Known-Word Colors:** Known-word and N+1 annotation colors moved to `subtitleStyle.knownWordColor` and `subtitleStyle.nPlusOneColor`. Legacy Anki color keys remain accepted with deprecation warnings.
-  - N+1 highlighting is preserved for configs that already had it enabled; new configs leave it disabled unless `ankiConnect.nPlusOne.enabled` is set explicitly.
-
-- **Character Dictionary:** Entries are now scoped to the current AniList media and generate Japanese name aliases only, so raw romanized or English aliases no longer appear as separate results.
-  - A new `Ctrl/Cmd+D` manager modal lets you remove, reorder, or override loaded dictionary entries.
-  - The in-app AniList title selector now waits for an explicit search rather than triggering automatically; the search box is prefilled from the current filename guess.
-
-- **Linux Updater:** Tray "Check for Updates" now installs the new AppImage automatically via `electron-updater`, matching the macOS and Windows update flow. System-package-managed AppImages and non-AppImage launches fall back to the GitHub-asset flow.
-
-- **Subsync:** The subtitle sync dialog now always opens the manual picker; the `subsync.defaultMode` config option has been removed.
-
-- **Jellyfin Setup:** The server presets dropdown is replaced by a single editable server URL field.
-
-- **Defaults:** Jellyfin remote-session startup warmup and character-name subtitle highlighting now default to off.
-
-- **Runtime:** The bundled Electron runtime is updated from 39.8.6 to 42.2.0.
-
-- **Subtitle Delay Shortcuts:** Default overlay subtitle delay and step bindings now match mpv's own defaults.
-  - `z`, `Z`, and `x` adjust sub-delay; `Ctrl+Shift+Left/Right` runs native sub-step and shows the current delay on the OSD.
-  - The previous SubMiner-only adjacent-cue delay action has been removed.
-
-- **Update Notifications:** New installs default to overlay-only update notifications instead of overlay plus system notifications.
+- **YouTube Media Cache Mode**: A new `youtube.mediaCache.mode` setting (`direct` or `background`) lets you choose how SubMiner extracts audio and image from YouTube cards.
+  - In background mode, SubMiner creates a text-only card immediately, downloads a yt-dlp media cache (capped at 720p by default), and fills audio and image fields once the file is ready — with overlay and OSD notifications when the download starts and when media is available.
+  - If a background download fails, SubMiner now notifies you and clears any pending media updates rather than leaving cards silently incomplete.
 
 ### Fixed
 
-- **macOS Overlay:** Significantly improved overlay focus and stability across a range of scenarios.
-  - The overlay hides when mpv loses focus, is minimized, or is no longer the foreground target; stays stable through transient window-tracking misses; remains correctly layered during stats mouse passthrough; and opens over fullscreen mpv without switching Spaces.
-  - Passthrough is fixed so mpv controls stay clickable before hovering a subtitle bar. The compiled mpv window helper is now correctly bundled, preventing the overlay from falling back to a slower startup path on first launch.
-  - Yomitan popup focus is restored after card mining or popup reload; clicking transparent overlay space correctly closes the popup and returns passthrough to mpv without a hide/reappear cycle.
+- **Log Export**: Log filenames now use your local date, so exporting logs near midnight no longer pulls stale files from the previous UTC day. Export redaction has also been expanded to mask a broader range of sensitive data, including IP addresses, email addresses, authentication and cookie headers, yt-dlp cookie arguments, URL credentials, and signed YouTube media URLs.
 
-- **Linux/Hyprland Overlay:** Overlay placement refreshes after leaving mpv fullscreen so the visible overlay stays aligned to the player.
-  - The overlay stays stacked above mpv after click-to-focus events and is suspended while the in-player stats window is open.
-  - Settings windows (SubMiner and Yomitan) now open above the subtitle overlay; the overlay hides immediately when the character dictionary modal opens, including while AniList lookup is in progress.
-  - Auto-paused startup is more reliable: the overlay stays interactive during the first measurement gap, startup subtitle cache misses paint raw text before tokenization finishes, and temporarily empty mpv subtitle reads are resolved before synthetic warm readiness can resume playback.
-
-- **Playback:** The first subtitle is primed before autoplay resumes so the overlay renders text before video playback begins. Launcher-owned videos quit SubMiner when playback ends while background and tray sessions stay alive.
-  - The visible overlay and subtitle stream stay alive after restarting SubMiner from the `y-r` shortcut, with correct Linux bounds reapplication and user-paused playback preserved through readiness gates.
-  - The visible overlay remains active when mpv advances to the next playlist item, even when the next episode loads after the warm transition delay.
-
-- **Jellyfin Playback:** Resolved a wide range of discovery and playback issues: the active item is no longer reloaded during startup, paused mpv is no longer misreported as playing, startup unpause no longer repeats after a manual pause or `y-t` toggle, and duplicate ready signals no longer re-show the overlay.
-  - Discovery now correctly handles delayed Japanese subtitle selection and prevents later-loading foreign tracks from stealing the active Japanese track.
-  - Discovery resume correctly handles `StartPositionTicks: 0` for items with saved progress.
-
-- **Jellyfin Subtitles:** Improved subtitle timing by preferring default embedded streams over external sidecars, stripping Jellyfin's server-selected stream from playback URLs, suppressing mpv auto-selection while SubMiner stages managed tracks, and automatically correcting Japanese-vs-English cue timeline offsets.
-  - Per-stream subtitle delay shifts are restored on load. Track selection now tolerates transient `track-list` read failures and numeric string track IDs on Linux.
-
-- **Jellyfin Overlay:** The visible subtitle overlay now shows automatically during Jellyfin playback so `subtitleStyle` appearance applies, and the bundled mpv plugin is injected when SubMiner auto-launches mpv so mpv-side keybindings work without overlay focus.
-  - The `y-t` overlay toggle is reliable and remains sticky across stream redirects.
-  - Passive Linux/Hyprland overlay shows no longer steal keyboard focus from mpv.
-
-- **Jellyfin Remote Progress:** Fixed progress sync for mpv/SubMiner seek jumps, stopped sessions, startup path changes, and Linux websocket reconnect windows.
-  - Play and Resume are now distinct: Play starts from the beginning while Resume starts at the saved position.
-  - Final progress reports use SubMiner's last known position when mpv resets during stop.
-
-- **Jellyfin Identity:** Cast device identity is now derived from the OS hostname. Multiple SubMiner installs no longer share the same remote-session identity.
-
-- **Jellyfin Tray:** The discovery tray checkbox stays in sync on Linux after tray, CLI, or startup remote-session changes. Stale discovery sessions restart automatically when the server no longer lists the SubMiner cast target.
-
-- **Jellyfin Setup:** Fixed the Windows login flow with an IPC bridge and immediate progress feedback; unreachable servers time out with an inline error instead of hanging.
-
-- **AniList Progress:** Threshold checks now use fresh playback position data so updates fire correctly when playback reaches or skips past the watched threshold.
-  - Season-specific results are preferred for multi-season files, with a clear message when the matched season is not in Planning or Watching status.
-  - Repeated missing-token checks no longer exhaust AniList retry attempts or create duplicate dead-letter entries for the same episode.
-  - Manual AniList linking from the stats anime page now strips generated `Season N` suffixes from automatic searches so results match the base title correctly.
-
-- **Anki:** Sentence-audio padding is now opt-in by default; animated AVIF freeze-frame duration is correctly aligned to word audio length without double-counting padding.
-  - Multi-line sentence mining stays aligned for repeated subtitle text; Kiku duplicate-card detection and merge flow are fixed; clipboard card updates from YouTube use mpv's resolved stream URLs; sentence cards refresh the secondary subtitle before saving.
-  - Known-word cache append is fixed when no default Anki mining deck is configured but multiple known-word deck field mappings are present.
-
-- **YouTube:** Primary subtitles are downloaded to temporary local files so the primary bar and sidebar read the same source, with cleanup on reload and quit.
-  - False load-failure notifications are suppressed. Launcher-managed playback creates the tray icon when attaching to an already-running process, and app-owned playback no longer lets the mpv plugin start a second SubMiner instance.
-
-- **Character Dictionary:** Surname honorifics are now matched for Japanese localized aliases embedded in AniList alternative names; cached snapshots are regenerated to include them.
-  - Cached media matches are reused when loading a title with an existing snapshot, avoiding redundant AniList search requests. Manager keyboard shortcuts are correctly forwarded to the mpv plugin.
-
-- **Updater:** Update checks are more stable across platforms: Linux uses GitHub release metadata; `subminer -u` can update independently of the tray app; macOS update dialogs reliably appear in the foreground.
-  - Builds that cannot apply native updates show a manual-install message instead of a restart prompt. Windows retains the native NSIS update path while routing updater HTTP through the main process.
-  - Linux updates now correctly create and refresh the launcher runtime plugin copy and rofi theme alongside AppImage and launcher updates; both support assets are auto-installed from the bundled app on first launch if either is missing.
-
-- **Setup - macOS:** First-run setup recognizes existing `subminer` installs in Homebrew or user PATH directories and avoids writing into Homebrew-owned paths.
-  - `subminer app --setup` opens the setup flow even when SubMiner is already running. The standalone setup app quits after completing first-run setup, and `subminer settings` exits cleanly when the window is closed.
-
-- **Tray App:** Fixed several lifecycle issues: the tray stays running when Yomitan settings are closed; a close-only menu prevents accidentally quitting the tray app; an in-page close button is available on Hyprland where native window controls are unavailable.
-  - Settings loading no longer blocks other tray actions; the embedded popup preview is disabled to prevent renderer hangs during sidebar navigation; extension refreshes at startup are serialized; the session help modal closes correctly without mpv running.
-  - On Windows, "Open SubMiner Setup" now correctly opens the setup window after first-run setup is complete. System notifications now correctly show the SubMiner app icon when no custom notification image is provided.
-
-- **Launcher:** Launcher-opened videos reuse an already-running background SubMiner instance and correctly reapply preferred subtitles on warm launches. Videos stay paused until subtitle priming and tokenization readiness complete.
-  - `subminer settings` on macOS no longer emits Electron menu diagnostics and exits cleanly when the window is closed. Linux first-run launcher installs build with a valid Bun shebang; `subminer app` on Linux returns control to the terminal immediately.
-  - On Windows, managed mpv launches from a background instance correctly retarget the new mpv socket, bind to the player window, and receive startup overlay options.
-
-- **Subtitle Frequency:** Frequency highlighting is preserved for determiner-led noun compounds like `その場` while standalone determiners are still filtered. Annotations are corrected for Yomitan single-token compounds with internal particles like `目の前`.
-
-- **Subtitle Annotation Prefetch:** Cached annotations and character images are ready for more live subtitle changes without delaying raw subtitle display.
-
-- **Shortcuts:** Native mpv menu shortcuts are disabled during managed macOS playback so SubMiner shortcuts also work while mpv has focus. Session shortcuts including `stats.markWatchedKey` are correctly wired through mpv. The visible overlay receives focus when entering multi-line copy/mine selection so number keys work on macOS and Windows.
-
-- **Stats:** In-player stats layering is fixed so delete confirmations, overlay modals, and update-check dialogs appear above the stats window. Jellyfin playback stats are grouped under item metadata so watched episodes merge with matching local library titles and display clean names.
-
-- **Sidebar:** Yomitan lookup popups opened from the subtitle sidebar now correctly pause playback when popup auto-pause is enabled. Mined cards use audio and images from the clicked subtitle line rather than the current primary line.
-
-- **Controller:** Config and debug shortcuts stay closed while controller support is disabled, with a notice to enable `controller.enabled`. Learn mode can be entered from the edit pencil or binding badge; remaps are saved per controller profile, and individual bindings can be reset to their defaults.
-
-- **Discord Rich Presence:** Presence no longer falls back to Jellyfin stream URLs; Jellyfin playback titles are primed before loading tokenized streams so presence shows the show/episode title.
-
-- **WebSocket:** The regular subtitle WebSocket now sends plain text only; annotation spans and token metadata are sent exclusively on the annotation WebSocket.
-
-- **Windows Startup:** Fatal startup errors now show a native error dialog and write details to the app log instead of exiting silently.
-
-- **Yomitan:** Fixed popups not opening when overlay startup races the Yomitan extension load.
-
-- **Subtitle Sync Modal:** Fixed a macOS issue where the modal would flash and disappear on the first attempt, or leave stale state after syncing.
-
-### Docs
-
-- **Versioned Docs:** Stable docs are now published at the site root with current development docs under `/main/`.
-  - Fixed versioned docs navigation so archived pages keep local links under the selected version, the version switcher no longer nests paths incorrectly, and local dev version routes serve warmed archive files instead of redirecting to production.
-
-- **Configuration Reference:** All previously undocumented config options are now covered, including `subtitleStyle.primaryDefaultMode`, `stats.markWatchedKey`, `immersionTracking.lifetimeSummaries.*`, and all seven `mpv.*` launcher options. Updated known-word cache docs and examples to recommend expression/word fields.
-
-- **Architecture Docs:** Added a Playback Startup Flow diagram and a Runtime Sockets section and diagram to the IPC + Runtime Contracts page, with cross-reference pointers in the MPV Plugin and Troubleshooting pages.
-
-- **Linux Update Flow:** Documented that Linux update flows manage the launcher runtime plugin copy and rofi theme from `subminer-assets.tar.gz`, and that normal launcher playback auto-installs those managed support assets if either one is missing.
+- **YouTube Card Media Reliability**: Direct stream extraction now uses safer ffmpeg options and skips stale or cached stream map entries to reduce failed media generation. Background cache downloads are hardened with IPv4 and extractor retry flags, stale cache files are cleaned up on startup and before new downloads, and in-flight background downloads are stopped automatically when switching back to direct mode.
 
 ## What's Changed
 
-- Replace subtitle delay actions with native mpv keybindings by @ksyasuda in #120
-- fix(stats): strip Season N suffix from AniList title searches by @ksyasuda in #121
-- fix(overlay): preserve visible state across playlist item transitions by @ksyasuda in #124
-- fix(overlay): restore macOS Yomitan popup focus without breaking click-away by @ksyasuda in #125
-- fix(linux): auto-install managed plugin copy; include in asset updates by @ksyasuda in #127
+- feat(youtube): add mediaCache mode and safer stream media extraction by @ksyasuda in #130
+- fix(logs): use local date for log filenames and expand export redaction by @ksyasuda in #131
 
 ## Installation
 

scripts/build-changelog.test.ts

@@ -605,6 +605,7 @@ test('writePrereleaseNotesForVersion writes cumulative beta notes without mutati
 
     const prereleaseNotes = fs.readFileSync(outputPath, 'utf8');
     assert.match(prereleaseNotes, /^> This is a prerelease build for testing\./m);
+    assert.match(prereleaseNotes, /<!-- prerelease-base-version: 0\.11\.3 -->/);
     assert.match(prereleaseNotes, /## Highlights\n### Added\n- Polished: added entry\./);
     assert.match(prereleaseNotes, /### Fixed\n- Polished: fixed entry\./);
     assert.match(prereleaseNotes, /## Installation\n\nSee the README and docs\/installation guide/);
@@ -620,6 +621,8 @@ test('writePrereleaseNotesForVersion reuses existing prerelease notes when addin
   const existingNotes = [
     '> This is a prerelease build for testing. Stable changelog and docs-site updates remain pending until the final stable release.',
     '',
+    '<!-- prerelease-base-version: 0.11.3 -->',
+    '',
     '## Highlights',
     '### Added',
     '- Overlay: Previous beta entry.',
@@ -679,6 +682,61 @@ test('writePrereleaseNotesForVersion reuses existing prerelease notes when addin
   }
 });
 
+test('writePrereleaseNotesForVersion ignores unmarked prerelease notes from an older release line', async () => {
+  const { writePrereleaseNotesForVersion } = await loadModule();
+  const workspace = createWorkspace('prerelease-ignore-unmarked-old-notes');
+  const projectRoot = path.join(workspace, 'SubMiner');
+  const existingNotes = [
+    '> This is a prerelease build for testing. Stable changelog and docs-site updates remain pending until the final stable release.',
+    '',
+    '## Highlights',
+    '### Added',
+    '- Settings Window: Previous release line entry.',
+    '',
+    '## Installation',
+    '',
+    'See the README and docs/installation guide for full setup steps.',
+    '',
+  ].join('\n');
+
+  fs.mkdirSync(path.join(projectRoot, 'changes'), { recursive: true });
+  fs.mkdirSync(path.join(projectRoot, 'release'), { recursive: true });
+  fs.writeFileSync(
+    path.join(projectRoot, 'package.json'),
+    JSON.stringify({ name: 'subminer', version: '0.17.0-beta.1' }, null, 2),
+    'utf8',
+  );
+  fs.writeFileSync(path.join(projectRoot, 'release', 'prerelease-notes.md'), existingNotes, 'utf8');
+  fs.writeFileSync(
+    path.join(projectRoot, 'changes', '001.md'),
+    [
+      'type: changed',
+      'area: overlay',
+      '',
+      '- Replaced subtitle delay actions with native mpv keybindings.',
+    ].join('\n'),
+    'utf8',
+  );
+
+  try {
+    const stub = defaultStubClaude();
+    const outputPath = writePrereleaseNotesForVersion({
+      cwd: projectRoot,
+      version: '0.17.0-beta.1',
+      deps: { runClaude: stub.runClaude },
+    });
+
+    assert.equal(stub.calls.length, 1, 'prerelease should issue exactly one Claude call');
+    assert.doesNotMatch(stub.calls[0]!.input, /EXISTING PRERELEASE NOTES/);
+    assert.doesNotMatch(stub.calls[0]!.input, /Settings Window: Previous release line entry/);
+
+    const prereleaseNotes = fs.readFileSync(outputPath, 'utf8');
+    assert.match(prereleaseNotes, /### Changed\n- Polished: changed entry\./);
+  } finally {
+    fs.rmSync(workspace, { recursive: true, force: true });
+  }
+});
+
 test('writePrereleaseNotesForVersion prompts Claude to revise stale prerelease bullets instead of appending fix churn', async () => {
   const { writePrereleaseNotesForVersion } = await loadModule();
   const workspace = createWorkspace('prerelease-net-outcome-prompt');
@@ -686,6 +744,8 @@ test('writePrereleaseNotesForVersion prompts Claude to revise stale prerelease b
   const existingNotes = [
     '> This is a prerelease build for testing. Stable changelog and docs-site updates remain pending until the final stable release.',
     '',
+    '<!-- prerelease-base-version: 0.12.0 -->',
+    '',
     '## Highlights',
     '### Added',
     '- Config Window: Previous beta entry.',

scripts/build-changelog.ts

@@ -93,6 +93,36 @@ function isSupportedPrereleaseVersion(version: string): boolean {
   return /^\d+\.\d+\.\d+-(beta|rc)\.\d+$/u.test(normalizeVersion(version));
 }
 
+function resolvePrereleaseBaseVersion(version: string): string {
+  const match = /^(\d+\.\d+\.\d+)-(?:beta|rc)\.\d+$/u.exec(normalizeVersion(version));
+  if (!match) {
+    throw new Error(
+      `Unsupported prerelease version (${version}). Expected x.y.z-beta.N or x.y.z-rc.N.`,
+    );
+  }
+  return match[1]!;
+}
+
+function renderPrereleaseBaseVersionMarker(version: string): string {
+  return `<!-- prerelease-base-version: ${resolvePrereleaseBaseVersion(version)} -->`;
+}
+
+function extractPrereleaseBaseVersionMarker(notes: string): string | null {
+  return /<!--\s*prerelease-base-version:\s*(\d+\.\d+\.\d+)\s*-->/u.exec(notes)?.[1] ?? null;
+}
+
+function stripPrereleaseMetadata(notes: string): string {
+  return notes.replace(/<!--\s*prerelease-base-version:\s*\d+\.\d+\.\d+\s*-->\s*/u, '').trim();
+}
+
+function resolveReusablePrereleaseNotes(notes: string, version: string): string | undefined {
+  const existingBaseVersion = extractPrereleaseBaseVersionMarker(notes);
+  if (existingBaseVersion !== resolvePrereleaseBaseVersion(version)) {
+    return undefined;
+  }
+  return stripPrereleaseMetadata(notes);
+}
+
 function verifyRequestedVersionMatchesPackageVersion(
   options: Pick<ChangelogOptions, 'cwd' | 'version' | 'deps'>,
 ): void {
@@ -669,13 +699,16 @@ function renderReleaseNotes(
     disclaimer?: string;
     contributions?: Contribution[];
     contributorSections?: string[];
+    metadata?: string[];
   },
 ): string {
   const prefix = options?.disclaimer ? [options.disclaimer, ''] : [];
+  const metadata = options?.metadata?.length ? [...options.metadata, ''] : [];
   const contributorSections =
     options?.contributorSections ?? renderContributorsSections(options?.contributions ?? []);
   return [
     ...prefix,
+    ...metadata,
     '## Highlights',
     changes,
     '',
@@ -705,6 +738,7 @@ function writeReleaseNotesFile(
     outputPath?: string;
     contributions?: Contribution[];
     contributorSections?: string[];
+    metadata?: string[];
   },
 ): string {
   const mkdirSync = deps?.mkdirSync ?? fs.mkdirSync;
@@ -1038,7 +1072,7 @@ export function writePrereleaseNotesForVersion(options?: ChangelogOptions): stri
 
   const prereleaseNotesPath = path.join(cwd, PRERELEASE_NOTES_PATH);
   const existingReleaseNotes = existsSync(prereleaseNotesPath)
-    ? readFileSync(prereleaseNotesPath, 'utf8')
+    ? resolveReusablePrereleaseNotes(readFileSync(prereleaseNotesPath, 'utf8'), version)
     : undefined;
   const changes = polishFragmentsWithClaude(fragments, {
     mode: 'release-notes',
@@ -1052,6 +1086,7 @@ export function writePrereleaseNotesForVersion(options?: ChangelogOptions): stri
       '> This is a prerelease build for testing. Stable changelog and docs-site updates remain pending until the final stable release.',
     outputPath: PRERELEASE_NOTES_PATH,
     contributions,
+    metadata: [renderPrereleaseBaseVersionMarker(version)],
   });
 }
 

src/anki-integration.test.ts

@@ -5,6 +5,7 @@ import * as os from 'os';
 import * as path from 'path';
 import { AnkiIntegration } from './anki-integration';
 import { FieldGroupingMergeCollaborator } from './anki-integration/field-grouping-merge';
+import type { MediaInput } from './media-input';
 import { AnkiConnectConfig } from './types';
 
 type TestOverlayNotificationPayload = {
@@ -24,6 +25,13 @@ interface IntegrationTestContext {
   stateDir: string;
 }
 
+function describeMediaInputForTest(input: MediaInput): string {
+  if (typeof input === 'string') {
+    return input;
+  }
+  return `${input.path}:${input.source ?? 'raw'}`;
+}
+
 function createIntegrationTestContext(
   options: {
     highlightEnabled?: boolean;
@@ -95,7 +103,7 @@ function createIntegrationTestContext(
     knownWordsScope: string;
     knownWordsLastRefreshedAtMs: number;
   };
-  privateState.knownWordsScope = 'is:note';
+  privateState.knownWordsScope = 'all';
   privateState.knownWordsLastRefreshedAtMs = Date.now();
 
   return {
@@ -324,6 +332,119 @@ test('AnkiIntegration resolves merged-away note ids to the kept note id', () =>
   }
 });
 
+function processSentenceWithConfig(
+  config: Partial<AnkiConnectConfig>,
+  mpvSentence: string,
+  noteFields: Record<string, string>,
+): string {
+  const integration = new AnkiIntegration(config as AnkiConnectConfig, {} as never, {} as never);
+  return (
+    integration as unknown as {
+      processSentence: (sentence: string, fields: Record<string, string>) => string;
+    }
+  ).processSentence(mpvSentence, noteFields);
+}
+
+function processSentenceFuriganaWithConfig(
+  config: Partial<AnkiConnectConfig>,
+  sentenceFurigana: string,
+  noteFields: Record<string, string>,
+): string {
+  const integration = new AnkiIntegration(config as AnkiConnectConfig, {} as never, {} as never);
+  return (
+    integration as unknown as {
+      processSentenceFurigana: (sentence: string, fields: Record<string, string>) => string;
+    }
+  ).processSentenceFurigana(sentenceFurigana, noteFields);
+}
+
+test('AnkiIntegration highlights mined word from expression field when sentence has no bold marker', () => {
+  const processed = processSentenceWithConfig(
+    {
+      fields: {
+        word: 'Expression',
+        sentence: 'Sentence',
+      },
+      behavior: {
+        highlightWord: true,
+      },
+    },
+    '先日 貴様らが潜入した キールダンジョンから―',
+    {
+      expression: '潜入',
+      sentence: '先日 貴様らが潜入した キールダンジョンから―',
+    },
+  );
+
+  assert.equal(processed, '先日 貴様らが<b>潜入</b>した キールダンジョンから―');
+});
+
+test('AnkiIntegration keeps existing Yomitan bold target when present', () => {
+  const processed = processSentenceWithConfig(
+    {
+      fields: {
+        word: 'Expression',
+        sentence: 'Sentence',
+      },
+      behavior: {
+        highlightWord: true,
+      },
+    },
+    '先日 貴様らが潜入した キールダンジョンから―',
+    {
+      expression: '潜入',
+      sentence: '<b>潜入した</b>',
+    },
+  );
+
+  assert.equal(processed, '先日 貴様らが<b>潜入した</b> キールダンジョンから―');
+});
+
+test('AnkiIntegration leaves sentence plain when word highlighting is disabled', () => {
+  const processed = processSentenceWithConfig(
+    {
+      fields: {
+        word: 'Expression',
+        sentence: 'Sentence',
+      },
+      behavior: {
+        highlightWord: false,
+      },
+    },
+    '先日 貴様らが潜入した キールダンジョンから―',
+    {
+      expression: '潜入',
+      sentence: '<b>潜入</b>',
+    },
+  );
+
+  assert.equal(processed, '先日 貴様らが潜入した キールダンジョンから―');
+});
+
+test('AnkiIntegration highlights mined word in sentence furigana field', () => {
+  const processed = processSentenceFuriganaWithConfig(
+    {
+      fields: {
+        word: 'Expression',
+        sentence: 'Sentence',
+      },
+      behavior: {
+        highlightWord: true,
+      },
+    },
+    '<span class="term"><ruby>不思議<rt>ふしぎ</rt></ruby></span><span class="term">な</span><span class="term"><ruby>特技<rt>とくぎ</rt></ruby></span><span class="term">を</span>',
+    {
+      expression: '特技',
+      sentence: '不思議な特技を',
+    },
+  );
+
+  assert.equal(
+    processed,
+    '<span class="term"><ruby>不思議<rt>ふしぎ</rt></ruby></span><span class="term">な</span><b><span class="term"><ruby>特技<rt>とくぎ</rt></ruby></span></b><span class="term">を</span>',
+  );
+});
+
 test('AnkiIntegration does not allocate proxy server when proxy transport is disabled', () => {
   const integration = new AnkiIntegration(
     {
@@ -414,6 +535,460 @@ test('AnkiIntegration marks partial update notifications as failures in OSD mode
   assert.deepEqual(osdMessages, ['x Updated card: taberu (image failed)']);
 });
 
+test('AnkiIntegration applies ready YouTube cache media to every queued note id', async () => {
+  const osdMessages: string[] = [];
+  const updatedNotes: Array<{ noteId: number; fields: Record<string, string> }> = [];
+  const storedMedia: string[] = [];
+  const mediaInputs: string[] = [];
+
+  const integration = new AnkiIntegration(
+    {
+      fields: {
+        image: 'Picture',
+      },
+      media: {
+        imageFormat: 'jpg',
+      },
+      behavior: {
+        notificationType: 'osd',
+      },
+    },
+    {} as never,
+    {} as never,
+    (text) => {
+      osdMessages.push(text);
+    },
+  );
+
+  const internals = integration as unknown as {
+    client: {
+      notesInfo: (noteIds: number[]) => Promise<unknown[]>;
+      updateNoteFields: (noteId: number, fields: Record<string, string>) => Promise<void>;
+      storeMediaFile: (filename: string, data: Buffer) => Promise<void>;
+    };
+    mediaGenerator: {
+      generateAudio: (
+        path: MediaInput,
+        startTime: number,
+        endTime: number,
+        audioPadding?: number,
+        audioStreamIndex?: number,
+      ) => Promise<Buffer>;
+      generateScreenshot: (path: MediaInput) => Promise<Buffer>;
+    };
+    queuePendingYoutubeMediaUpdate: (job: {
+      sourceUrl: string;
+      noteId: number;
+      startTime: number;
+      endTime: number;
+      label: string | number;
+      audioStreamIndex?: number;
+      audioFieldName?: string;
+      imageFieldName?: string;
+      generateAudio: boolean;
+      generateImage: boolean;
+    }) => void;
+  };
+  internals.client = {
+    notesInfo: async (noteIds) =>
+      noteIds.map((noteId) => ({
+        noteId,
+        fields: {
+          SentenceAudio: { value: '' },
+          Picture: { value: '' },
+        },
+      })),
+    updateNoteFields: async (noteId, fields) => {
+      updatedNotes.push({ noteId, fields });
+    },
+    storeMediaFile: async (filename) => {
+      storedMedia.push(filename);
+    },
+  };
+  internals.mediaGenerator = {
+    generateAudio: async (mediaPath, _startTime, _endTime, _audioPadding, audioStreamIndex) => {
+      mediaInputs.push(
+        `audio:${describeMediaInputForTest(mediaPath)}:${audioStreamIndex ?? 'auto'}`,
+      );
+      return Buffer.from('audio');
+    },
+    generateScreenshot: async (mediaPath) => {
+      mediaInputs.push(`image:${describeMediaInputForTest(mediaPath)}`);
+      return Buffer.from('image');
+    },
+  };
+  internals.queuePendingYoutubeMediaUpdate({
+    sourceUrl: 'https://www.youtube.com/watch?v=abc123',
+    noteId: 101,
+    startTime: 10,
+    endTime: 12,
+    label: 'first',
+    audioStreamIndex: 22,
+    audioFieldName: 'SentenceAudio',
+    imageFieldName: 'Picture',
+    generateAudio: true,
+    generateImage: true,
+  });
+  internals.queuePendingYoutubeMediaUpdate({
+    sourceUrl: 'https://youtu.be/abc123',
+    noteId: 202,
+    startTime: 20,
+    endTime: 22,
+    label: 'second',
+    audioStreamIndex: 23,
+    audioFieldName: 'SentenceAudio',
+    imageFieldName: 'Picture',
+    generateAudio: true,
+    generateImage: true,
+  });
+
+  await integration.handleYoutubeMediaCacheReady('https://youtu.be/abc123', '/tmp/media.mkv');
+
+  assert.deepEqual(mediaInputs, [
+    'audio:/tmp/media.mkv:youtube-cache:auto',
+    'image:/tmp/media.mkv:youtube-cache',
+    'audio:/tmp/media.mkv:youtube-cache:auto',
+    'image:/tmp/media.mkv:youtube-cache',
+  ]);
+  assert.deepEqual(
+    updatedNotes.map((update) => update.noteId),
+    [101, 202],
+  );
+  const firstUpdate = updatedNotes[0];
+  const secondUpdate = updatedNotes[1];
+  assert.ok(firstUpdate);
+  assert.ok(secondUpdate);
+  assert.match(firstUpdate.fields.SentenceAudio ?? '', /^\[sound:audio_/);
+  assert.match(firstUpdate.fields.Picture ?? '', /^<img src="image_/);
+  assert.match(secondUpdate.fields.SentenceAudio ?? '', /^\[sound:audio_/);
+  assert.match(secondUpdate.fields.Picture ?? '', /^<img src="image_/);
+  assert.equal(storedMedia.length, 4);
+  assert.equal(
+    osdMessages.some((message) =>
+      message.includes('YouTube media cache ready. Adding media to 2 queued cards.'),
+    ),
+    true,
+  );
+});
+
+test('AnkiIntegration reports partial queued YouTube media updates separately from failures', async () => {
+  const osdMessages: string[] = [];
+  const updatedNotes: Array<{ noteId: number; fields: Record<string, string> }> = [];
+  const notifications: Array<{ noteId: number; label: string | number; suffix?: string }> = [];
+
+  const integration = new AnkiIntegration(
+    {
+      fields: {
+        image: 'Picture',
+      },
+      media: {
+        imageFormat: 'jpg',
+      },
+      behavior: {
+        notificationType: 'osd',
+      },
+    },
+    {} as never,
+    {} as never,
+    (text) => {
+      osdMessages.push(text);
+    },
+  );
+
+  const internals = integration as unknown as {
+    client: {
+      notesInfo: (noteIds: number[]) => Promise<unknown[]>;
+      updateNoteFields: (noteId: number, fields: Record<string, string>) => Promise<void>;
+      storeMediaFile: () => Promise<void>;
+    };
+    mediaGenerator: {
+      generateAudio: () => Promise<Buffer>;
+      generateScreenshot: () => Promise<Buffer>;
+    };
+    queuePendingYoutubeMediaUpdate: (job: {
+      sourceUrl: string;
+      noteId: number;
+      startTime: number;
+      endTime: number;
+      label: string | number;
+      audioFieldName?: string;
+      imageFieldName?: string;
+      generateAudio: boolean;
+      generateImage: boolean;
+    }) => void;
+    showNotification: (noteId: number, label: string | number, suffix?: string) => Promise<void>;
+  };
+  internals.client = {
+    notesInfo: async (noteIds) =>
+      noteIds.map((noteId) => ({
+        noteId,
+        fields: {
+          SentenceAudio: { value: '' },
+          Picture: { value: '' },
+        },
+      })),
+    updateNoteFields: async (noteId, fields) => {
+      updatedNotes.push({ noteId, fields });
+    },
+    storeMediaFile: async () => undefined,
+  };
+  internals.mediaGenerator = {
+    generateAudio: async () => {
+      throw new Error('audio stream not found');
+    },
+    generateScreenshot: async () => Buffer.from('image'),
+  };
+  internals.showNotification = async (noteId, label, suffix) => {
+    notifications.push({ noteId, label, suffix });
+  };
+
+  internals.queuePendingYoutubeMediaUpdate({
+    sourceUrl: 'https://www.youtube.com/watch?v=partial',
+    noteId: 303,
+    startTime: 10,
+    endTime: 12,
+    label: 'partial',
+    audioFieldName: 'SentenceAudio',
+    imageFieldName: 'Picture',
+    generateAudio: true,
+    generateImage: true,
+  });
+
+  await integration.handleYoutubeMediaCacheReady('https://youtu.be/partial', '/tmp/media.mkv');
+
+  assert.equal(updatedNotes.length, 1);
+  assert.match(updatedNotes[0]?.fields.Picture ?? '', /^<img src="image_/);
+  assert.equal(updatedNotes[0]?.fields.SentenceAudio, undefined);
+  assert.deepEqual(notifications, [{ noteId: 303, label: 'partial', suffix: 'audio failed' }]);
+  assert.equal(
+    osdMessages.some((message) =>
+      message.includes('Queued YouTube media finished with 0 updated, 1 partial, and 0 failed.'),
+    ),
+    true,
+  );
+});
+
+test('AnkiIntegration queues YouTube media updates against recovered source URLs', async () => {
+  const updatedNotes: Array<{ noteId: number; fields: Record<string, string> }> = [];
+  const storedMedia: string[] = [];
+
+  const integration = new AnkiIntegration(
+    {
+      fields: {
+        image: 'Picture',
+      },
+      media: {
+        imageFormat: 'jpg',
+      },
+    },
+    {} as never,
+    {
+      currentVideoPath: 'https://rr1---sn.example.googlevideo.com/videoplayback?expire=1777777777',
+      currentSubStart: 10,
+      currentSubEnd: 12,
+      currentTimePos: 11,
+    } as never,
+    () => undefined,
+    undefined,
+    undefined,
+    undefined,
+    {},
+    undefined,
+    undefined,
+    async () => null,
+    () => true,
+    () => 'https://www.youtube.com/watch?v=abc123',
+  );
+
+  const internals = integration as unknown as {
+    client: {
+      notesInfo: (noteIds: number[]) => Promise<unknown[]>;
+      updateNoteFields: (noteId: number, fields: Record<string, string>) => Promise<void>;
+      storeMediaFile: (filename: string) => Promise<void>;
+    };
+    mediaGenerator: {
+      generateAudio: () => Promise<Buffer>;
+      generateScreenshot: () => Promise<Buffer>;
+    };
+    queuePendingYoutubeMediaUpdateForNote: (job: {
+      noteId: number;
+      noteInfo: { noteId: number; fields: Record<string, { value: string }> };
+      label: string | number;
+    }) => Promise<boolean>;
+    showNotification: () => Promise<void>;
+  };
+  internals.client = {
+    notesInfo: async (noteIds) =>
+      noteIds.map((noteId) => ({
+        noteId,
+        fields: {
+          SentenceAudio: { value: '' },
+          Picture: { value: '' },
+        },
+      })),
+    updateNoteFields: async (noteId, fields) => {
+      updatedNotes.push({ noteId, fields });
+    },
+    storeMediaFile: async (filename) => {
+      storedMedia.push(filename);
+    },
+  };
+  internals.mediaGenerator = {
+    generateAudio: async () => Buffer.from('audio'),
+    generateScreenshot: async () => Buffer.from('image'),
+  };
+  internals.showNotification = async () => undefined;
+
+  const queued = await internals.queuePendingYoutubeMediaUpdateForNote({
+    noteId: 404,
+    noteInfo: {
+      noteId: 404,
+      fields: {
+        SentenceAudio: { value: '' },
+        Picture: { value: '' },
+      },
+    },
+    label: 'resolved source',
+  });
+  await integration.handleYoutubeMediaCacheReady('https://youtu.be/abc123', '/tmp/media.mkv');
+
+  assert.equal(queued, true);
+  assert.equal(updatedNotes.length, 1);
+  assert.equal(updatedNotes[0]?.noteId, 404);
+  assert.match(updatedNotes[0]?.fields.SentenceAudio ?? '', /^\[sound:audio_/);
+  assert.match(updatedNotes[0]?.fields.Picture ?? '', /^<img src="image_/);
+  assert.equal(storedMedia.length, 2);
+});
+
+test('AnkiIntegration passes audio normalization config for ready cached YouTube audio', async () => {
+  const audioCalls: Array<{
+    path: string;
+    audioStreamIndex?: number;
+    normalizeAudio?: boolean;
+  }> = [];
+
+  const integration = new AnkiIntegration(
+    {
+      media: {
+        audioPadding: 0,
+        normalizeAudio: false,
+      },
+    },
+    {} as never,
+    {
+      currentVideoPath: 'https://www.youtube.com/watch?v=abc123',
+      currentAudioStreamIndex: 0,
+      currentSubStart: 10,
+      currentSubEnd: 12,
+      currentTimePos: 11,
+    } as never,
+    () => undefined,
+    undefined,
+    undefined,
+    undefined,
+    {},
+    undefined,
+    undefined,
+    async () => '/tmp/subminer-youtube-media-cache/media.mkv',
+    () => true,
+  );
+
+  const internals = integration as unknown as {
+    mediaGenerator: {
+      generateAudio: (
+        path: { path: string },
+        startTime: number,
+        endTime: number,
+        audioPadding?: number,
+        audioStreamIndex?: number,
+        normalizeAudio?: boolean,
+      ) => Promise<Buffer>;
+    };
+    generateAudio: () => Promise<Buffer | null>;
+  };
+  internals.mediaGenerator = {
+    generateAudio: async (
+      path,
+      _startTime,
+      _endTime,
+      _audioPadding,
+      audioStreamIndex,
+      normalizeAudio,
+    ) => {
+      audioCalls.push({ path: path.path, audioStreamIndex, normalizeAudio });
+      return Buffer.from('audio');
+    },
+  };
+
+  await internals.generateAudio();
+
+  assert.deepEqual(audioCalls, [
+    {
+      path: '/tmp/subminer-youtube-media-cache/media.mkv',
+      audioStreamIndex: undefined,
+      normalizeAudio: false,
+    },
+  ]);
+});
+
+test('AnkiIntegration announces ready YouTube cache when no queued notes exist', async () => {
+  const overlayNotifications: TestOverlayNotificationPayload[] = [];
+
+  const integration = new AnkiIntegration(
+    {
+      behavior: {
+        notificationType: 'overlay',
+      },
+    },
+    {} as never,
+    {} as never,
+    undefined,
+    undefined,
+    undefined,
+    undefined,
+    {},
+    undefined,
+    (payload) => {
+      overlayNotifications.push(payload as TestOverlayNotificationPayload);
+    },
+  );
+
+  await integration.handleYoutubeMediaCacheReady('https://youtu.be/abc123', '/tmp/media.mkv');
+
+  assert.equal(overlayNotifications.length, 1);
+  assert.equal(overlayNotifications[0]?.title, 'SubMiner');
+  assert.equal(overlayNotifications[0]?.body, 'YouTube media cache ready.');
+});
+
+test('AnkiIntegration can let caller own no-queued YouTube cache ready notification', async () => {
+  const overlayNotifications: TestOverlayNotificationPayload[] = [];
+
+  const integration = new AnkiIntegration(
+    {
+      behavior: {
+        notificationType: 'overlay',
+      },
+    },
+    {} as never,
+    {} as never,
+    undefined,
+    undefined,
+    undefined,
+    undefined,
+    {},
+    undefined,
+    (payload) => {
+      overlayNotifications.push(payload as TestOverlayNotificationPayload);
+    },
+  );
+
+  await integration.handleYoutubeMediaCacheReady('https://youtu.be/abc123', '/tmp/media.mkv', {
+    notifyNoQueued: false,
+  });
+
+  assert.deepEqual(overlayNotifications, []);
+});
+
 test('AnkiIntegration embeds generated notification image on overlay mined-card notifications', async () => {
   const desktopNotifications: Array<{ title: string; body?: string; icon?: string }> = [];
   const overlayNotifications: TestOverlayNotificationPayload[] = [];

src/anki-integration.ts

@@ -61,7 +61,18 @@ import { NoteUpdateWorkflow } from './anki-integration/note-update-workflow';
 import { FieldGroupingWorkflow } from './anki-integration/field-grouping-workflow';
 import { resolveAnimatedImageLeadInSeconds } from './anki-integration/animated-image-sync';
 import { AnkiIntegrationRuntime, normalizeAnkiIntegrationConfig } from './anki-integration/runtime';
-import { resolveMediaGenerationInputPath } from './anki-integration/media-source';
+import {
+  resolveAudioStreamIndexForMediaGeneration,
+  resolveMediaGenerationInput,
+  resolveMediaGenerationInputPath,
+  type MediaGenerationInputResolverOptions,
+} from './anki-integration/media-source';
+import type { PendingYoutubeMediaUpdate } from './anki-integration/pending-youtube-media';
+import { PendingYoutubeMediaQueue } from './anki-integration/pending-youtube-media-queue';
+import type {
+  PendingYoutubeMediaQueueFailedOptions,
+  PendingYoutubeMediaQueueReadyOptions,
+} from './anki-integration/pending-youtube-media-queue';
 
 const log = createLogger('anki').child('integration');
 
@@ -70,7 +81,7 @@ interface NoteInfo {
   fields: Record<string, { value: string }>;
 }
 
-type CardKind = 'sentence' | 'audio';
+type CardKind = 'sentence' | 'audio' | 'word-and-sentence';
 
 function trimToNonEmptyString(value: unknown): string | null {
   if (typeof value !== 'string') return null;
@@ -78,6 +89,66 @@ function trimToNonEmptyString(value: unknown): string | null {
   return trimmed.length > 0 ? trimmed : null;
 }
 
+function stripRubyReadingText(value: string): string {
+  return value
+    .replace(/<rt\b[^>]*>[\s\S]*?<\/rt>/gi, '')
+    .replace(/<rp\b[^>]*>[\s\S]*?<\/rp>/gi, '');
+}
+
+function stripHtmlTags(value: string): string {
+  return value.replace(/<[^>]+>/g, '');
+}
+
+function getVisibleFuriganaText(value: string): string {
+  return stripHtmlTags(stripRubyReadingText(value));
+}
+
+function boldMatchingFuriganaTerms(sentenceFurigana: string, highlightedText: string): string {
+  if (!sentenceFurigana || !highlightedText || /<b\b/i.test(sentenceFurigana)) {
+    return sentenceFurigana;
+  }
+
+  const spanRegex = /<span\b[^>]*>[\s\S]*?<\/span>/gi;
+  const spans: Array<{ start: number; end: number; visibleStart: number; visibleEnd: number }> = [];
+  let visibleSentence = '';
+  let match: RegExpExecArray | null;
+  while ((match = spanRegex.exec(sentenceFurigana)) !== null) {
+    const visibleStart = visibleSentence.length;
+    visibleSentence += getVisibleFuriganaText(match[0] || '');
+    spans.push({
+      start: match.index,
+      end: match.index + match[0].length,
+      visibleStart,
+      visibleEnd: visibleSentence.length,
+    });
+  }
+
+  if (spans.length === 0) {
+    return sentenceFurigana.replace(highlightedText, `<b>${highlightedText}</b>`);
+  }
+
+  const highlightStart = visibleSentence.indexOf(highlightedText);
+  if (highlightStart === -1) {
+    return sentenceFurigana;
+  }
+  const highlightEnd = highlightStart + highlightedText.length;
+  const matchingSpans = spans.filter(
+    (span) => span.visibleEnd > highlightStart && span.visibleStart < highlightEnd,
+  );
+  if (matchingSpans.length === 0) {
+    return sentenceFurigana;
+  }
+
+  let result = sentenceFurigana;
+  for (const span of [...matchingSpans].reverse()) {
+    result = `${result.slice(0, span.start)}<b>${result.slice(
+      span.start,
+      span.end,
+    )}</b>${result.slice(span.end)}`;
+  }
+  return result;
+}
+
 function decodeURIComponentSafe(value: string): string {
   try {
     return decodeURIComponent(value);
@@ -165,6 +236,13 @@ export class AnkiIntegration {
   private consumeSubtitleMiningContextCallback: (() => SubtitleMiningContext | null) | null = null;
   private noteIdRedirects = new Map<number, number>();
   private trackedDuplicateNoteIds = new Map<number, number[]>();
+  private getCachedMediaPath: MediaGenerationInputResolverOptions['getCachedMediaPath'] | null =
+    null;
+  private shouldRequireRemoteMediaCache: (() => boolean) | null = null;
+  private getYoutubeMediaSourceUrl:
+    | (() => Promise<string | null | undefined> | string | null | undefined)
+    | null = null;
+  private pendingYoutubeMediaQueue: PendingYoutubeMediaQueue;
 
   constructor(
     config: AnkiConnectConfig,
@@ -180,6 +258,9 @@ export class AnkiIntegration {
     aiConfig: AiConfig = {},
     recordCardsMined?: (count: number, noteIds?: number[]) => void,
     overlayNotificationCallback?: (payload: OverlayNotificationPayload) => void,
+    getCachedMediaPath?: MediaGenerationInputResolverOptions['getCachedMediaPath'],
+    shouldRequireRemoteMediaCache?: () => boolean,
+    getYoutubeMediaSourceUrl?: () => Promise<string | null | undefined> | string | null | undefined,
   ) {
     this.config = normalizeAnkiIntegrationConfig(config);
     this.aiConfig = { ...aiConfig };
@@ -192,6 +273,10 @@ export class AnkiIntegration {
     this.overlayNotificationCallback = overlayNotificationCallback || null;
     this.fieldGroupingCallback = fieldGroupingCallback || null;
     this.recordCardsMinedCallback = recordCardsMined ?? null;
+    this.getCachedMediaPath = getCachedMediaPath ?? null;
+    this.shouldRequireRemoteMediaCache = shouldRequireRemoteMediaCache ?? null;
+    this.getYoutubeMediaSourceUrl = getYoutubeMediaSourceUrl ?? null;
+    this.pendingYoutubeMediaQueue = this.createPendingYoutubeMediaQueue();
     this.knownWordCache = this.createKnownWordCache(knownWordCacheStatePath);
     this.pollingRunner = this.createPollingRunner();
     this.cardCreationService = this.createCardCreationService();
@@ -235,6 +320,82 @@ export class AnkiIntegration {
     }
   }
 
+  private getMediaResolverOptions(): MediaGenerationInputResolverOptions {
+    const options: MediaGenerationInputResolverOptions = {
+      logDebug: (message) => log.debug(message),
+    };
+    if (this.getCachedMediaPath) {
+      options.getCachedMediaPath = this.getCachedMediaPath;
+    }
+    if (this.shouldRequireRemoteMediaCache?.()) {
+      options.remoteCacheMode = 'required';
+    }
+    return options;
+  }
+
+  private createPendingYoutubeMediaQueue(): PendingYoutubeMediaQueue {
+    return new PendingYoutubeMediaQueue({
+      client: {
+        notesInfo: async (noteIds) => (await this.client.notesInfo(noteIds)) as unknown,
+        updateNoteFields: (noteId, fields) => this.client.updateNoteFields(noteId, fields),
+        storeMediaFile: (filename, data) => this.client.storeMediaFile(filename, data),
+      },
+      mediaGenerator: {
+        generateAudio: (videoPath, startTime, endTime, audioPadding, audioStreamIndex) =>
+          this.mediaGenerator.generateAudio(
+            videoPath,
+            startTime,
+            endTime,
+            audioPadding,
+            audioStreamIndex,
+            this.config.media?.normalizeAudio !== false,
+          ),
+        generateScreenshot: (videoPath, timestamp, options) =>
+          this.mediaGenerator.generateScreenshot(videoPath, timestamp, options),
+        generateAnimatedImage: (videoPath, startTime, endTime, audioPadding, options) =>
+          this.mediaGenerator.generateAnimatedImage(
+            videoPath,
+            startTime,
+            endTime,
+            audioPadding,
+            options,
+          ),
+      },
+      getConfig: () => this.config,
+      getCurrentVideoPath: () => this.getCurrentYoutubeMediaSourceUrl(),
+      getCachedMediaPath: this.getCachedMediaPath,
+      shouldRequireRemoteMediaCache: () => this.shouldRequireRemoteMediaCache?.() === true,
+      getSubtitleMediaRange: (context) => this.getSubtitleMediaRange(context),
+      getResolvedSentenceAudioFieldName: (noteInfo) =>
+        this.getResolvedSentenceAudioFieldName(noteInfo),
+      resolveConfiguredFieldName: (noteInfo, ...preferredNames) =>
+        this.resolveConfiguredFieldName(noteInfo, ...preferredNames),
+      mergeFieldValue: (existing, newValue, overwrite) =>
+        this.mergeFieldValue(existing, newValue, overwrite),
+      getAnimatedImageLeadInSeconds: (noteInfo) => this.getAnimatedImageLeadInSeconds(noteInfo),
+      generateAudioFilename: () => this.generateAudioFilename(),
+      generateImageFilename: () => this.generateImageFilename(),
+      formatMiscInfoPatternForMediaPath: (
+        fallbackFilename,
+        startTimeSeconds,
+        mediaPath,
+        mediaTitle,
+      ) =>
+        this.formatMiscInfoPatternForMediaPath(
+          fallbackFilename,
+          startTimeSeconds,
+          mediaPath,
+          mediaTitle,
+        ),
+      showStatusNotification: (message) => this.showStatusNotification(message),
+      showNotification: (noteId, label, errorSuffix) =>
+        this.showNotification(noteId, label, errorSuffix),
+      logInfo: (...args) => log.info(args[0] as string, ...args.slice(1)),
+      logWarn: (...args) => log.warn(args[0] as string, ...args.slice(1)),
+      logError: (...args) => log.error(args[0] as string, ...args.slice(1)),
+    });
+  }
+
   private createKnownWordCache(knownWordCacheStatePath?: string): KnownWordCacheManager {
     return new KnownWordCacheManager({
       client: {
@@ -317,6 +478,10 @@ export class AnkiIntegration {
       getAiConfig: () => this.aiConfig,
       getTimingTracker: () => this.timingTracker,
       getMpvClient: () => this.mpvClient,
+      ...(this.getCachedMediaPath ? { getCachedMediaPath: this.getCachedMediaPath } : {}),
+      shouldRequireRemoteMediaCache: () => this.shouldRequireRemoteMediaCache?.() === true,
+      getYoutubeMediaSourceUrl: () => this.getCurrentYoutubeMediaSourceUrl(),
+      queuePendingYoutubeMediaUpdate: (job) => this.queuePendingYoutubeMediaUpdate(job),
       getDeck: () => this.config.deck,
       client: {
         addNote: (deck, modelName, fields, tags) =>
@@ -338,6 +503,7 @@ export class AnkiIntegration {
             endTime,
             audioPadding,
             audioStreamIndex,
+            this.config.media?.normalizeAudio !== false,
           ),
         generateScreenshot: (videoPath, timestamp, options) =>
           this.mediaGenerator.generateScreenshot(videoPath, timestamp, options),
@@ -461,6 +627,10 @@ export class AnkiIntegration {
       handleFieldGroupingManual: (originalNoteId, newNoteId, newNoteInfo, expression) =>
         this.handleFieldGroupingManual(originalNoteId, newNoteId, newNoteInfo, expression),
       processSentence: (mpvSentence, noteFields) => this.processSentence(mpvSentence, noteFields),
+      processSentenceFurigana: (sentenceFurigana, noteFields) =>
+        this.processSentenceFurigana(sentenceFurigana, noteFields),
+      setCardTypeFields: (updatedFields, availableFieldNames, cardKind) =>
+        this.setCardTypeFields(updatedFields, availableFieldNames, cardKind),
       resolveConfiguredFieldName: (noteInfo, ...preferredNames) =>
         this.resolveConfiguredFieldName(noteInfo, ...preferredNames),
       getResolvedSentenceAudioFieldName: (noteInfo) =>
@@ -476,6 +646,7 @@ export class AnkiIntegration {
       formatMiscInfoPattern: (fallbackFilename, startTimeSeconds) =>
         this.formatMiscInfoPattern(fallbackFilename, startTimeSeconds),
       consumeSubtitleMiningContext: () => this.consumeSubtitleMiningContext(),
+      queuePendingYoutubeMediaUpdate: (job) => this.queuePendingYoutubeMediaUpdateForNote(job),
       addConfiguredTagsToNote: (noteId) => this.addConfiguredTagsToNote(noteId),
       showNotification: (noteId, label) => this.showNotification(noteId, label),
       showOsdNotification: (message) => this.showStatusNotification(message),
@@ -677,20 +848,25 @@ export class AnkiIntegration {
     return result;
   }
 
+  private getSentenceHighlightText(noteFields: Record<string, string>): string {
+    const sentenceFieldName = this.config.fields?.sentence?.toLowerCase() || 'sentence';
+    const existingSentence = noteFields[sentenceFieldName] || '';
+    return (
+      existingSentence.match(/<b>(.*?)<\/b>/)?.[1] ||
+      getPreferredWordValueFromExtractedFields(noteFields, this.config).trim()
+    );
+  }
+
   private processSentence(mpvSentence: string, noteFields: Record<string, string>): string {
     if (this.config.behavior?.highlightWord === false) {
       return mpvSentence;
     }
 
-    const sentenceFieldName = this.config.fields?.sentence?.toLowerCase() || 'sentence';
-    const existingSentence = noteFields[sentenceFieldName] || '';
-
-    const highlightMatch = existingSentence.match(/<b>(.*?)<\/b>/);
-    if (!highlightMatch || !highlightMatch[1]) {
+    const highlightedText = this.getSentenceHighlightText(noteFields);
+    if (!highlightedText) {
       return mpvSentence;
     }
 
-    const highlightedText = highlightMatch[1];
     const index = mpvSentence.indexOf(highlightedText);
 
     if (index === -1) {
@@ -702,6 +878,20 @@ export class AnkiIntegration {
     return `${prefix}<b>${highlightedText}</b>${suffix}`;
   }
 
+  private processSentenceFurigana(
+    sentenceFurigana: string,
+    noteFields: Record<string, string>,
+  ): string {
+    if (this.config.behavior?.highlightWord === false) {
+      return sentenceFurigana;
+    }
+
+    const highlightedText = this.getSentenceHighlightText(noteFields);
+    return highlightedText
+      ? boldMatchingFuriganaTerms(sentenceFurigana, highlightedText)
+      : sentenceFurigana;
+  }
+
   private consumeSubtitleMiningContext(): SubtitleMiningContext | null {
     if (!this.consumeSubtitleMiningContextCallback) {
       return null;
@@ -750,13 +940,53 @@ export class AnkiIntegration {
     };
   }
 
+  private queuePendingYoutubeMediaUpdate(job: PendingYoutubeMediaUpdate): void {
+    this.pendingYoutubeMediaQueue.enqueue(job);
+  }
+
+  private async queuePendingYoutubeMediaUpdateForNote(job: {
+    noteId: number;
+    noteInfo: NoteInfo;
+    context?: SubtitleMiningContext;
+    label: string | number;
+  }): Promise<boolean> {
+    return this.pendingYoutubeMediaQueue.queueFromNote(job);
+  }
+
+  private async getCurrentYoutubeMediaSourceUrl(): Promise<string> {
+    return (
+      trimToNonEmptyString(await this.getYoutubeMediaSourceUrl?.()) ??
+      trimToNonEmptyString(this.mpvClient.currentVideoPath) ??
+      ''
+    );
+  }
+
+  async handleYoutubeMediaCacheReady(
+    sourceUrl: string,
+    cachedPath: string,
+    options?: PendingYoutubeMediaQueueReadyOptions,
+  ): Promise<void> {
+    await this.pendingYoutubeMediaQueue.handleReady(sourceUrl, cachedPath, options);
+  }
+
+  async handleYoutubeMediaCacheFailed(
+    sourceUrl: string,
+    options?: PendingYoutubeMediaQueueFailedOptions,
+  ): Promise<void> {
+    await this.pendingYoutubeMediaQueue.handleFailed(sourceUrl, options);
+  }
+
   private async generateAudio(context?: SubtitleMiningContext): Promise<Buffer | null> {
     const mpvClient = this.mpvClient;
     if (!mpvClient || !mpvClient.currentVideoPath) {
       return null;
     }
 
-    const videoPath = await resolveMediaGenerationInputPath(mpvClient, 'audio');
+    const videoPath = await resolveMediaGenerationInput(
+      mpvClient,
+      'audio',
+      this.getMediaResolverOptions(),
+    );
     if (!videoPath) {
       return null;
     }
@@ -767,7 +997,8 @@ export class AnkiIntegration {
       startTime,
       endTime,
       this.config.media?.audioPadding,
-      this.mpvClient.currentAudioStreamIndex,
+      resolveAudioStreamIndexForMediaGeneration(videoPath, this.mpvClient.currentAudioStreamIndex),
+      this.config.media?.normalizeAudio !== false,
     );
   }
 
@@ -779,7 +1010,11 @@ export class AnkiIntegration {
       return null;
     }
 
-    const videoPath = await resolveMediaGenerationInputPath(this.mpvClient, 'video');
+    const videoPath = await resolveMediaGenerationInput(
+      this.mpvClient,
+      'video',
+      this.getMediaResolverOptions(),
+    );
     if (!videoPath) {
       return null;
     }
@@ -813,17 +1048,30 @@ export class AnkiIntegration {
   }
 
   private formatMiscInfoPattern(fallbackFilename: string, startTimeSeconds?: number): string {
+    return this.formatMiscInfoPatternForMediaPath(
+      fallbackFilename,
+      startTimeSeconds,
+      this.mpvClient.currentVideoPath || '',
+      this.mpvClient.currentMediaTitle ?? undefined,
+    );
+  }
+
+  private formatMiscInfoPatternForMediaPath(
+    fallbackFilename: string,
+    startTimeSeconds: number | undefined,
+    mediaPath: string,
+    mediaTitle?: string,
+  ): string {
     if (!this.config.metadata?.pattern) {
       return '';
     }
 
-    const currentVideoPath = this.mpvClient.currentVideoPath || '';
-    const videoFilename = extractFilenameFromMediaPath(currentVideoPath);
-    const mediaTitle = trimToNonEmptyString(this.mpvClient.currentMediaTitle);
+    const videoFilename = extractFilenameFromMediaPath(mediaPath);
+    const resolvedMediaTitle = trimToNonEmptyString(mediaTitle);
     const filenameWithExt =
-      (shouldPreferMediaTitleForMiscInfo(currentVideoPath, videoFilename)
-        ? mediaTitle || videoFilename
-        : videoFilename || mediaTitle) || fallbackFilename;
+      (shouldPreferMediaTitleForMiscInfo(mediaPath, videoFilename)
+        ? resolvedMediaTitle || videoFilename
+        : videoFilename || resolvedMediaTitle) || fallbackFilename;
     const filenameWithoutExt = filenameWithExt.replace(/\.[^.]+$/, '');
 
     const currentTimePos =
@@ -1030,6 +1278,30 @@ export class AnkiIntegration {
   ): void {
     const audioFlagNames = ['IsAudioCard'];
 
+    if (cardKind === 'word-and-sentence') {
+      const wordAndSentenceFlag = this.resolveFieldName(
+        availableFieldNames,
+        'IsWordAndSentenceCard',
+      );
+      if (!wordAndSentenceFlag) {
+        return;
+      }
+      updatedFields[wordAndSentenceFlag] = 'x';
+
+      const sentenceFlag = this.resolveFieldName(availableFieldNames, 'IsSentenceCard');
+      if (sentenceFlag && sentenceFlag !== wordAndSentenceFlag) {
+        updatedFields[sentenceFlag] = '';
+      }
+
+      for (const audioFlagName of audioFlagNames) {
+        const resolved = this.resolveFieldName(availableFieldNames, audioFlagName);
+        if (resolved && resolved !== wordAndSentenceFlag) {
+          updatedFields[resolved] = '';
+        }
+      }
+      return;
+    }
+
     if (cardKind === 'sentence') {
       const sentenceFlag = this.resolveFieldName(availableFieldNames, 'IsSentenceCard');
       if (sentenceFlag) {
@@ -1350,7 +1622,7 @@ export class AnkiIntegration {
       miscInfoValue?: string;
     } = {};
 
-    if (this.config.media?.generateAudio && this.mpvClient?.currentVideoPath) {
+    if (this.config.media?.generateAudio !== false && this.mpvClient?.currentVideoPath) {
       try {
         const audioFilename = this.generateAudioFilename();
         const audioBuffer = await this.generateAudio();
@@ -1370,7 +1642,7 @@ export class AnkiIntegration {
       }
     }
 
-    if (this.config.media?.generateImage && this.mpvClient?.currentVideoPath) {
+    if (this.config.media?.generateImage !== false && this.mpvClient?.currentVideoPath) {
       try {
         const animatedLeadInSeconds = noteInfo
           ? await this.getAnimatedImageLeadInSeconds(noteInfo)

src/anki-integration/card-creation-manual-update.test.ts

@@ -2,10 +2,33 @@ import assert from 'node:assert/strict';
 import test from 'node:test';
 
 import { CardCreationService } from './card-creation';
+import { toMpvEdlValue } from './mpv-edl-test-utils';
+import type { MediaInput } from '../media-generator';
 import type { AnkiConnectConfig } from '../types/anki';
 
 type CardCreationDeps = ConstructorParameters<typeof CardCreationService>[0];
 
+function setWordAndSentenceCardTypeFields(
+  updatedFields: Record<string, string>,
+  availableFieldNames: string[],
+  cardKind: 'sentence' | 'audio' | 'word-and-sentence',
+): void {
+  if (cardKind !== 'word-and-sentence') return;
+
+  const resolveFieldName = (preferredName: string): string | null =>
+    availableFieldNames.find((name) => name.toLowerCase() === preferredName.toLowerCase()) ?? null;
+  const wordAndSentenceFlag = resolveFieldName('IsWordAndSentenceCard');
+  if (!wordAndSentenceFlag) return;
+
+  updatedFields[wordAndSentenceFlag] = 'x';
+  for (const flagName of ['IsSentenceCard', 'IsAudioCard']) {
+    const resolved = resolveFieldName(flagName);
+    if (resolved && resolved !== wordAndSentenceFlag) {
+      updatedFields[resolved] = '';
+    }
+  }
+}
+
 function createManualUpdateService(overrides: Partial<CardCreationDeps> = {}): {
   service: CardCreationService;
   updatedFields: Record<string, string>[];
@@ -142,6 +165,72 @@ test('manual clipboard subtitle update replaces sentence audio without touching
   );
 });
 
+test('manual clipboard subtitle update marks Kiku word cards as word-and-sentence cards when enabled', async () => {
+  const { service, updatedFields } = createManualUpdateService({
+    getConfig: () =>
+      ({
+        deck: 'Mining',
+        fields: {
+          word: 'Expression',
+          sentence: 'Sentence',
+          audio: 'ExpressionAudio',
+        },
+        media: {
+          generateAudio: false,
+          generateImage: false,
+          maxMediaDuration: 30,
+        },
+        behavior: {
+          overwriteAudio: false,
+          overwriteImage: false,
+        },
+        ai: false,
+      }) as AnkiConnectConfig,
+    client: {
+      addNote: async () => 0,
+      addTags: async () => undefined,
+      notesInfo: async () => [
+        {
+          noteId: 42,
+          fields: {
+            Expression: { value: '単語' },
+            Sentence: { value: '' },
+            IsWordAndSentenceCard: { value: '' },
+            IsSentenceCard: { value: '' },
+            IsAudioCard: { value: '' },
+          },
+        },
+      ],
+      updateNoteFields: async (_noteId, fields) => {
+        updatedFields.push(fields);
+      },
+      storeMediaFile: async () => undefined,
+      findNotes: async () => [42],
+      retrieveMediaFile: async () => '',
+    },
+    getEffectiveSentenceCardConfig: () => ({
+      model: 'Sentence',
+      sentenceField: 'Sentence',
+      audioField: 'SentenceAudio',
+      lapisEnabled: false,
+      kikuEnabled: true,
+      kikuFieldGrouping: 'disabled',
+      kikuDeleteDuplicateInAuto: false,
+    }),
+    setCardTypeFields: setWordAndSentenceCardTypeFields,
+  });
+
+  await service.updateLastAddedFromClipboard('字幕');
+
+  assert.equal(updatedFields.length, 1);
+  assert.deepEqual(updatedFields[0], {
+    Sentence: '字幕',
+    IsWordAndSentenceCard: 'x',
+    IsSentenceCard: '',
+    IsAudioCard: '',
+  });
+});
+
 test('manual clipboard subtitle update skips audio when sentence audio field is missing', async () => {
   const { service, updatedFields, mergeCalls, storedMedia } = createManualUpdateService({
     client: {
@@ -179,9 +268,13 @@ test('manual clipboard subtitle update skips audio when sentence audio field is
 test('manual clipboard subtitle update uses resolved mpv stream URLs for remote media', async () => {
   const audioPaths: string[] = [];
   const imagePaths: string[] = [];
+  const recordMediaPath = (mediaInput: MediaInput): string =>
+    typeof mediaInput === 'string' ? mediaInput : mediaInput.path;
+  const audioUrl = 'https://audio.example/videoplayback?mime=audio%2Fwebm';
+  const videoUrl = 'https://video.example/videoplayback?mime=video%2Fmp4';
   const edlSource = [
-    'edl://!new_stream;!no_clip;!no_chapters;%70%https://audio.example/videoplayback?mime=audio%2Fwebm',
-    '!new_stream;!no_clip;!no_chapters;%69%https://video.example/videoplayback?mime=video%2Fmp4',
+    `edl://!new_stream;!no_clip;!no_chapters;${toMpvEdlValue(audioUrl)}`,
+    `!new_stream;!no_clip;!no_chapters;${toMpvEdlValue(videoUrl)}`,
     '!global_tags,title=test',
   ].join(';');
 
@@ -251,11 +344,11 @@ test('manual clipboard subtitle update uses resolved mpv stream URLs for remote
     },
     mediaGenerator: {
       generateAudio: async (path) => {
-        audioPaths.push(path);
+        audioPaths.push(recordMediaPath(path));
         return Buffer.from('audio');
       },
       generateScreenshot: async (path) => {
-        imagePaths.push(path);
+        imagePaths.push(recordMediaPath(path));
         return Buffer.from('image');
       },
       generateAnimatedImage: async () => null,
@@ -264,8 +357,8 @@ test('manual clipboard subtitle update uses resolved mpv stream URLs for remote
 
   await service.updateLastAddedFromClipboard('一行目\n\n二行目');
 
-  assert.deepEqual(audioPaths, ['https://audio.example/videoplayback?mime=audio%2Fwebm']);
-  assert.deepEqual(imagePaths, ['https://video.example/videoplayback?mime=video%2Fmp4']);
+  assert.deepEqual(audioPaths, [audioUrl]);
+  assert.deepEqual(imagePaths, [videoUrl]);
   assert.equal(storedMedia.length, 2);
   assert.equal(updatedFields.length, 1);
   assert.equal(updatedFields[0]?.Sentence, '一行目 二行目');

src/anki-integration/card-creation.test.ts

@@ -2,6 +2,8 @@ import assert from 'node:assert/strict';
 import test from 'node:test';
 
 import { CardCreationService } from './card-creation';
+import { toMpvEdlValue } from './mpv-edl-test-utils';
+import type { MediaInput } from '../media-generator';
 import type { AnkiConnectConfig } from '../types/anki';
 
 test('CardCreationService counts locally created sentence cards', async () => {
@@ -287,9 +289,13 @@ test('CardCreationService keeps updating after recordCardsMinedCallback throws',
 test('CardCreationService uses stream-open-filename for remote media generation', async () => {
   const audioPaths: string[] = [];
   const imagePaths: string[] = [];
+  const recordMediaPath = (mediaInput: MediaInput): string =>
+    typeof mediaInput === 'string' ? mediaInput : mediaInput.path;
+  const audioUrl = 'https://audio.example/videoplayback?mime=audio%2Fwebm';
+  const videoUrl = 'https://video.example/videoplayback?mime=video%2Fmp4';
   const edlSource = [
-    'edl://!new_stream;!no_clip;!no_chapters;%70%https://audio.example/videoplayback?mime=audio%2Fwebm',
-    '!new_stream;!no_clip;!no_chapters;%69%https://video.example/videoplayback?mime=video%2Fmp4',
+    `edl://!new_stream;!no_clip;!no_chapters;${toMpvEdlValue(audioUrl)}`,
+    `!new_stream;!no_clip;!no_chapters;${toMpvEdlValue(videoUrl)}`,
     '!global_tags,title=test',
   ].join(';');
 
@@ -345,11 +351,11 @@ test('CardCreationService uses stream-open-filename for remote media generation'
     },
     mediaGenerator: {
       generateAudio: async (path) => {
-        audioPaths.push(path);
+        audioPaths.push(recordMediaPath(path));
         return Buffer.from('audio');
       },
       generateScreenshot: async (path) => {
-        imagePaths.push(path);
+        imagePaths.push(recordMediaPath(path));
         return Buffer.from('image');
       },
       generateAnimatedImage: async () => null,
@@ -394,8 +400,261 @@ test('CardCreationService uses stream-open-filename for remote media generation'
   const created = await service.createSentenceCard('テスト', 0, 1);
 
   assert.equal(created, true);
-  assert.deepEqual(audioPaths, ['https://audio.example/videoplayback?mime=audio%2Fwebm']);
-  assert.deepEqual(imagePaths, ['https://video.example/videoplayback?mime=video%2Fmp4']);
+  assert.deepEqual(audioPaths, [audioUrl]);
+  assert.deepEqual(imagePaths, [videoUrl]);
+});
+
+test('CardCreationService does not use mpv stream indexes for ready cached YouTube media', async () => {
+  const audioCalls: Array<{ path: string; audioStreamIndex?: number }> = [];
+
+  const service = new CardCreationService({
+    getConfig: () =>
+      ({
+        deck: 'Mining',
+        fields: {
+          sentence: 'Sentence',
+          audio: 'SentenceAudio',
+          image: 'Picture',
+        },
+        media: {
+          generateAudio: true,
+          generateImage: false,
+          imageFormat: 'jpg',
+        },
+        behavior: {},
+        ai: false,
+      }) as AnkiConnectConfig,
+    getAiConfig: () => ({}),
+    getTimingTracker: () => ({}) as never,
+    getMpvClient: () =>
+      ({
+        currentVideoPath: 'https://www.youtube.com/watch?v=abc123',
+        currentSubText: '字幕',
+        currentSubStart: 10,
+        currentSubEnd: 12,
+        currentTimePos: 11,
+        currentAudioStreamIndex: 0,
+      }) as never,
+    getCachedMediaPath: async () => '/tmp/subminer-youtube-media-cache/media.mkv',
+    shouldRequireRemoteMediaCache: () => true,
+    client: {
+      addNote: async () => 42,
+      addTags: async () => undefined,
+      notesInfo: async () => [
+        {
+          noteId: 42,
+          fields: {
+            Sentence: { value: '' },
+            SentenceAudio: { value: '' },
+            Picture: { value: '' },
+          },
+        },
+      ],
+      updateNoteFields: async () => undefined,
+      storeMediaFile: async () => undefined,
+      findNotes: async () => [],
+      retrieveMediaFile: async () => '',
+    },
+    mediaGenerator: {
+      generateAudio: async (path, _startTime, _endTime, _padding, audioStreamIndex) => {
+        audioCalls.push({ path: typeof path === 'string' ? path : path.path, audioStreamIndex });
+        return Buffer.from('audio');
+      },
+      generateScreenshot: async () => null,
+      generateAnimatedImage: async () => null,
+    },
+    showOsdNotification: () => undefined,
+    showUpdateResult: () => undefined,
+    showStatusNotification: () => undefined,
+    showNotification: async () => undefined,
+    beginUpdateProgress: () => undefined,
+    endUpdateProgress: () => undefined,
+    withUpdateProgress: async (_message, action) => action(),
+    resolveConfiguredFieldName: (noteInfo, preferredName) => {
+      if (!preferredName) return null;
+      return Object.keys(noteInfo.fields).find((field) => field === preferredName) ?? null;
+    },
+    resolveNoteFieldName: (noteInfo, preferredName) => {
+      if (!preferredName) return null;
+      return Object.keys(noteInfo.fields).find((field) => field === preferredName) ?? null;
+    },
+    getAnimatedImageLeadInSeconds: async () => 0,
+    extractFields: () => ({}),
+    processSentence: (sentence) => sentence,
+    setCardTypeFields: () => undefined,
+    mergeFieldValue: (_existing, newValue) => newValue,
+    formatMiscInfoPattern: () => '',
+    getEffectiveSentenceCardConfig: () => ({
+      model: 'Sentence',
+      sentenceField: 'Sentence',
+      audioField: 'SentenceAudio',
+      lapisEnabled: false,
+      kikuEnabled: false,
+      kikuFieldGrouping: 'disabled',
+      kikuDeleteDuplicateInAuto: false,
+    }),
+    getFallbackDurationSeconds: () => 10,
+    appendKnownWordsFromNoteInfo: () => undefined,
+    isUpdateInProgress: () => false,
+    setUpdateInProgress: () => undefined,
+    trackLastAddedNoteId: () => undefined,
+  });
+
+  const created = await service.createSentenceCard('テスト', 10, 12);
+
+  assert.equal(created, true);
+  assert.deepEqual(audioCalls, [
+    {
+      path: '/tmp/subminer-youtube-media-cache/media.mkv',
+      audioStreamIndex: undefined,
+    },
+  ]);
+});
+
+test('CardCreationService queues YouTube media when required cache is not ready', async () => {
+  const mediaCalls: string[] = [];
+  const updates: Array<{ noteId: number; fields: Record<string, string> }> = [];
+  const queuedUpdates: Array<{
+    sourceUrl: string;
+    noteId: number;
+    startTime: number;
+    endTime: number;
+    label: string | number;
+    audioFieldName?: string;
+    imageFieldName?: string;
+    miscInfoFieldName?: string;
+    generateAudio: boolean;
+    generateImage: boolean;
+  }> = [];
+  let streamRequests = 0;
+
+  const service = new CardCreationService({
+    getConfig: () =>
+      ({
+        deck: 'Mining',
+        fields: {
+          sentence: 'Sentence',
+          audio: 'SentenceAudio',
+          image: 'Picture',
+          miscInfo: 'MiscInfo',
+        },
+        media: {
+          generateAudio: true,
+          generateImage: true,
+          imageFormat: 'jpg',
+        },
+        behavior: {},
+        ai: false,
+      }) as AnkiConnectConfig,
+    getAiConfig: () => ({}),
+    getTimingTracker: () => ({}) as never,
+    getMpvClient: () =>
+      ({
+        currentVideoPath: 'https://www.youtube.com/watch?v=abc123',
+        currentSubText: '字幕',
+        currentSubStart: 10,
+        currentSubEnd: 12,
+        currentTimePos: 11,
+        currentAudioStreamIndex: 2,
+        requestProperty: async () => {
+          streamRequests += 1;
+          return 'https://rr1---sn.example.googlevideo.com/videoplayback?id=123';
+        },
+      }) as never,
+    getCachedMediaPath: async () => null,
+    shouldRequireRemoteMediaCache: () => true,
+    queuePendingYoutubeMediaUpdate: (job) => {
+      queuedUpdates.push(job);
+    },
+    client: {
+      addNote: async () => 42,
+      addTags: async () => undefined,
+      notesInfo: async () => [
+        {
+          noteId: 42,
+          fields: {
+            Sentence: { value: '' },
+            SentenceAudio: { value: '' },
+            Picture: { value: '' },
+            MiscInfo: { value: '' },
+          },
+        },
+      ],
+      updateNoteFields: async (noteId, fields) => {
+        updates.push({ noteId, fields });
+      },
+      storeMediaFile: async () => undefined,
+      findNotes: async () => [],
+      retrieveMediaFile: async () => '',
+    },
+    mediaGenerator: {
+      generateAudio: async () => {
+        mediaCalls.push('audio');
+        return Buffer.from('audio');
+      },
+      generateScreenshot: async () => {
+        mediaCalls.push('image');
+        return Buffer.from('image');
+      },
+      generateAnimatedImage: async () => null,
+    },
+    showOsdNotification: () => undefined,
+    showUpdateResult: () => undefined,
+    showStatusNotification: () => undefined,
+    showNotification: async () => undefined,
+    beginUpdateProgress: () => undefined,
+    endUpdateProgress: () => undefined,
+    withUpdateProgress: async (_message, action) => action(),
+    resolveConfiguredFieldName: (noteInfo, preferredName) => {
+      if (!preferredName) return null;
+      return Object.keys(noteInfo.fields).find((field) => field === preferredName) ?? null;
+    },
+    resolveNoteFieldName: (noteInfo, preferredName) => {
+      if (!preferredName) return null;
+      return Object.keys(noteInfo.fields).find((field) => field === preferredName) ?? null;
+    },
+    getAnimatedImageLeadInSeconds: async () => 0,
+    extractFields: () => ({}),
+    processSentence: (sentence) => sentence,
+    setCardTypeFields: () => undefined,
+    mergeFieldValue: (_existing, newValue) => newValue,
+    formatMiscInfoPattern: () => '',
+    getEffectiveSentenceCardConfig: () => ({
+      model: 'Sentence',
+      sentenceField: 'Sentence',
+      audioField: 'SentenceAudio',
+      lapisEnabled: false,
+      kikuEnabled: false,
+      kikuFieldGrouping: 'disabled',
+      kikuDeleteDuplicateInAuto: false,
+    }),
+    getFallbackDurationSeconds: () => 10,
+    appendKnownWordsFromNoteInfo: () => undefined,
+    isUpdateInProgress: () => false,
+    setUpdateInProgress: () => undefined,
+    trackLastAddedNoteId: () => undefined,
+  });
+
+  const created = await service.createSentenceCard('テスト', 10, 12);
+
+  assert.equal(created, true);
+  assert.equal(streamRequests, 0);
+  assert.deepEqual(mediaCalls, []);
+  assert.deepEqual(queuedUpdates, [
+    {
+      sourceUrl: 'https://www.youtube.com/watch?v=abc123',
+      noteId: 42,
+      startTime: 10,
+      endTime: 12,
+      label: 'テスト',
+      audioFieldName: 'SentenceAudio',
+      imageFieldName: 'Picture',
+      miscInfoFieldName: 'MiscInfo',
+      generateAudio: true,
+      generateImage: true,
+    },
+  ]);
+  assert.deepEqual(updates, []);
 });
 
 test('CardCreationService tracks pre-add duplicate note ids for kiku sentence cards', async () => {

src/anki-integration/card-creation.ts

@@ -5,20 +5,43 @@ import {
 } from '../anki-field-config';
 import { AnkiConnectConfig } from '../types/anki';
 import { createLogger } from '../logger';
+import type { MediaInput } from '../media-input';
 import { SubtitleTimingTracker } from '../subtitle-timing-tracker';
 import { AiConfig } from '../types/integrations';
 import { MpvClient } from '../types/runtime';
 import { resolveSentenceBackText } from './ai';
-import { resolveMediaGenerationInputPath } from './media-source';
+import {
+  resolveMediaGenerationInput,
+  resolveAudioStreamIndexForMediaGeneration,
+  type MediaGenerationInputResolverOptions,
+} from './media-source';
+import { shouldMarkWordAndSentenceCard } from './note-field-utils';
+import type { PendingYoutubeMediaUpdate } from './pending-youtube-media';
 
 const log = createLogger('anki').child('integration.card-creation');
 
+function shouldGenerateAudio(config: AnkiConnectConfig): boolean {
+  return config.media?.generateAudio !== false;
+}
+
+function shouldGenerateImage(config: AnkiConnectConfig): boolean {
+  return config.media?.generateImage !== false;
+}
+
+function trimToNonEmptyString(value: unknown): string | null {
+  if (typeof value !== 'string') {
+    return null;
+  }
+  const trimmed = value.trim();
+  return trimmed.length > 0 ? trimmed : null;
+}
+
 export interface CardCreationNoteInfo {
   noteId: number;
   fields: Record<string, { value: string }>;
 }
 
-type CardKind = 'sentence' | 'audio';
+type CardKind = 'sentence' | 'audio' | 'word-and-sentence';
 
 interface CardCreationClient {
   addNote(
@@ -37,14 +60,15 @@ interface CardCreationClient {
 
 interface CardCreationMediaGenerator {
   generateAudio(
-    path: string,
+    path: MediaInput,
     startTime: number,
     endTime: number,
     audioPadding?: number,
     audioStreamIndex?: number,
+    normalizeAudio?: boolean,
   ): Promise<Buffer | null>;
   generateScreenshot(
-    path: string,
+    path: MediaInput,
     timestamp: number,
     options: {
       format: 'jpg' | 'png' | 'webp';
@@ -54,7 +78,7 @@ interface CardCreationMediaGenerator {
     },
   ): Promise<Buffer | null>;
   generateAnimatedImage(
-    path: string,
+    path: MediaInput,
     startTime: number,
     endTime: number,
     audioPadding?: number,
@@ -73,6 +97,10 @@ interface CardCreationDeps {
   getAiConfig: () => AiConfig;
   getTimingTracker: () => SubtitleTimingTracker;
   getMpvClient: () => MpvClient;
+  getCachedMediaPath?: MediaGenerationInputResolverOptions['getCachedMediaPath'];
+  shouldRequireRemoteMediaCache?: () => boolean;
+  getYoutubeMediaSourceUrl?: () => Promise<string | null | undefined> | string | null | undefined;
+  queuePendingYoutubeMediaUpdate?: (job: PendingYoutubeMediaUpdate) => void;
   getDeck?: () => string | undefined;
   client: CardCreationClient;
   mediaGenerator: CardCreationMediaGenerator;
@@ -120,6 +148,19 @@ interface CardCreationDeps {
 export class CardCreationService {
   constructor(private readonly deps: CardCreationDeps) {}
 
+  private getMediaResolverOptions(): MediaGenerationInputResolverOptions {
+    const options: MediaGenerationInputResolverOptions = {
+      logDebug: (message) => log.debug(message),
+    };
+    if (this.deps.getCachedMediaPath) {
+      options.getCachedMediaPath = this.deps.getCachedMediaPath;
+    }
+    if (this.deps.shouldRequireRemoteMediaCache?.()) {
+      options.remoteCacheMode = 'required';
+    }
+    return options;
+  }
+
   private getConfiguredAnkiTags(): string[] {
     const tags = this.deps.getConfig().tags;
     if (!Array.isArray(tags)) {
@@ -219,7 +260,8 @@ export class CardCreationService {
           this.deps.getConfig(),
         );
         const sentenceAudioField = this.getResolvedSentenceOnlyAudioFieldName(noteInfo);
-        const sentenceField = this.deps.getEffectiveSentenceCardConfig().sentenceField;
+        const sentenceCardConfig = this.deps.getEffectiveSentenceCardConfig();
+        const sentenceField = sentenceCardConfig.sentenceField;
 
         const sentence = blocks.join(' ');
         const updatedFields: Record<string, string> = {};
@@ -230,6 +272,13 @@ export class CardCreationService {
         if (sentenceField) {
           const processedSentence = this.deps.processSentence(sentence, fields);
           updatedFields[sentenceField] = processedSentence;
+          if (shouldMarkWordAndSentenceCard(noteInfo, sentenceCardConfig)) {
+            this.deps.setCardTypeFields(
+              updatedFields,
+              Object.keys(noteInfo.fields),
+              'word-and-sentence',
+            );
+          }
           updatePerformed = true;
         }
 
@@ -237,14 +286,18 @@ export class CardCreationService {
           `Clipboard update: timing range ${rangeStart.toFixed(2)}s - ${rangeEnd.toFixed(2)}s`,
         );
 
-        const audioSourcePath = this.deps.getConfig().media?.generateAudio
-          ? await resolveMediaGenerationInputPath(mpvClient, 'audio')
+        const config = this.deps.getConfig();
+        const generateAudio = shouldGenerateAudio(config);
+        const generateImage = shouldGenerateImage(config);
+        const mediaResolverOptions = this.getMediaResolverOptions();
+        const audioSourcePath = generateAudio
+          ? await resolveMediaGenerationInput(mpvClient, 'audio', mediaResolverOptions)
           : null;
-        const videoPath = this.deps.getConfig().media?.generateImage
-          ? await resolveMediaGenerationInputPath(mpvClient, 'video')
+        const videoPath = generateImage
+          ? await resolveMediaGenerationInput(mpvClient, 'video', mediaResolverOptions)
           : null;
 
-        if (this.deps.getConfig().media?.generateAudio) {
+        if (generateAudio) {
           try {
             const audioFilename = this.generateAudioFilename();
             const audioBuffer = audioSourcePath
@@ -272,7 +325,7 @@ export class CardCreationService {
           }
         }
 
-        if (this.deps.getConfig().media?.generateImage) {
+        if (generateImage) {
           try {
             const animatedLeadInSeconds = await this.deps.getAnimatedImageLeadInSeconds(noteInfo);
             const imageFilename = this.generateImageFilename();
@@ -432,7 +485,7 @@ export class CardCreationService {
           errors.push('audio');
         }
 
-        if (this.deps.getConfig().media?.generateImage) {
+        if (shouldGenerateImage(this.deps.getConfig())) {
           try {
             const animatedLeadInSeconds = await this.deps.getAnimatedImageLeadInSeconds(noteInfo);
             const imageFilename = this.generateImageFilename();
@@ -513,9 +566,23 @@ export class CardCreationService {
 
     try {
       return await this.deps.withUpdateProgress('Creating sentence card', async () => {
-        const videoPath = await resolveMediaGenerationInputPath(mpvClient, 'video');
-        const audioSourcePath = await resolveMediaGenerationInputPath(mpvClient, 'audio');
-        if (!videoPath) {
+        const config = this.deps.getConfig();
+        const generateAudio = shouldGenerateAudio(config);
+        const generateImage = shouldGenerateImage(config);
+        const mediaResolverOptions = this.getMediaResolverOptions();
+        const videoPath = generateImage
+          ? await resolveMediaGenerationInput(mpvClient, 'video', mediaResolverOptions)
+          : null;
+        const audioSourcePath = generateAudio
+          ? await resolveMediaGenerationInput(mpvClient, 'audio', mediaResolverOptions)
+          : null;
+        const missingRequestedMediaInput =
+          (generateImage && !videoPath) || (generateAudio && !audioSourcePath);
+        const shouldQueuePendingYoutubeMedia =
+          missingRequestedMediaInput &&
+          this.deps.shouldRequireRemoteMediaCache?.() === true &&
+          typeof this.deps.queuePendingYoutubeMediaUpdate === 'function';
+        if (missingRequestedMediaInput && !shouldQueuePendingYoutubeMedia) {
           this.deps.showOsdNotification('No video loaded');
           return false;
         }
@@ -525,13 +592,13 @@ export class CardCreationService {
 
         const sentenceField = sentenceCardConfig.sentenceField;
         const audioFieldName = sentenceCardConfig.audioField || 'SentenceAudio';
-        const translationField = this.deps.getConfig().fields?.translation || 'SelectionText';
+        const translationField = config.fields?.translation || 'SelectionText';
         let resolvedMiscInfoField: string | null = null;
         let resolvedSentenceAudioField: string = audioFieldName;
 
         fields[sentenceField] = sentence;
 
-        const ankiAiConfig = this.deps.getConfig().ai;
+        const ankiAiConfig = config.ai;
         const ankiAiEnabled =
           typeof ankiAiConfig === 'object' && ankiAiConfig !== null
             ? ankiAiConfig.enabled === true
@@ -554,7 +621,7 @@ export class CardCreationService {
 
         if (sentenceCardConfig.lapisEnabled || sentenceCardConfig.kikuEnabled) {
           fields.IsSentenceCard = 'x';
-          fields[getConfiguredWordFieldName(this.deps.getConfig())] = sentence;
+          fields[getConfiguredWordFieldName(config)] = sentence;
         }
 
         const pendingNoteInfo = this.createPendingNoteInfo(fields);
@@ -563,7 +630,7 @@ export class CardCreationService {
         );
         const pendingExpressionText = getPreferredWordValueFromExtractedFields(
           pendingNoteFields,
-          this.deps.getConfig(),
+          config,
         ).trim();
         let duplicateNoteIds: number[] = [];
         if (
@@ -581,7 +648,7 @@ export class CardCreationService {
           }
         }
 
-        const deck = this.deps.getConfig().deck || 'Default';
+        const deck = config.deck || 'Default';
         let noteId: number;
         try {
           noteId = await this.deps.client.addNote(
@@ -627,7 +694,7 @@ export class CardCreationService {
               this.deps.resolveNoteFieldName(createdNoteInfo, audioFieldName) || audioFieldName;
             resolvedMiscInfoField = this.deps.resolveConfiguredFieldName(
               createdNoteInfo,
-              this.deps.getConfig().fields?.miscInfo,
+              config.fields?.miscInfo,
             );
 
             const cardTypeFields: Record<string, string> = {};
@@ -645,41 +712,69 @@ export class CardCreationService {
           errors.push('card type fields');
         }
 
+        const label = sentence.length > 30 ? sentence.substring(0, 30) + '...' : sentence;
+        if (shouldQueuePendingYoutubeMedia) {
+          this.deps.queuePendingYoutubeMediaUpdate?.({
+            sourceUrl:
+              trimToNonEmptyString(await this.deps.getYoutubeMediaSourceUrl?.()) ??
+              mpvClient.currentVideoPath,
+            noteId,
+            startTime,
+            endTime,
+            label,
+            audioFieldName: resolvedSentenceAudioField,
+            imageFieldName: config.fields?.image,
+            miscInfoFieldName: resolvedMiscInfoField ?? undefined,
+            generateAudio,
+            generateImage,
+          });
+          await this.deps.showNotification(noteId, label, 'media queued');
+          return true;
+        }
+
+        if (missingRequestedMediaInput) {
+          return false;
+        }
+
         const mediaFields: Record<string, string> = {};
 
-        try {
-          const audioFilename = this.generateAudioFilename();
-          const audioBuffer = audioSourcePath
-            ? await this.mediaGenerateAudio(audioSourcePath, startTime, endTime)
-            : null;
+        if (generateAudio) {
+          try {
+            const audioFilename = this.generateAudioFilename();
+            const audioBuffer = audioSourcePath
+              ? await this.mediaGenerateAudio(audioSourcePath, startTime, endTime)
+              : null;
 
-          if (audioBuffer) {
-            await this.deps.client.storeMediaFile(audioFilename, audioBuffer);
-            const audioValue = `[sound:${audioFilename}]`;
-            mediaFields[resolvedSentenceAudioField] = audioValue;
-            miscInfoFilename = audioFilename;
+            if (audioBuffer) {
+              await this.deps.client.storeMediaFile(audioFilename, audioBuffer);
+              const audioValue = `[sound:${audioFilename}]`;
+              mediaFields[resolvedSentenceAudioField] = audioValue;
+              miscInfoFilename = audioFilename;
+            }
+          } catch (error) {
+            log.error('Failed to generate sentence audio:', (error as Error).message);
+            errors.push('audio');
           }
-        } catch (error) {
-          log.error('Failed to generate sentence audio:', (error as Error).message);
-          errors.push('audio');
         }
 
-        try {
-          const imageFilename = this.generateImageFilename();
-          const imageBuffer = await this.generateImageBuffer(videoPath, startTime, endTime);
+        if (generateImage) {
+          try {
+            const imageFilename = this.generateImageFilename();
+            const imageBuffer = await this.generateImageBuffer(videoPath!, startTime, endTime);
 
-          const imageField = this.deps.getConfig().fields?.image;
-          if (imageBuffer && imageField) {
-            await this.deps.client.storeMediaFile(imageFilename, imageBuffer);
-            mediaFields[imageField] = `<img src="${imageFilename}">`;
-            miscInfoFilename = imageFilename;
+            const imageField = config.fields?.image;
+            if (imageBuffer && imageField) {
+              await this.deps.client.storeMediaFile(imageFilename, imageBuffer);
+              mediaFields[imageField] = `<img src="${imageFilename}">`;
+              miscInfoFilename = imageFilename;
+            }
+          } catch (error) {
+            log.error('Failed to generate sentence image:', (error as Error).message);
+            errors.push('image');
           }
-        } catch (error) {
-          log.error('Failed to generate sentence image:', (error as Error).message);
-          errors.push('image');
         }
 
-        if (this.deps.getConfig().fields?.miscInfo) {
+        if (config.fields?.miscInfo) {
           const miscInfo = this.deps.formatMiscInfoPattern(miscInfoFilename || '', startTime);
           if (miscInfo && resolvedMiscInfoField) {
             mediaFields[resolvedMiscInfoField] = miscInfo;
@@ -695,7 +790,6 @@ export class CardCreationService {
           }
         }
 
-        const label = sentence.length > 30 ? sentence.substring(0, 30) + '...' : sentence;
         const errorSuffix = errors.length > 0 ? `${errors.join(', ')} failed` : undefined;
         await this.deps.showNotification(noteId, label, errorSuffix);
         return true;
@@ -731,7 +825,7 @@ export class CardCreationService {
   }
 
   private async mediaGenerateAudio(
-    videoPath: string,
+    videoPath: MediaInput,
     startTime: number,
     endTime: number,
   ): Promise<Buffer | null> {
@@ -745,12 +839,16 @@ export class CardCreationService {
       startTime,
       endTime,
       this.deps.getConfig().media?.audioPadding,
-      mpvClient.currentAudioStreamIndex ?? undefined,
+      resolveAudioStreamIndexForMediaGeneration(
+        videoPath,
+        mpvClient.currentAudioStreamIndex ?? undefined,
+      ),
+      this.deps.getConfig().media?.normalizeAudio !== false,
     );
   }
 
   private async generateImageBuffer(
-    videoPath: string,
+    videoPath: MediaInput,
     startTime: number,
     endTime: number,
     animatedLeadInSeconds = 0,

src/anki-integration/known-word-cache.test.ts

@@ -94,7 +94,7 @@ test('KnownWordCacheManager startLifecycle keeps fresh persisted cache without i
       JSON.stringify({
         version: 2,
         refreshedAtMs: 120_000,
-        scope: '{"refreshMinutes":60,"scope":"is:note","fieldsWord":""}',
+        scope: '{"refreshMinutes":60,"scope":"all","fieldsWord":""}',
         words: ['猫'],
         notes: {
           '1': ['猫'],
@@ -143,7 +143,7 @@ test('KnownWordCacheManager startLifecycle immediately refreshes stale persisted
       JSON.stringify({
         version: 2,
         refreshedAtMs: 59_000,
-        scope: '{"refreshMinutes":1,"scope":"is:note","fieldsWord":"Word"}',
+        scope: '{"refreshMinutes":1,"scope":"all","fieldsWord":"Word"}',
         words: ['猫'],
         notes: {
           '1': ['猫'],
@@ -229,7 +229,7 @@ test('KnownWordCacheManager refresh incrementally reconciles deleted and edited
       JSON.stringify({
         version: 2,
         refreshedAtMs: 1,
-        scope: '{"refreshMinutes":1440,"scope":"is:note","fieldsWord":"Word"}',
+        scope: '{"refreshMinutes":1440,"scope":"all","fieldsWord":"Word"}',
         words: ['猫', '犬'],
         notes: {
           '1': ['猫'],
@@ -276,6 +276,36 @@ test('KnownWordCacheManager refresh incrementally reconciles deleted and edited
   }
 });
 
+test('KnownWordCacheManager uses empty query when no known-word deck is configured', async () => {
+  const config: AnkiConnectConfig = {
+    fields: {
+      word: 'Word',
+    },
+    knownWords: {
+      highlightEnabled: true,
+    },
+  };
+  const { manager, clientState, cleanup } = createKnownWordCacheHarness(config);
+
+  try {
+    clientState.findNotesByQuery.set('', [1]);
+    clientState.notesInfoResult = [
+      {
+        noteId: 1,
+        fields: {
+          Word: { value: '猫' },
+        },
+      },
+    ];
+
+    await manager.refresh(true);
+
+    assert.equal(manager.isKnownWord('猫'), true);
+  } finally {
+    cleanup();
+  }
+});
+
 test('KnownWordCacheManager skips malformed note info without fields', async () => {
   const config: AnkiConnectConfig = {
     fields: {
@@ -364,7 +394,7 @@ test('KnownWordCacheManager preserves cache state key captured before refresh wo
       scope: string;
       words: string[];
     };
-    assert.equal(persisted.scope, '{"refreshMinutes":1,"scope":"is:note","fieldsWord":"Word"}');
+    assert.equal(persisted.scope, '{"refreshMinutes":1,"scope":"all","fieldsWord":"Word"}');
     assert.deepEqual(persisted.words, ['猫']);
   } finally {
     fs.rmSync(stateDir, { recursive: true, force: true });
@@ -568,7 +598,7 @@ test('KnownWordCacheManager reports immediate append cache clears as mutations',
       JSON.stringify({
         version: 2,
         refreshedAtMs: Date.now(),
-        scope: '{"refreshMinutes":60,"scope":"is:note","fieldsWord":"Expression"}',
+        scope: '{"refreshMinutes":60,"scope":"all","fieldsWord":"Expression"}',
         words: ['猫'],
         notes: {
           '1': ['猫'],

src/anki-integration/known-word-cache.ts

@@ -48,7 +48,7 @@ export function getKnownWordCacheScopeForConfig(config: AnkiConnectConfig): stri
   }
 
   const configuredDeck = trimToNonEmptyString(config.deck);
-  return configuredDeck ? `deck:${configuredDeck}` : 'is:note';
+  return configuredDeck ? `deck:${configuredDeck}` : 'all';
 }
 
 export function getKnownWordCacheLifecycleConfig(config: AnkiConnectConfig): string {
@@ -396,7 +396,7 @@ export class KnownWordCacheManager {
   private buildKnownWordsQuery(): string {
     const decks = this.getKnownWordDecks();
     if (decks.length === 0) {
-      return 'is:note';
+      return '';
     }
 
     if (decks.length === 1) {

src/anki-integration/media-source.test.ts

@@ -1,7 +1,34 @@
 import assert from 'node:assert/strict';
 import test from 'node:test';
 
-import { resolveMediaGenerationInputPath } from './media-source';
+import * as mediaSource from './media-source';
+import { toMpvEdlValue } from './mpv-edl-test-utils';
+
+const { resolveMediaGenerationInputPath } = mediaSource;
+
+type StructuredMediaInput = {
+  path: string;
+  source: string;
+  singleResolvedStream: boolean;
+  inputOptions?: {
+    reconnect?: boolean;
+    userAgent?: string;
+    headers?: Record<string, string>;
+  };
+};
+
+type StructuredMediaResolver = (
+  mpvClient: Parameters<typeof resolveMediaGenerationInputPath>[0],
+  kind?: Parameters<typeof resolveMediaGenerationInputPath>[1],
+  options?: {
+    getCachedMediaPath?: (
+      currentVideoPath: string,
+      kind: Parameters<typeof resolveMediaGenerationInputPath>[1],
+    ) => Promise<string | null>;
+    remoteCacheMode?: 'optional' | 'required';
+    logDebug?: (message: string) => void;
+  },
+) => Promise<StructuredMediaInput | null>;
 
 test('resolveMediaGenerationInputPath keeps local file paths', async () => {
   const result = await resolveMediaGenerationInputPath({
@@ -27,9 +54,11 @@ test('resolveMediaGenerationInputPath prefers stream-open-filename for remote me
 });
 
 test('resolveMediaGenerationInputPath unwraps mpv edl source for audio and video', async () => {
+  const audioUrl = 'https://audio.example/videoplayback?mime=audio%2Fwebm';
+  const videoUrl = 'https://video.example/videoplayback?mime=video%2Fmp4';
   const edlSource = [
-    'edl://!new_stream;!no_clip;!no_chapters;%70%https://audio.example/videoplayback?mime=audio%2Fwebm',
-    '!new_stream;!no_clip;!no_chapters;%69%https://video.example/videoplayback?mime=video%2Fmp4',
+    `edl://!new_stream;!no_clip;!no_chapters;${toMpvEdlValue(audioUrl)}`,
+    `!new_stream;!no_clip;!no_chapters;${toMpvEdlValue(videoUrl)}`,
     '!global_tags,title=test',
   ].join(';');
 
@@ -48,8 +77,52 @@ test('resolveMediaGenerationInputPath unwraps mpv edl source for audio and video
     'video',
   );
 
-  assert.equal(audioResult, 'https://audio.example/videoplayback?mime=audio%2Fwebm');
-  assert.equal(videoResult, 'https://video.example/videoplayback?mime=video%2Fmp4');
+  assert.equal(audioResult, audioUrl);
+  assert.equal(videoResult, videoUrl);
+});
+
+test('resolveMediaGenerationInputPath strips mpv edl segment options from unwrapped streams', async () => {
+  const audioUrl = 'https://audio.example/videoplayback?mime=audio%2Fwebm';
+  const signedVideoUrl =
+    'https://rr1---sn.example.googlevideo.com/videoplayback?mime=video%2Fmp4&mn=sn-a,sn-b&lsig=abc%3D';
+  const edlSource = [
+    `edl://!new_stream;!no_clip;!no_chapters;${toMpvEdlValue(audioUrl)}`,
+    `!new_stream;!no_clip;!no_chapters;${toMpvEdlValue(signedVideoUrl)},title=clip,length=73,timestamps=chapters`,
+    '!global_tags,title=test',
+  ].join(';');
+
+  const result = await resolveMediaGenerationInputPath(
+    {
+      currentVideoPath: 'https://www.youtube.com/watch?v=abc123',
+      requestProperty: async () => edlSource,
+    },
+    'video',
+  );
+
+  assert.equal(result, signedVideoUrl);
+});
+
+test('resolveMediaGenerationInputPath ignores length-guarded URLs in mpv edl headers', async () => {
+  const initUrl = 'https://init.example/init.mp4';
+  const audioUrl = 'https://audio.example/stream';
+  const videoUrl = 'https://video.example/stream';
+  const edlSource = [
+    `edl://!mp4_dash,init=${toMpvEdlValue(initUrl)}`,
+    '!new_stream',
+    toMpvEdlValue(audioUrl),
+    '!new_stream',
+    toMpvEdlValue(videoUrl),
+  ].join(';');
+
+  const audioResult = await resolveMediaGenerationInputPath(
+    {
+      currentVideoPath: 'https://www.youtube.com/watch?v=abc123',
+      requestProperty: async () => edlSource,
+    },
+    'audio',
+  );
+
+  assert.equal(audioResult, audioUrl);
 });
 
 test('resolveMediaGenerationInputPath falls back to currentVideoPath when stream-open-filename fails', async () => {
@@ -62,3 +135,213 @@ test('resolveMediaGenerationInputPath falls back to currentVideoPath when stream
 
   assert.equal(result, 'https://www.youtube.com/watch?v=abc123');
 });
+
+test('resolveMediaGenerationInput returns single-stream metadata for mpv EDL URLs', async () => {
+  const resolver = (
+    mediaSource as typeof mediaSource & {
+      resolveMediaGenerationInput?: StructuredMediaResolver;
+    }
+  ).resolveMediaGenerationInput;
+  assert.equal(typeof resolver, 'function');
+
+  const audioUrl = 'https://audio.example/videoplayback?mime=audio%2Fwebm';
+  const videoUrl = 'https://video.example/videoplayback?mime=video%2Fmp4';
+  const edlSource = [
+    `edl://!new_stream;!no_clip;!no_chapters;${toMpvEdlValue(audioUrl)}`,
+    `!new_stream;!no_clip;!no_chapters;${toMpvEdlValue(videoUrl)}`,
+  ].join(';');
+
+  const result = await resolver!(
+    {
+      currentVideoPath: 'https://www.youtube.com/watch?v=abc123',
+      requestProperty: async (name: string) => {
+        if (name === 'stream-open-filename') return edlSource;
+        if (name === 'user-agent') return 'Mozilla/5.0';
+        if (name === 'http-header-fields') {
+          return ['Cookie: SID=secret', 'Referer: https://www.youtube.com/', 'X-Test: ok'];
+        }
+        return null;
+      },
+    },
+    'audio',
+  );
+
+  assert.equal(result?.path, audioUrl);
+  assert.equal(result?.singleResolvedStream, true);
+  assert.equal(result?.inputOptions?.reconnect, true);
+  assert.equal(result?.inputOptions?.userAgent, 'Mozilla/5.0');
+  assert.deepEqual(result?.inputOptions?.headers, {
+    Referer: 'https://www.youtube.com/',
+    'X-Test': 'ok',
+  });
+});
+
+test('resolveMediaGenerationInput reads file-local mpv request options', async () => {
+  const resolver = (
+    mediaSource as typeof mediaSource & {
+      resolveMediaGenerationInput?: StructuredMediaResolver;
+    }
+  ).resolveMediaGenerationInput;
+  assert.equal(typeof resolver, 'function');
+
+  const result = await resolver!(
+    {
+      currentVideoPath: 'https://www.youtube.com/watch?v=abc123',
+      requestProperty: async (name: string) => {
+        if (name === 'stream-open-filename') {
+          return 'https://rr1---sn.example.googlevideo.com/videoplayback?id=123';
+        }
+        if (name === 'file-local-options/user-agent') return 'SubMiner Test Agent';
+        if (name === 'options/http-header-fields') return ['X-Shared: ok'];
+        if (name === 'file-local-options/http-header-fields') {
+          return ['Cookie: SID=secret', 'Referer: https://m.youtube.com/', 'X-Local: yes'];
+        }
+        return null;
+      },
+    },
+    'video',
+  );
+
+  assert.equal(result?.path, 'https://rr1---sn.example.googlevideo.com/videoplayback?id=123');
+  assert.equal(result?.singleResolvedStream, true);
+  assert.equal(result?.inputOptions?.userAgent, 'SubMiner Test Agent');
+  assert.deepEqual(result?.inputOptions?.headers, {
+    'X-Shared': 'ok',
+    Referer: 'https://m.youtube.com/',
+    'X-Local': 'yes',
+    Origin: 'https://www.youtube.com',
+  });
+});
+
+test('resolveMediaGenerationInput prefers a ready cached media file for YouTube extraction', async () => {
+  const resolver = (
+    mediaSource as typeof mediaSource & {
+      resolveMediaGenerationInput?: StructuredMediaResolver;
+    }
+  ).resolveMediaGenerationInput;
+  assert.equal(typeof resolver, 'function');
+
+  const result = await resolver!(
+    {
+      currentVideoPath: 'https://www.youtube.com/watch?v=abc123',
+      requestProperty: async () => 'https://rr1---sn.example.googlevideo.com/videoplayback?id=123',
+    },
+    'video',
+    {
+      getCachedMediaPath: async () => '/tmp/subminer-youtube-media-cache/abc123/media.mkv',
+    },
+  );
+
+  assert.equal(result?.path, '/tmp/subminer-youtube-media-cache/abc123/media.mkv');
+  assert.equal(result?.source, 'youtube-cache');
+  assert.equal(result?.singleResolvedStream, false);
+  assert.equal(result?.inputOptions, undefined);
+});
+
+test('resolveMediaGenerationInput debug-logs sanitized YouTube cache hits', async () => {
+  const resolver = (
+    mediaSource as typeof mediaSource & {
+      resolveMediaGenerationInput?: StructuredMediaResolver;
+    }
+  ).resolveMediaGenerationInput;
+  assert.equal(typeof resolver, 'function');
+  const logs: string[] = [];
+
+  const result = await resolver!(
+    {
+      currentVideoPath: 'https://www.youtube.com/watch?v=abc123&signature=secret',
+      requestProperty: async () => 'https://rr1---sn.example.googlevideo.com/videoplayback?id=123',
+    },
+    'video',
+    {
+      getCachedMediaPath: async () => '/tmp/subminer-youtube-media-cache/abc123/media.mkv',
+      logDebug: (message) => logs.push(message),
+    },
+  );
+
+  assert.equal(result?.source, 'youtube-cache');
+  assert.match(logs.join('\n'), /kind=video source=youtube-cache/);
+  assert.match(
+    logs.join('\n'),
+    /input=local:\/tmp\/subminer-youtube-media-cache\/abc123\/media\.mkv/,
+  );
+  assert.match(logs.join('\n'), /current=remote:www\.youtube\.com/);
+  assert.doesNotMatch(logs.join('\n'), /signature=secret|videoplayback/);
+});
+
+test('resolveMediaGenerationInput does not fall back to direct remote streams when cache is required', async () => {
+  const resolver = (
+    mediaSource as typeof mediaSource & {
+      resolveMediaGenerationInput?: StructuredMediaResolver;
+    }
+  ).resolveMediaGenerationInput;
+  assert.equal(typeof resolver, 'function');
+
+  const result = await resolver!(
+    {
+      currentVideoPath: 'https://www.youtube.com/watch?v=abc123',
+      requestProperty: async () => 'https://rr1---sn.example.googlevideo.com/videoplayback?id=123',
+    },
+    'video',
+    {
+      getCachedMediaPath: async () => null,
+      remoteCacheMode: 'required',
+    },
+  );
+
+  assert.equal(result, null);
+});
+
+test('resolveMediaGenerationInput falls back when optional cache lookup fails', async () => {
+  const resolver = (
+    mediaSource as typeof mediaSource & {
+      resolveMediaGenerationInput?: StructuredMediaResolver;
+    }
+  ).resolveMediaGenerationInput;
+  assert.equal(typeof resolver, 'function');
+
+  const result = await resolver!(
+    {
+      currentVideoPath: 'https://www.youtube.com/watch?v=abc123',
+      requestProperty: async () => 'https://rr1---sn.example.googlevideo.com/videoplayback?id=123',
+    },
+    'video',
+    {
+      getCachedMediaPath: async () => {
+        throw new Error('cache unavailable');
+      },
+    },
+  );
+
+  assert.equal(result?.path, 'https://rr1---sn.example.googlevideo.com/videoplayback?id=123');
+  assert.equal(result?.source, 'stream-open-filename');
+});
+
+test('resolveMediaGenerationInput debug-logs sanitized required-cache misses', async () => {
+  const resolver = (
+    mediaSource as typeof mediaSource & {
+      resolveMediaGenerationInput?: StructuredMediaResolver;
+    }
+  ).resolveMediaGenerationInput;
+  assert.equal(typeof resolver, 'function');
+  const logs: string[] = [];
+
+  const result = await resolver!(
+    {
+      currentVideoPath: 'https://www.youtube.com/watch?v=abc123&signature=secret',
+      requestProperty: async () => 'https://rr1---sn.example.googlevideo.com/videoplayback?id=123',
+    },
+    'video',
+    {
+      getCachedMediaPath: async () => null,
+      remoteCacheMode: 'required',
+      logDebug: (message) => logs.push(message),
+    },
+  );
+
+  assert.equal(result, null);
+  assert.match(logs.join('\n'), /kind=video source=cache-miss/);
+  assert.match(logs.join('\n'), /mode=required/);
+  assert.match(logs.join('\n'), /current=remote:www\.youtube\.com/);
+  assert.doesNotMatch(logs.join('\n'), /signature=secret|videoplayback|googlevideo/);
+});

src/anki-integration/media-source.ts

@@ -1,7 +1,58 @@
 import { isRemoteMediaPath } from '../jimaku/utils';
+import type { MediaInput, MediaInputOptions } from '../media-input';
 import type { MpvClient } from '../types/runtime';
+import { extractFileUrlsFromMpvEdlSource } from './mpv-edl';
 
 export type MediaGenerationKind = 'audio' | 'video';
+export type MediaGenerationInputSource =
+  | 'current-path'
+  | 'stream-open-filename'
+  | 'edl-stream'
+  | 'youtube-cache';
+
+export interface ResolvedMediaGenerationInput {
+  path: string;
+  kind: MediaGenerationKind;
+  source: MediaGenerationInputSource;
+  singleResolvedStream: boolean;
+  inputOptions?: MediaInputOptions;
+}
+
+export interface MediaGenerationInputResolverOptions {
+  getCachedMediaPath?: (
+    currentVideoPath: string,
+    kind: MediaGenerationKind,
+  ) => Promise<string | null>;
+  remoteCacheMode?: 'optional' | 'required';
+  logDebug?: (message: string) => void;
+}
+
+export function resolveAudioStreamIndexForMediaGeneration(
+  input: MediaInput,
+  audioStreamIndex: number | null | undefined,
+): number | undefined {
+  if (typeof input === 'object' && 'source' in input && input.source === 'youtube-cache') {
+    return undefined;
+  }
+  return audioStreamIndex ?? undefined;
+}
+
+const BLOCKED_HTTP_HEADER_NAMES = new Set(['authorization', 'cookie', 'proxy-authorization']);
+const HTTP_HEADER_FIELD_PROPERTY_NAMES = [
+  'http-header-fields',
+  'options/http-header-fields',
+  'file-local-options/http-header-fields',
+] as const;
+const USER_AGENT_PROPERTY_NAMES = [
+  'file-local-options/user-agent',
+  'options/user-agent',
+  'user-agent',
+] as const;
+const REFERRER_PROPERTY_NAMES = [
+  'file-local-options/referrer',
+  'options/referrer',
+  'referrer',
+] as const;
 
 function trimToNonEmptyString(value: unknown): string | null {
   if (typeof value !== 'string') {
@@ -11,10 +62,20 @@ function trimToNonEmptyString(value: unknown): string | null {
   return trimmed.length > 0 ? trimmed : null;
 }
 
+function normalizeHeaderName(value: string): string | null {
+  const trimmed = value.trim();
+  if (!/^[A-Za-z0-9!#$%&'*+.^_`|~-]+$/.test(trimmed)) {
+    return null;
+  }
+  if (BLOCKED_HTTP_HEADER_NAMES.has(trimmed.toLowerCase())) {
+    return null;
+  }
+  return trimmed;
+}
+
 function extractUrlsFromMpvEdlSource(source: string): string[] {
-  const matches = source.matchAll(/%\d+%(https?:\/\/.*?)(?=;!new_stream|;!global_tags|$)/gms);
-  return [...matches]
-    .map((match) => trimToNonEmptyString(match[1]))
+  return extractFileUrlsFromMpvEdlSource(source)
+    .map((value) => trimToNonEmptyString(value))
     .filter((value): value is string => value !== null);
 }
 
@@ -53,6 +114,317 @@ function resolvePreferredUrlFromMpvEdlSource(
   return kind === 'audio' ? (urls[0] ?? null) : (urls[urls.length - 1] ?? null);
 }
 
+function getHostname(value: string): string | null {
+  try {
+    return new URL(value).hostname.toLowerCase();
+  } catch {
+    return null;
+  }
+}
+
+function matchesHost(hostname: string, expectedHost: string): boolean {
+  return hostname === expectedHost || hostname.endsWith(`.${expectedHost}`);
+}
+
+function isGoogleVideoMediaPath(value: string): boolean {
+  const host = getHostname(value);
+  return Boolean(host && matchesHost(host, 'googlevideo.com'));
+}
+
+function describeMediaPathForDebugLog(value: string): string {
+  try {
+    const url = new URL(value);
+    if (url.protocol === 'http:' || url.protocol === 'https:') {
+      return `remote:${url.hostname.toLowerCase() || 'unknown'}`;
+    }
+    return `${url.protocol.replace(/:$/, '')}:`;
+  } catch {
+    // Not a URL; treat as a local file path below.
+  }
+
+  if (value.startsWith('edl://')) {
+    return 'edl:';
+  }
+
+  return `local:${value}`;
+}
+
+function logMediaResolutionDebug(
+  options: MediaGenerationInputResolverOptions,
+  message: string,
+): void {
+  if (!options.logDebug) {
+    return;
+  }
+  try {
+    options.logDebug(`[media-source] ${message}`);
+  } catch {
+    // Debug logging should not affect media generation.
+  }
+}
+
+function logResolvedMediaGenerationInput(
+  options: MediaGenerationInputResolverOptions,
+  currentVideoPath: string,
+  result: ResolvedMediaGenerationInput,
+): void {
+  logMediaResolutionDebug(
+    options,
+    [
+      `kind=${result.kind}`,
+      `source=${result.source}`,
+      `input=${describeMediaPathForDebugLog(result.path)}`,
+      `current=${describeMediaPathForDebugLog(currentVideoPath)}`,
+      `singleResolvedStream=${result.singleResolvedStream}`,
+    ].join(' '),
+  );
+}
+
+function logMediaGenerationInputMiss(
+  options: MediaGenerationInputResolverOptions,
+  kind: MediaGenerationKind,
+  currentVideoPath: string,
+  reason: string,
+): void {
+  logMediaResolutionDebug(
+    options,
+    [
+      `kind=${kind}`,
+      'source=cache-miss',
+      `reason=${reason}`,
+      `mode=${options.remoteCacheMode ?? 'optional'}`,
+      `current=${describeMediaPathForDebugLog(currentVideoPath)}`,
+    ].join(' '),
+  );
+}
+
+function setHeaderIfMissing(headers: Record<string, string>, name: string, value: string): void {
+  const lowerName = name.toLowerCase();
+  if (!Object.keys(headers).some((existing) => existing.toLowerCase() === lowerName)) {
+    headers[name] = value;
+  }
+}
+
+function parseMpvHeaderField(value: string): [string, string] | null {
+  const separatorIndex = value.indexOf(':');
+  if (separatorIndex <= 0) {
+    return null;
+  }
+
+  const name = normalizeHeaderName(value.slice(0, separatorIndex));
+  const headerValue = trimToNonEmptyString(value.slice(separatorIndex + 1));
+  if (!name || !headerValue) {
+    return null;
+  }
+  return [name, headerValue.replace(/[\r\n]+/g, ' ')];
+}
+
+function toHeaderFields(value: unknown): string[] {
+  if (Array.isArray(value)) {
+    return value.filter((entry): entry is string => typeof entry === 'string');
+  }
+  if (typeof value === 'string') {
+    return value
+      .split(/\r?\n/)
+      .map((entry) => entry.trim())
+      .filter((entry) => entry.length > 0);
+  }
+  return [];
+}
+
+async function requestOptionalMpvProperty(
+  mpvClient: Pick<MpvClient, 'requestProperty'>,
+  name: string,
+): Promise<unknown> {
+  if (!mpvClient.requestProperty) {
+    return null;
+  }
+
+  try {
+    return await mpvClient.requestProperty(name);
+  } catch {
+    return null;
+  }
+}
+
+async function requestFirstNonEmptyStringProperty(
+  mpvClient: Pick<MpvClient, 'requestProperty'>,
+  names: readonly string[],
+): Promise<string | null> {
+  for (const name of names) {
+    const value = trimToNonEmptyString(await requestOptionalMpvProperty(mpvClient, name));
+    if (value) {
+      return value;
+    }
+  }
+  return null;
+}
+
+async function resolveRemoteInputOptions(
+  mpvClient: Pick<MpvClient, 'requestProperty'>,
+  resolvedPath: string,
+): Promise<MediaInputOptions | undefined> {
+  if (!isRemoteMediaPath(resolvedPath) || !mpvClient.requestProperty) {
+    return undefined;
+  }
+
+  const headers: Record<string, string> = {};
+  for (const propertyName of HTTP_HEADER_FIELD_PROPERTY_NAMES) {
+    const mpvHeaderFields = toHeaderFields(
+      await requestOptionalMpvProperty(mpvClient, propertyName),
+    );
+    for (const field of mpvHeaderFields) {
+      const parsed = parseMpvHeaderField(field);
+      if (parsed) {
+        headers[parsed[0]] = parsed[1];
+      }
+    }
+  }
+
+  const userAgent = await requestFirstNonEmptyStringProperty(mpvClient, USER_AGENT_PROPERTY_NAMES);
+  const referrer = await requestFirstNonEmptyStringProperty(mpvClient, REFERRER_PROPERTY_NAMES);
+  if (referrer) {
+    setHeaderIfMissing(headers, 'Referer', referrer);
+  }
+  if (isGoogleVideoMediaPath(resolvedPath)) {
+    setHeaderIfMissing(headers, 'Referer', 'https://www.youtube.com/');
+    setHeaderIfMissing(headers, 'Origin', 'https://www.youtube.com');
+  }
+
+  return {
+    reconnect: true,
+    ...(userAgent ? { userAgent } : {}),
+    ...(Object.keys(headers).length > 0 ? { headers } : {}),
+  };
+}
+
+async function toResolvedMediaGenerationInput(
+  mpvClient: Pick<MpvClient, 'requestProperty'>,
+  path: string,
+  kind: MediaGenerationKind,
+  source: MediaGenerationInputSource,
+  singleResolvedStream: boolean,
+): Promise<ResolvedMediaGenerationInput> {
+  const inputOptions = await resolveRemoteInputOptions(mpvClient, path);
+  return {
+    path,
+    kind,
+    source,
+    singleResolvedStream,
+    ...(inputOptions ? { inputOptions } : {}),
+  };
+}
+
+export async function resolveMediaGenerationInput(
+  mpvClient: Pick<MpvClient, 'currentVideoPath' | 'requestProperty'> | null | undefined,
+  kind: MediaGenerationKind = 'video',
+  options: MediaGenerationInputResolverOptions = {},
+): Promise<ResolvedMediaGenerationInput | null> {
+  const currentVideoPath = trimToNonEmptyString(mpvClient?.currentVideoPath);
+  if (!currentVideoPath) {
+    logMediaResolutionDebug(options, `kind=${kind} source=none reason=no-current-video`);
+    return null;
+  }
+
+  if (!isRemoteMediaPath(currentVideoPath)) {
+    const result: ResolvedMediaGenerationInput = {
+      path: currentVideoPath,
+      kind,
+      source: 'current-path',
+      singleResolvedStream: false,
+    };
+    logResolvedMediaGenerationInput(options, currentVideoPath, result);
+    return result;
+  }
+
+  let cachedPath: string | null = null;
+  if (options.getCachedMediaPath) {
+    try {
+      cachedPath = trimToNonEmptyString(await options.getCachedMediaPath(currentVideoPath, kind));
+    } catch {
+      cachedPath = null;
+    }
+  }
+  if (cachedPath) {
+    const result: ResolvedMediaGenerationInput = {
+      path: cachedPath,
+      kind,
+      source: 'youtube-cache',
+      singleResolvedStream: false,
+    };
+    logResolvedMediaGenerationInput(options, currentVideoPath, result);
+    return result;
+  }
+
+  if (options.remoteCacheMode === 'required') {
+    logMediaGenerationInputMiss(options, kind, currentVideoPath, 'required-cache-unavailable');
+    return null;
+  }
+
+  if (!mpvClient?.requestProperty) {
+    const result: ResolvedMediaGenerationInput = {
+      path: currentVideoPath,
+      kind,
+      source: 'current-path',
+      singleResolvedStream: false,
+    };
+    logResolvedMediaGenerationInput(options, currentVideoPath, result);
+    return result;
+  }
+
+  try {
+    const streamOpenFilename = trimToNonEmptyString(
+      await mpvClient.requestProperty('stream-open-filename'),
+    );
+    if (streamOpenFilename?.startsWith('edl://')) {
+      const preferredUrl = resolvePreferredUrlFromMpvEdlSource(streamOpenFilename, kind);
+      if (preferredUrl) {
+        const result = await toResolvedMediaGenerationInput(
+          mpvClient,
+          preferredUrl,
+          kind,
+          'edl-stream',
+          true,
+        );
+        logResolvedMediaGenerationInput(options, currentVideoPath, result);
+        return result;
+      }
+      const result = await toResolvedMediaGenerationInput(
+        mpvClient,
+        streamOpenFilename,
+        kind,
+        'stream-open-filename',
+        false,
+      );
+      logResolvedMediaGenerationInput(options, currentVideoPath, result);
+      return result;
+    }
+    if (streamOpenFilename) {
+      const result = await toResolvedMediaGenerationInput(
+        mpvClient,
+        streamOpenFilename,
+        kind,
+        'stream-open-filename',
+        isRemoteMediaPath(streamOpenFilename),
+      );
+      logResolvedMediaGenerationInput(options, currentVideoPath, result);
+      return result;
+    }
+  } catch {
+    // Fall back to the current path when mpv does not expose a resolved stream URL.
+  }
+
+  const result = await toResolvedMediaGenerationInput(
+    mpvClient,
+    currentVideoPath,
+    kind,
+    'current-path',
+    false,
+  );
+  logResolvedMediaGenerationInput(options, currentVideoPath, result);
+  return result;
+}
+
 export async function resolveMediaGenerationInputPath(
   mpvClient: Pick<MpvClient, 'currentVideoPath' | 'requestProperty'> | null | undefined,
   kind: MediaGenerationKind = 'video',

src/anki-integration/mpv-edl-test-utils.ts

@@ -0,0 +1,3 @@
+export function toMpvEdlValue(value: string): string {
+  return `%${Buffer.byteLength(value, 'utf8')}%${value}`;
+}

src/anki-integration/mpv-edl.test.ts

@@ -0,0 +1,37 @@
+import assert from 'node:assert/strict';
+import test from 'node:test';
+
+import { extractFileUrlsFromMpvEdlSource } from './mpv-edl';
+import { toMpvEdlValue } from './mpv-edl-test-utils';
+
+test('extractFileUrlsFromMpvEdlSource honors length-guarded file values', () => {
+  const url =
+    'https://rr1---sn.example.googlevideo.com/videoplayback?mime=video%2Fmp4&mn=sn-a,sn-b&lsig=abc%3D';
+  const source = `edl://!new_stream;${toMpvEdlValue(url)},title=clip,length=73`;
+
+  assert.deepEqual(extractFileUrlsFromMpvEdlSource(source), [url]);
+});
+
+test('extractFileUrlsFromMpvEdlSource reads file parameters', () => {
+  const initUrl = 'https://init.example/init.mp4';
+  const fileUrl = 'https://video.example/videoplayback?mime=video%2Fmp4';
+  const source = `edl://!mp4_dash,init=${toMpvEdlValue(initUrl)};file=${toMpvEdlValue(
+    fileUrl,
+  )},length=42`;
+
+  assert.deepEqual(extractFileUrlsFromMpvEdlSource(source), [fileUrl]);
+});
+
+test('extractFileUrlsFromMpvEdlSource aggregates file URLs across entries', () => {
+  const audioUrl = 'https://audio.example/videoplayback?mime=audio%2Fwebm';
+  const videoUrl = 'https://video.example/videoplayback?mime=video%2Fmp4';
+  const source = [
+    'edl://!new_stream',
+    toMpvEdlValue(audioUrl),
+    '!new_stream',
+    `file=${toMpvEdlValue(videoUrl)},length=50`,
+    '!global_tags,title=test',
+  ].join(';');
+
+  assert.deepEqual(extractFileUrlsFromMpvEdlSource(source), [audioUrl, videoUrl]);
+});

src/anki-integration/mpv-edl.ts

@@ -0,0 +1,195 @@
+const EDL_URI_PREFIX = 'edl://';
+
+const BYTE_COMMA = ','.charCodeAt(0);
+const BYTE_CR = '\r'.charCodeAt(0);
+const BYTE_EQUALS = '='.charCodeAt(0);
+const BYTE_EXCLAMATION = '!'.charCodeAt(0);
+const BYTE_LF = '\n'.charCodeAt(0);
+const BYTE_PERCENT = '%'.charCodeAt(0);
+const BYTE_SEMICOLON = ';'.charCodeAt(0);
+
+function isDigitByte(value: number | undefined): value is number {
+  return value !== undefined && value >= 48 && value <= 57;
+}
+
+function isEntrySeparator(value: number | undefined): boolean {
+  return value === BYTE_SEMICOLON || value === BYTE_LF || value === BYTE_CR;
+}
+
+function isParamSeparator(value: number | undefined): boolean {
+  return value === BYTE_COMMA || isEntrySeparator(value);
+}
+
+function decodeBytes(buffer: Buffer, start: number, end: number): string {
+  return buffer.subarray(start, end).toString('utf8');
+}
+
+function isHttpUrl(value: string): boolean {
+  return /^https?:\/\//i.test(value);
+}
+
+function toEdlDataBuffer(source: string): Buffer {
+  const data = source.startsWith(EDL_URI_PREFIX) ? source.slice(EDL_URI_PREFIX.length) : source;
+  return Buffer.from(data, 'utf8');
+}
+
+function parseLengthGuardedValue(
+  buffer: Buffer,
+  position: number,
+): { value: string; end: number } | null {
+  if (buffer[position] !== BYTE_PERCENT) {
+    return null;
+  }
+
+  let cursor = position + 1;
+  if (!isDigitByte(buffer[cursor])) {
+    return null;
+  }
+
+  let byteLength = 0;
+  while (true) {
+    const digit = buffer[cursor];
+    if (!isDigitByte(digit)) {
+      break;
+    }
+    byteLength = byteLength * 10 + (digit - 48);
+    cursor += 1;
+  }
+
+  if (buffer[cursor] !== BYTE_PERCENT) {
+    return null;
+  }
+
+  const valueStart = cursor + 1;
+  const valueEnd = valueStart + byteLength;
+  if (valueEnd > buffer.length) {
+    return null;
+  }
+
+  return {
+    value: decodeBytes(buffer, valueStart, valueEnd),
+    end: valueEnd,
+  };
+}
+
+function skipEntrySeparators(buffer: Buffer, position: number): number {
+  let cursor = position;
+  while (cursor < buffer.length && isEntrySeparator(buffer[cursor])) {
+    cursor += 1;
+  }
+  return cursor;
+}
+
+function skipEntry(buffer: Buffer, position: number): number {
+  let cursor = position;
+  while (cursor < buffer.length) {
+    const guardedValue = parseLengthGuardedValue(buffer, cursor);
+    if (guardedValue) {
+      cursor = guardedValue.end;
+      continue;
+    }
+    if (isEntrySeparator(buffer[cursor])) {
+      break;
+    }
+    cursor += 1;
+  }
+  return cursor;
+}
+
+function parseRawValue(buffer: Buffer, position: number): { value: string; end: number } {
+  let cursor = position;
+  while (
+    cursor < buffer.length &&
+    !isParamSeparator(buffer[cursor]) &&
+    buffer[cursor] !== BYTE_EXCLAMATION
+  ) {
+    cursor += 1;
+  }
+  return {
+    value: decodeBytes(buffer, position, cursor),
+    end: cursor,
+  };
+}
+
+function parseParamValue(buffer: Buffer, position: number): { value: string; end: number } {
+  return parseLengthGuardedValue(buffer, position) ?? parseRawValue(buffer, position);
+}
+
+function parseOptionalParamName(
+  buffer: Buffer,
+  position: number,
+): { name: string | null; valueStart: number } {
+  let cursor = position;
+  while (
+    cursor < buffer.length &&
+    !isParamSeparator(buffer[cursor]) &&
+    buffer[cursor] !== BYTE_PERCENT &&
+    buffer[cursor] !== BYTE_EXCLAMATION
+  ) {
+    if (buffer[cursor] === BYTE_EQUALS) {
+      return {
+        name: decodeBytes(buffer, position, cursor),
+        valueStart: cursor + 1,
+      };
+    }
+    cursor += 1;
+  }
+
+  return { name: null, valueStart: position };
+}
+
+function parseSegmentEntry(buffer: Buffer, position: number): { urls: string[]; end: number } {
+  const urls: string[] = [];
+  let cursor = position;
+  let unnamedParamIndex = 0;
+
+  while (cursor < buffer.length && !isEntrySeparator(buffer[cursor])) {
+    const { name, valueStart } = parseOptionalParamName(buffer, cursor);
+    const value = parseParamValue(buffer, valueStart);
+    const lowerName = name?.toLowerCase() ?? null;
+    const isFileParam = lowerName === 'file' || (lowerName === null && unnamedParamIndex === 0);
+
+    if (isFileParam && isHttpUrl(value.value)) {
+      urls.push(value.value);
+    }
+
+    if (lowerName === null) {
+      unnamedParamIndex += 1;
+    }
+
+    cursor = value.end;
+    if (buffer[cursor] === BYTE_COMMA) {
+      cursor += 1;
+      continue;
+    }
+    if (!isEntrySeparator(buffer[cursor])) {
+      cursor = skipEntry(buffer, cursor);
+    }
+  }
+
+  return { urls, end: cursor };
+}
+
+export function extractFileUrlsFromMpvEdlSource(source: string): string[] {
+  const buffer = toEdlDataBuffer(source);
+  const urls: string[] = [];
+  let cursor = 0;
+
+  while (cursor < buffer.length) {
+    cursor = skipEntrySeparators(buffer, cursor);
+    if (cursor >= buffer.length) {
+      break;
+    }
+
+    if (buffer[cursor] === BYTE_EXCLAMATION) {
+      cursor = skipEntry(buffer, cursor);
+      continue;
+    }
+
+    const segment = parseSegmentEntry(buffer, cursor);
+    urls.push(...segment.urls);
+    cursor = segment.end;
+  }
+
+  return urls;
+}

src/anki-integration/note-field-utils.ts

@@ -0,0 +1,37 @@
+export interface NoteFieldValueInfo {
+  fields: Record<string, { value: string }>;
+}
+
+export function getNoteFieldValue(
+  noteInfo: NoteFieldValueInfo,
+  preferredName: string,
+): string | null {
+  const resolvedFieldName = Object.keys(noteInfo.fields).find(
+    (fieldName) => fieldName.toLowerCase() === preferredName.toLowerCase(),
+  );
+  return resolvedFieldName ? (noteInfo.fields[resolvedFieldName]?.value ?? '') : null;
+}
+
+export function hasNoteFieldValue(noteInfo: NoteFieldValueInfo, preferredName: string): boolean {
+  return (getNoteFieldValue(noteInfo, preferredName) ?? '').trim().length > 0;
+}
+
+export function shouldMarkWordAndSentenceCard(
+  noteInfo: NoteFieldValueInfo,
+  sentenceCardConfig: { lapisEnabled: boolean; kikuEnabled: boolean },
+): boolean {
+  if (!sentenceCardConfig.lapisEnabled && !sentenceCardConfig.kikuEnabled) {
+    return false;
+  }
+
+  const wordAndSentenceValue = getNoteFieldValue(noteInfo, 'IsWordAndSentenceCard');
+  if (wordAndSentenceValue === null) {
+    return false;
+  }
+  if (wordAndSentenceValue.trim().length > 0) {
+    return true;
+  }
+  return (
+    !hasNoteFieldValue(noteInfo, 'IsSentenceCard') && !hasNoteFieldValue(noteInfo, 'IsAudioCard')
+  );
+}

src/anki-integration/note-update-workflow.test.ts

@@ -7,6 +7,27 @@ import {
 } from './note-update-workflow';
 import type { SubtitleMiningContext } from '../types/subtitle';
 
+function setWordAndSentenceCardTypeFields(
+  updatedFields: Record<string, string>,
+  availableFieldNames: string[],
+  cardKind: 'word-and-sentence',
+): void {
+  assert.equal(cardKind, 'word-and-sentence');
+  const resolveFieldName = (preferredName: string): string | null =>
+    availableFieldNames.find((name) => name.toLowerCase() === preferredName.toLowerCase()) ?? null;
+
+  const wordAndSentenceFlag = resolveFieldName('IsWordAndSentenceCard');
+  if (!wordAndSentenceFlag) return;
+
+  updatedFields[wordAndSentenceFlag] = 'x';
+  for (const flagName of ['IsSentenceCard', 'IsAudioCard']) {
+    const resolved = resolveFieldName(flagName);
+    if (resolved && resolved !== wordAndSentenceFlag) {
+      updatedFields[resolved] = '';
+    }
+  }
+}
+
 function createWorkflowHarness() {
   const updates: Array<{ noteId: number; fields: Record<string, string> }> = [];
   const notifications: Array<{ noteId: number; label: string | number }> = [];
@@ -40,6 +61,7 @@ function createWorkflowHarness() {
     getCurrentSubtitleStart: () => 12.3,
     getEffectiveSentenceCardConfig: () => ({
       sentenceField: 'Sentence',
+      lapisEnabled: false,
       kikuEnabled: false,
       kikuFieldGrouping: 'disabled' as const,
     }),
@@ -57,6 +79,7 @@ function createWorkflowHarness() {
     handleFieldGroupingManual: async (_originalNoteId, _newNoteId, _newNoteInfo, _expression) =>
       false,
     processSentence: (text: string, _noteFields: Record<string, string>) => text,
+    setCardTypeFields: setWordAndSentenceCardTypeFields,
     resolveConfiguredFieldName: (noteInfo: NoteUpdateWorkflowNoteInfo, preferred?: string) => {
       if (!preferred) return null;
       const names = Object.keys(noteInfo.fields);
@@ -102,6 +125,118 @@ test('NoteUpdateWorkflow updates sentence field and emits notification', async (
   assert.equal(harness.notifications.length, 1);
 });
 
+test('NoteUpdateWorkflow updates sentence furigana when highlight processor changes it', async () => {
+  const harness = createWorkflowHarness();
+  harness.deps.client.notesInfo = async () =>
+    [
+      {
+        noteId: 42,
+        fields: {
+          Expression: { value: 'tokugi' },
+          Sentence: { value: '' },
+          SentenceFurigana: { value: '<span class="term">tokugi</span>' },
+        },
+      },
+    ] satisfies NoteUpdateWorkflowNoteInfo[];
+  harness.deps.processSentenceFurigana = (sentenceFurigana) =>
+    sentenceFurigana.replace('tokugi', '<b>tokugi</b>');
+
+  await harness.workflow.execute(42);
+
+  assert.equal(harness.updates.length, 1);
+  assert.deepEqual(harness.updates[0]?.fields, {
+    Sentence: 'subtitle-text',
+    SentenceFurigana: '<span class="term"><b>tokugi</b></span>',
+  });
+});
+
+test('NoteUpdateWorkflow marks enriched Kiku word cards as word-and-sentence cards', async () => {
+  const harness = createWorkflowHarness();
+  harness.deps.getEffectiveSentenceCardConfig = () => ({
+    sentenceField: 'Sentence',
+    lapisEnabled: false,
+    kikuEnabled: true,
+    kikuFieldGrouping: 'manual',
+  });
+  harness.deps.client.notesInfo = async () =>
+    [
+      {
+        noteId: 42,
+        fields: {
+          Expression: { value: 'taberu' },
+          Sentence: { value: '' },
+          IsWordAndSentenceCard: { value: '' },
+          IsSentenceCard: { value: '' },
+          IsAudioCard: { value: '' },
+        },
+      },
+    ] satisfies NoteUpdateWorkflowNoteInfo[];
+
+  await harness.workflow.execute(42);
+
+  assert.equal(harness.updates.length, 1);
+  assert.deepEqual(harness.updates[0]?.fields, {
+    Sentence: 'subtitle-text',
+    IsWordAndSentenceCard: 'x',
+    IsSentenceCard: '',
+    IsAudioCard: '',
+  });
+});
+
+test('NoteUpdateWorkflow does not set Kiku card flags when Lapis and Kiku are disabled', async () => {
+  const harness = createWorkflowHarness();
+  harness.deps.client.notesInfo = async () =>
+    [
+      {
+        noteId: 42,
+        fields: {
+          Expression: { value: 'taberu' },
+          Sentence: { value: '' },
+          IsWordAndSentenceCard: { value: '' },
+          IsSentenceCard: { value: '' },
+          IsAudioCard: { value: '' },
+        },
+      },
+    ] satisfies NoteUpdateWorkflowNoteInfo[];
+
+  await harness.workflow.execute(42);
+
+  assert.equal(harness.updates.length, 1);
+  assert.deepEqual(harness.updates[0]?.fields, {
+    Sentence: 'subtitle-text',
+  });
+});
+
+test('NoteUpdateWorkflow preserves explicit sentence card type during sentence enrichment', async () => {
+  const harness = createWorkflowHarness();
+  harness.deps.getEffectiveSentenceCardConfig = () => ({
+    sentenceField: 'Sentence',
+    lapisEnabled: true,
+    kikuEnabled: false,
+    kikuFieldGrouping: 'disabled',
+  });
+  harness.deps.client.notesInfo = async () =>
+    [
+      {
+        noteId: 42,
+        fields: {
+          Expression: { value: 'sentence expression' },
+          Sentence: { value: '' },
+          IsWordAndSentenceCard: { value: '' },
+          IsSentenceCard: { value: 'x' },
+          IsAudioCard: { value: '' },
+        },
+      },
+    ] satisfies NoteUpdateWorkflowNoteInfo[];
+
+  await harness.workflow.execute(42);
+
+  assert.equal(harness.updates.length, 1);
+  assert.deepEqual(harness.updates[0]?.fields, {
+    Sentence: 'subtitle-text',
+  });
+});
+
 test('NoteUpdateWorkflow no-ops when note info is missing', async () => {
   const harness = createWorkflowHarness();
   harness.deps.client.notesInfo = async () => [];
@@ -119,6 +254,7 @@ test('NoteUpdateWorkflow updates note before auto field grouping merge', async (
   let notesInfoCallCount = 0;
   harness.deps.getEffectiveSentenceCardConfig = () => ({
     sentenceField: 'Sentence',
+    lapisEnabled: false,
     kikuEnabled: true,
     kikuFieldGrouping: 'auto',
   });
@@ -273,3 +409,59 @@ test('NoteUpdateWorkflow uses subtitle sidebar context for sentence media timing
   assert.deepEqual(imageContext, sidebarContext);
   assert.equal(miscInfoStartTime, 10);
 });
+
+test('NoteUpdateWorkflow queues media updates when YouTube cache is pending', async () => {
+  const harness = createWorkflowHarness();
+  const queuedUpdates: Array<{
+    noteId: number;
+    noteInfo: NoteUpdateWorkflowNoteInfo;
+    context?: SubtitleMiningContext;
+    label: string | number;
+  }> = [];
+  const mediaCalls: string[] = [];
+
+  harness.deps.client.notesInfo = async () =>
+    [
+      {
+        noteId: 42,
+        fields: {
+          Expression: { value: 'taberu' },
+          Sentence: { value: '' },
+          SentenceAudio: { value: '' },
+          Picture: { value: '' },
+        },
+      },
+    ] satisfies NoteUpdateWorkflowNoteInfo[];
+  harness.deps.getConfig = () => ({
+    fields: {
+      sentence: 'Sentence',
+      image: 'Picture',
+    },
+    media: {
+      generateAudio: true,
+      generateImage: true,
+    },
+    behavior: {},
+  });
+  harness.deps.generateAudio = async () => {
+    mediaCalls.push('audio');
+    return Buffer.from('audio');
+  };
+  harness.deps.generateImage = async () => {
+    mediaCalls.push('image');
+    return Buffer.from('image');
+  };
+  harness.deps.queuePendingYoutubeMediaUpdate = async (job) => {
+    queuedUpdates.push(job);
+    return true;
+  };
+
+  await harness.workflow.execute(42);
+
+  assert.deepEqual(mediaCalls, []);
+  assert.equal(queuedUpdates.length, 1);
+  assert.equal(queuedUpdates[0]?.noteId, 42);
+  assert.equal(queuedUpdates[0]?.label, 'taberu');
+  assert.equal(queuedUpdates[0]?.context, undefined);
+  assert.deepEqual(harness.updates, [{ noteId: 42, fields: { Sentence: 'subtitle-text' } }]);
+});

src/anki-integration/note-update-workflow.ts

@@ -1,6 +1,7 @@
 import { DEFAULT_ANKI_CONNECT_CONFIG } from '../config';
 import { getPreferredWordValueFromExtractedFields } from '../anki-field-config';
 import type { SubtitleMiningContext } from '../types/subtitle';
+import { shouldMarkWordAndSentenceCard } from './note-field-utils';
 
 export interface NoteUpdateWorkflowNoteInfo {
   noteId: number;
@@ -35,6 +36,7 @@ export interface NoteUpdateWorkflowDeps {
   getCurrentSubtitleStart: () => number | undefined;
   getEffectiveSentenceCardConfig: () => {
     sentenceField: string;
+    lapisEnabled: boolean;
     kikuEnabled: boolean;
     kikuFieldGrouping: 'auto' | 'manual' | 'disabled';
   };
@@ -58,6 +60,15 @@ export interface NoteUpdateWorkflowDeps {
     expression: string,
   ) => Promise<boolean>;
   processSentence: (mpvSentence: string, noteFields: Record<string, string>) => string;
+  processSentenceFurigana?: (
+    sentenceFurigana: string,
+    noteFields: Record<string, string>,
+  ) => string;
+  setCardTypeFields: (
+    updatedFields: Record<string, string>,
+    availableFieldNames: string[],
+    cardKind: 'word-and-sentence',
+  ) => void;
   resolveConfiguredFieldName: (
     noteInfo: NoteUpdateWorkflowNoteInfo,
     ...preferredNames: (string | undefined)[]
@@ -74,6 +85,12 @@ export interface NoteUpdateWorkflowDeps {
   ) => Promise<Buffer | null>;
   formatMiscInfoPattern: (fallbackFilename: string, startTimeSeconds?: number) => string;
   consumeSubtitleMiningContext?: () => SubtitleMiningContext | null;
+  queuePendingYoutubeMediaUpdate?: (job: {
+    noteId: number;
+    noteInfo: NoteUpdateWorkflowNoteInfo;
+    context?: SubtitleMiningContext;
+    label: string | number;
+  }) => Promise<boolean>;
   addConfiguredTagsToNote: (noteId: number) => Promise<void>;
   showNotification: (noteId: number, label: string | number) => Promise<void>;
   showOsdNotification: (message: string) => void;
@@ -184,15 +201,52 @@ export class NoteUpdateWorkflow {
         sentenceField,
         config.fields?.sentence,
       );
+      const noteLabel = hasExpressionText ? expressionText : noteId;
 
       const currentSubtitleText = subtitleMiningContext?.text ?? this.deps.getCurrentSubtitleText();
       if (sentenceField && currentSubtitleText) {
         const processedSentence = this.deps.processSentence(currentSubtitleText, fields);
         updatedFields[sentenceField] = processedSentence;
+        if (shouldMarkWordAndSentenceCard(noteInfo, sentenceCardConfig)) {
+          this.deps.setCardTypeFields(
+            updatedFields,
+            Object.keys(noteInfo.fields),
+            'word-and-sentence',
+          );
+        }
         updatePerformed = true;
       }
+      const sentenceFuriganaField = this.deps.resolveConfiguredFieldName(
+        noteInfo,
+        'SentenceFurigana',
+      );
+      const existingSentenceFurigana = sentenceFuriganaField
+        ? noteInfo.fields[sentenceFuriganaField]?.value || ''
+        : '';
+      if (sentenceFuriganaField && existingSentenceFurigana && this.deps.processSentenceFurigana) {
+        const processedSentenceFurigana = this.deps.processSentenceFurigana(
+          existingSentenceFurigana,
+          fields,
+        );
+        if (processedSentenceFurigana !== existingSentenceFurigana) {
+          updatedFields[sentenceFuriganaField] = processedSentenceFurigana;
+          updatePerformed = true;
+        }
+      }
 
-      if (config.media?.generateAudio) {
+      const generateAudio = config.media?.generateAudio !== false;
+      const generateImage = config.media?.generateImage !== false;
+      const mediaCacheQueued =
+        (generateAudio || generateImage) && this.deps.queuePendingYoutubeMediaUpdate
+          ? await this.deps.queuePendingYoutubeMediaUpdate({
+              noteId,
+              noteInfo,
+              context: subtitleMiningContext ?? undefined,
+              label: noteLabel,
+            })
+          : false;
+
+      if (!mediaCacheQueued && generateAudio) {
         try {
           const audioFilename = this.deps.generateAudioFilename();
           const audioBuffer = await this.deps.generateAudio(subtitleMiningContext ?? undefined);
@@ -217,7 +271,7 @@ export class NoteUpdateWorkflow {
         }
       }
 
-      if (config.media?.generateImage) {
+      if (!mediaCacheQueued && generateImage) {
         try {
           const animatedLeadInSeconds = await this.deps.getAnimatedImageLeadInSeconds(noteInfo);
           const imageFilename = this.deps.generateImageFilename();
@@ -252,7 +306,7 @@ export class NoteUpdateWorkflow {
         }
       }
 
-      if (config.fields?.miscInfo) {
+      if (!mediaCacheQueued && config.fields?.miscInfo) {
         const miscInfo = this.deps.formatMiscInfoPattern(
           miscInfoFilename || '',
           subtitleMiningContext?.startTime ?? this.deps.getCurrentSubtitleStart(),
@@ -270,8 +324,8 @@ export class NoteUpdateWorkflow {
       if (updatePerformed) {
         await this.deps.client.updateNoteFields(noteId, updatedFields);
         await this.deps.addConfiguredTagsToNote(noteId);
-        this.deps.logInfo('Updated card fields for:', hasExpressionText ? expressionText : noteId);
-        await this.deps.showNotification(noteId, hasExpressionText ? expressionText : noteId);
+        this.deps.logInfo('Updated card fields for:', noteLabel);
+        await this.deps.showNotification(noteId, noteLabel);
       }
 
       if (shouldRunFieldGrouping && hasExpressionText && duplicateNoteId !== null) {

src/anki-integration/pending-youtube-media-queue.test.ts

@@ -0,0 +1,193 @@
+import assert from 'node:assert/strict';
+import test from 'node:test';
+
+import type { AnkiConnectConfig } from '../types/anki';
+import {
+  PendingYoutubeMediaQueue,
+  type PendingYoutubeMediaQueueDeps,
+} from './pending-youtube-media-queue';
+
+function createDeps(
+  overrides: Partial<PendingYoutubeMediaQueueDeps> = {},
+): PendingYoutubeMediaQueueDeps {
+  const warnings: unknown[][] = [];
+  const deps: PendingYoutubeMediaQueueDeps & { warnings: unknown[][] } = {
+    client: {
+      notesInfo: async () => [],
+      updateNoteFields: async () => {},
+      storeMediaFile: async () => {},
+    },
+    mediaGenerator: {
+      generateAudio: async () => Buffer.from('audio'),
+      generateScreenshot: async () => Buffer.from('image'),
+      generateAnimatedImage: async () => Buffer.from('image'),
+    },
+    getConfig: () =>
+      ({
+        media: { generateAudio: true, generateImage: true },
+        fields: {},
+      }) as AnkiConnectConfig,
+    getCurrentVideoPath: () => 'https://www.youtube.com/watch?v=abc123',
+    getCachedMediaPath: async () => null,
+    shouldRequireRemoteMediaCache: () => true,
+    getSubtitleMediaRange: () => ({ startTime: 1, endTime: 2 }),
+    getResolvedSentenceAudioFieldName: () => 'SentenceAudio',
+    resolveConfiguredFieldName: () => 'Picture',
+    mergeFieldValue: (_existing, newValue) => newValue,
+    getAnimatedImageLeadInSeconds: async () => 0,
+    generateAudioFilename: () => 'audio.mp3',
+    generateImageFilename: () => 'image.webp',
+    formatMiscInfoPatternForMediaPath: () => '',
+    showStatusNotification: () => {},
+    showNotification: async () => {},
+    logInfo: () => {},
+    logWarn: (...args) => {
+      warnings.push(args);
+    },
+    logError: () => {},
+    warnings,
+    ...overrides,
+  };
+  return deps;
+}
+
+test('PendingYoutubeMediaQueue treats cache lookup failures as an immediate generation fallback', async () => {
+  const deps = createDeps({
+    getCachedMediaPath: async () => {
+      throw new Error('cache unavailable');
+    },
+  });
+  const queue = new PendingYoutubeMediaQueue(deps);
+
+  const queued = await queue.queueFromNote({
+    noteId: 42,
+    noteInfo: { noteId: 42, fields: {} },
+    label: 'demo',
+  });
+
+  assert.equal(queued, false);
+  assert.deepEqual((deps as typeof deps & { warnings: unknown[][] }).warnings, [
+    [
+      'Failed to read YouTube cache state; falling back to immediate media generation:',
+      'cache unavailable',
+    ],
+  ]);
+});
+
+test('PendingYoutubeMediaQueue drains matching queued jobs when the cache download fails', async () => {
+  const statusMessages: string[] = [];
+  const notifications: Array<{ noteId: number; label: string | number; suffix?: string }> = [];
+  const updatedNotes: number[] = [];
+  const deps = createDeps({
+    client: {
+      notesInfo: async (noteIds) =>
+        noteIds.map((noteId) => ({
+          noteId,
+          fields: {
+            SentenceAudio: { value: '' },
+            Picture: { value: '' },
+          },
+        })),
+      updateNoteFields: async (noteId) => {
+        updatedNotes.push(noteId);
+      },
+      storeMediaFile: async () => {},
+    },
+    showStatusNotification: (message) => {
+      statusMessages.push(message);
+    },
+    showNotification: async (noteId, label, suffix) => {
+      notifications.push({ noteId, label, suffix });
+    },
+  });
+  const queue = new PendingYoutubeMediaQueue(deps);
+
+  queue.enqueue({
+    sourceUrl: 'https://www.youtube.com/watch?v=abc123',
+    noteId: 42,
+    startTime: 1,
+    endTime: 2,
+    label: 'queued',
+    generateAudio: true,
+    generateImage: true,
+  });
+
+  await queue.handleFailed('https://youtu.be/abc123');
+  await queue.handleReady('https://youtu.be/abc123', '/tmp/media.mkv');
+
+  assert.deepEqual(updatedNotes, []);
+  assert.deepEqual(notifications, [{ noteId: 42, label: 'queued', suffix: 'media cache failed' }]);
+  assert.equal(
+    statusMessages.includes('YouTube media cache failed. Media was not added to 1 queued card.'),
+    true,
+  );
+});
+
+test('PendingYoutubeMediaQueue defaults missing media flags to enabled when queuing from notes', async () => {
+  const updatedNotes: Array<{ noteId: number; fields: Record<string, string> }> = [];
+  const storedMedia: string[] = [];
+  const deps = createDeps({
+    client: {
+      notesInfo: async (noteIds) =>
+        noteIds.map((noteId) => ({
+          noteId,
+          fields: {
+            SentenceAudio: { value: '' },
+            Picture: { value: '' },
+          },
+        })),
+      updateNoteFields: async (noteId, fields) => {
+        updatedNotes.push({ noteId, fields });
+      },
+      storeMediaFile: async (filename) => {
+        storedMedia.push(filename);
+      },
+    },
+    getConfig: () => ({ media: {}, fields: { image: 'Picture' } }) as AnkiConnectConfig,
+  });
+  const queue = new PendingYoutubeMediaQueue(deps);
+
+  const queued = await queue.queueFromNote({
+    noteId: 42,
+    noteInfo: { noteId: 42, fields: {} },
+    label: 'demo',
+  });
+  await queue.handleReady('https://youtu.be/abc123', '/tmp/media.mkv');
+
+  assert.equal(queued, true);
+  assert.equal(updatedNotes.length, 1);
+  assert.equal(storedMedia.length, 2);
+  assert.match(updatedNotes[0]?.fields.SentenceAudio ?? '', /^\[sound:audio\.mp3\]$/);
+  assert.match(updatedNotes[0]?.fields.Picture ?? '', /^<img src="image\.webp">$/);
+});
+
+test('PendingYoutubeMediaQueue only announces a download once per source while jobs collect', () => {
+  const statusMessages: string[] = [];
+  const deps = createDeps({
+    showStatusNotification: (message) => {
+      statusMessages.push(message);
+    },
+  });
+  const queue = new PendingYoutubeMediaQueue(deps);
+
+  queue.enqueue({
+    sourceUrl: 'https://youtu.be/abc123',
+    noteId: 1,
+    startTime: 1,
+    endTime: 2,
+    label: 'first',
+    generateAudio: true,
+    generateImage: false,
+  });
+  queue.enqueue({
+    sourceUrl: 'https://www.youtube.com/watch?v=abc123',
+    noteId: 2,
+    startTime: 3,
+    endTime: 4,
+    label: 'second',
+    generateAudio: false,
+    generateImage: true,
+  });
+
+  assert.equal(statusMessages.length, 1);
+});

src/anki-integration/pending-youtube-media-queue.ts

@@ -0,0 +1,392 @@
+import { DEFAULT_ANKI_CONNECT_CONFIG } from '../config';
+import type { MediaInput } from '../media-input';
+import type { MediaGenerator } from '../media-generator';
+import type { AnkiConnectConfig } from '../types/anki';
+import type { SubtitleMiningContext } from '../types/subtitle';
+import { youtubeMediaUrlsMatch, type PendingYoutubeMediaUpdate } from './pending-youtube-media';
+import type { MediaGenerationInputResolverOptions } from './media-source';
+
+type PendingYoutubeMediaUpdateResult = 'updated' | 'partial' | 'failed';
+
+export interface PendingYoutubeMediaQueueReadyOptions {
+  notifyNoQueued?: boolean;
+}
+
+export interface PendingYoutubeMediaQueueFailedOptions {
+  notifyStatus?: boolean;
+}
+
+export interface PendingYoutubeMediaNoteInfo {
+  noteId: number;
+  fields: Record<string, { value: string }>;
+}
+
+export interface PendingYoutubeMediaQueueDeps {
+  client: {
+    notesInfo(noteIds: number[]): Promise<unknown>;
+    updateNoteFields(noteId: number, fields: Record<string, string>): Promise<void>;
+    storeMediaFile(filename: string, data: Buffer): Promise<void>;
+  };
+  mediaGenerator: Pick<
+    MediaGenerator,
+    'generateAudio' | 'generateScreenshot' | 'generateAnimatedImage'
+  >;
+  getConfig: () => AnkiConnectConfig;
+  getCurrentVideoPath: () => Promise<string | undefined> | string | undefined;
+  getCachedMediaPath: MediaGenerationInputResolverOptions['getCachedMediaPath'] | null;
+  shouldRequireRemoteMediaCache: () => boolean;
+  getSubtitleMediaRange: (context?: SubtitleMiningContext) => {
+    startTime: number;
+    endTime: number;
+  };
+  getResolvedSentenceAudioFieldName: (noteInfo: PendingYoutubeMediaNoteInfo) => string | null;
+  resolveConfiguredFieldName: (
+    noteInfo: PendingYoutubeMediaNoteInfo,
+    ...preferredNames: (string | undefined)[]
+  ) => string | null;
+  mergeFieldValue: (existing: string, newValue: string, overwrite: boolean) => string;
+  getAnimatedImageLeadInSeconds: (noteInfo: PendingYoutubeMediaNoteInfo) => Promise<number>;
+  generateAudioFilename: () => string;
+  generateImageFilename: () => string;
+  formatMiscInfoPatternForMediaPath: (
+    fallbackFilename: string,
+    startTimeSeconds: number | undefined,
+    mediaPath: string,
+    mediaTitle?: string,
+  ) => string;
+  showStatusNotification: (message: string) => void;
+  showNotification: (noteId: number, label: string | number, errorSuffix?: string) => Promise<void>;
+  logInfo: (message: string, ...args: unknown[]) => void;
+  logWarn: (message: string, ...args: unknown[]) => void;
+  logError: (message: string, ...args: unknown[]) => void;
+}
+
+function trimToNonEmptyString(value: unknown): string | null {
+  if (typeof value !== 'string') {
+    return null;
+  }
+  const trimmed = value.trim();
+  return trimmed.length > 0 ? trimmed : null;
+}
+
+function shouldGenerateAudio(config: AnkiConnectConfig): boolean {
+  return config.media?.generateAudio !== false;
+}
+
+function shouldGenerateImage(config: AnkiConnectConfig): boolean {
+  return config.media?.generateImage !== false;
+}
+
+export class PendingYoutubeMediaQueue {
+  private updates: PendingYoutubeMediaUpdate[] = [];
+
+  constructor(private readonly deps: PendingYoutubeMediaQueueDeps) {}
+
+  enqueue(job: PendingYoutubeMediaUpdate): void {
+    if (!job.generateAudio && !job.generateImage) {
+      return;
+    }
+    const isFirstQueuedForSource = !this.updates.some((existing) =>
+      youtubeMediaUrlsMatch(existing.sourceUrl, job.sourceUrl),
+    );
+    this.updates.push(job);
+    this.deps.logInfo('Queued YouTube media update for note:', job.noteId);
+    if (isFirstQueuedForSource) {
+      this.deps.showStatusNotification(
+        'YouTube media cache is still downloading. Card media will be added when the cache is ready.',
+      );
+    }
+  }
+
+  async queueFromNote(job: {
+    noteId: number;
+    noteInfo: PendingYoutubeMediaNoteInfo;
+    context?: SubtitleMiningContext;
+    label: string | number;
+  }): Promise<boolean> {
+    const sourceUrl = trimToNonEmptyString(await this.deps.getCurrentVideoPath());
+    const getCachedMediaPath = this.deps.getCachedMediaPath;
+    if (!sourceUrl || this.deps.shouldRequireRemoteMediaCache() !== true || !getCachedMediaPath) {
+      return false;
+    }
+
+    let cachedPath: string | null = null;
+    try {
+      cachedPath = await getCachedMediaPath(sourceUrl, 'video');
+    } catch (error) {
+      this.deps.logWarn(
+        'Failed to read YouTube cache state; falling back to immediate media generation:',
+        error instanceof Error ? error.message : String(error),
+      );
+      return false;
+    }
+    if (cachedPath) {
+      return false;
+    }
+
+    const config = this.deps.getConfig();
+    const mediaRange = this.deps.getSubtitleMediaRange(job.context);
+    this.enqueue({
+      sourceUrl,
+      noteId: job.noteId,
+      startTime: mediaRange.startTime,
+      endTime: mediaRange.endTime,
+      label: job.label,
+      audioFieldName: this.deps.getResolvedSentenceAudioFieldName(job.noteInfo) ?? undefined,
+      imageFieldName:
+        this.deps.resolveConfiguredFieldName(
+          job.noteInfo,
+          config.fields?.image,
+          DEFAULT_ANKI_CONNECT_CONFIG.fields.image,
+        ) ?? undefined,
+      miscInfoFieldName:
+        this.deps.resolveConfiguredFieldName(job.noteInfo, config.fields?.miscInfo) ?? undefined,
+      generateAudio: shouldGenerateAudio(config),
+      generateImage: shouldGenerateImage(config),
+    });
+    return true;
+  }
+
+  async handleReady(
+    sourceUrl: string,
+    cachedPath: string,
+    options: PendingYoutubeMediaQueueReadyOptions = {},
+  ): Promise<void> {
+    const jobs = this.takeMatchingUpdates(sourceUrl);
+    if (jobs.length === 0) {
+      if (options.notifyNoQueued !== false) {
+        this.deps.showStatusNotification('YouTube media cache ready.');
+      }
+      return;
+    }
+
+    this.deps.showStatusNotification(
+      `YouTube media cache ready. Adding media to ${jobs.length} queued card${
+        jobs.length === 1 ? '' : 's'
+      }.`,
+    );
+
+    let updatedCount = 0;
+    let partialCount = 0;
+    let failedCount = 0;
+    for (const job of jobs) {
+      try {
+        const result = await this.applyUpdate(job, cachedPath);
+        if (result === 'updated') {
+          updatedCount += 1;
+        } else if (result === 'partial') {
+          partialCount += 1;
+        } else {
+          failedCount += 1;
+        }
+      } catch (error) {
+        failedCount += 1;
+        this.deps.logError(
+          'Failed to apply queued YouTube media update:',
+          error instanceof Error ? error.message : String(error),
+        );
+      }
+    }
+
+    if (partialCount > 0 || failedCount > 0) {
+      this.deps.showStatusNotification(
+        `Queued YouTube media finished with ${updatedCount} updated, ${partialCount} partial, and ${failedCount} failed.`,
+      );
+    }
+  }
+
+  async handleFailed(
+    sourceUrl: string,
+    options: PendingYoutubeMediaQueueFailedOptions = {},
+  ): Promise<void> {
+    const jobs = this.takeMatchingUpdates(sourceUrl);
+    if (jobs.length === 0) {
+      if (options.notifyStatus !== false) {
+        this.deps.showStatusNotification('YouTube media cache failed.');
+      }
+      return;
+    }
+
+    if (options.notifyStatus !== false) {
+      this.deps.showStatusNotification(
+        `YouTube media cache failed. Media was not added to ${jobs.length} queued card${
+          jobs.length === 1 ? '' : 's'
+        }.`,
+      );
+    }
+    this.deps.logWarn('Discarding queued YouTube media updates after cache failure:', jobs.length);
+
+    for (const job of jobs) {
+      try {
+        await this.deps.showNotification(job.noteId, job.label, 'media cache failed');
+      } catch (error) {
+        this.deps.logWarn(
+          'Failed to show queued YouTube media failure notification:',
+          error instanceof Error ? error.message : String(error),
+        );
+      }
+    }
+  }
+
+  private takeMatchingUpdates(sourceUrl: string): PendingYoutubeMediaUpdate[] {
+    const matched: PendingYoutubeMediaUpdate[] = [];
+    const remaining: PendingYoutubeMediaUpdate[] = [];
+    for (const job of this.updates) {
+      if (youtubeMediaUrlsMatch(job.sourceUrl, sourceUrl)) {
+        matched.push(job);
+      } else {
+        remaining.push(job);
+      }
+    }
+    this.updates = remaining;
+    return matched;
+  }
+
+  private async applyUpdate(
+    job: PendingYoutubeMediaUpdate,
+    cachedPath: string,
+  ): Promise<PendingYoutubeMediaUpdateResult> {
+    const notesInfoResult = await this.deps.client.notesInfo([job.noteId]);
+    const notesInfo = notesInfoResult as unknown as PendingYoutubeMediaNoteInfo[];
+    const noteInfo = notesInfo[0];
+    if (!noteInfo) {
+      this.deps.logWarn('Queued YouTube media target note not found:', job.noteId);
+      return 'failed';
+    }
+
+    const config = this.deps.getConfig();
+    const mediaFields: Record<string, string> = {};
+    const errors: string[] = [];
+    let miscInfoFilename: string | null = null;
+    const cachedMediaInput: MediaInput = {
+      path: cachedPath,
+      source: 'youtube-cache',
+    };
+
+    if (job.generateAudio) {
+      try {
+        const audioFilename = this.deps.generateAudioFilename();
+        const audioBuffer = await this.deps.mediaGenerator.generateAudio(
+          cachedMediaInput,
+          job.startTime,
+          job.endTime,
+          config.media?.audioPadding,
+          undefined,
+          config.media?.normalizeAudio !== false,
+        );
+        if (audioBuffer) {
+          await this.deps.client.storeMediaFile(audioFilename, audioBuffer);
+          const audioField =
+            job.audioFieldName || this.deps.getResolvedSentenceAudioFieldName(noteInfo) || null;
+          if (audioField) {
+            const existingAudio = noteInfo.fields[audioField]?.value || '';
+            mediaFields[audioField] = this.deps.mergeFieldValue(
+              existingAudio,
+              `[sound:${audioFilename}]`,
+              config.behavior?.overwriteAudio !== false,
+            );
+          }
+          miscInfoFilename = audioFilename;
+        }
+      } catch (error) {
+        errors.push('audio');
+        this.deps.logError('Failed to generate queued YouTube audio:', (error as Error).message);
+      }
+    }
+
+    if (job.generateImage) {
+      try {
+        const imageFilename = this.deps.generateImageFilename();
+        const animatedLeadInSeconds = await this.deps.getAnimatedImageLeadInSeconds(noteInfo);
+        const imageBuffer = await this.generateImageFromInput(
+          cachedMediaInput,
+          job.startTime,
+          job.endTime,
+          animatedLeadInSeconds,
+        );
+        if (imageBuffer) {
+          await this.deps.client.storeMediaFile(imageFilename, imageBuffer);
+          const imageField =
+            job.imageFieldName ||
+            this.deps.resolveConfiguredFieldName(
+              noteInfo,
+              config.fields?.image,
+              DEFAULT_ANKI_CONNECT_CONFIG.fields.image,
+            );
+          if (imageField) {
+            const existingImage = noteInfo.fields[imageField]?.value || '';
+            mediaFields[imageField] = this.deps.mergeFieldValue(
+              existingImage,
+              `<img src="${imageFilename}">`,
+              config.behavior?.overwriteImage !== false,
+            );
+          } else {
+            this.deps.logWarn(
+              'Image field not found on queued YouTube media note, skipping image update',
+            );
+          }
+          miscInfoFilename = imageFilename;
+        }
+      } catch (error) {
+        errors.push('image');
+        this.deps.logError('Failed to generate queued YouTube image:', (error as Error).message);
+      }
+    }
+
+    if (config.fields?.miscInfo && miscInfoFilename) {
+      const miscInfoField =
+        job.miscInfoFieldName ||
+        this.deps.resolveConfiguredFieldName(noteInfo, config.fields.miscInfo);
+      const miscInfo = this.deps.formatMiscInfoPatternForMediaPath(
+        miscInfoFilename,
+        job.startTime,
+        job.sourceUrl,
+      );
+      if (miscInfoField && miscInfo) {
+        mediaFields[miscInfoField] = miscInfo;
+      }
+    }
+
+    if (Object.keys(mediaFields).length === 0) {
+      return 'failed';
+    }
+
+    await this.deps.client.updateNoteFields(job.noteId, mediaFields);
+    const errorSuffix = errors.length > 0 ? `${errors.join(', ')} failed` : undefined;
+    await this.deps.showNotification(job.noteId, job.label, errorSuffix);
+    this.deps.logInfo('Applied queued YouTube media update for note:', job.noteId);
+    return errors.length === 0 ? 'updated' : 'partial';
+  }
+
+  private async generateImageFromInput(
+    videoPath: MediaInput,
+    startTime: number,
+    endTime: number,
+    animatedLeadInSeconds = 0,
+  ): Promise<Buffer | null> {
+    const config = this.deps.getConfig();
+    if (config.media?.imageType === 'avif') {
+      return this.deps.mediaGenerator.generateAnimatedImage(
+        videoPath,
+        startTime,
+        endTime,
+        config.media?.audioPadding,
+        {
+          fps: config.media?.animatedFps,
+          maxWidth: config.media?.animatedMaxWidth,
+          maxHeight: config.media?.animatedMaxHeight,
+          crf: config.media?.animatedCrf,
+          leadingStillDuration: animatedLeadInSeconds,
+        },
+      );
+    }
+
+    const timestamp = startTime + (endTime - startTime) / 2;
+    return this.deps.mediaGenerator.generateScreenshot(videoPath, timestamp, {
+      format: config.media?.imageFormat as 'jpg' | 'png' | 'webp',
+      quality: config.media?.imageQuality,
+      maxWidth: config.media?.imageMaxWidth,
+      maxHeight: config.media?.imageMaxHeight,
+    });
+  }
+}

src/anki-integration/pending-youtube-media.ts

@@ -0,0 +1,53 @@
+export interface PendingYoutubeMediaUpdate {
+  sourceUrl: string;
+  noteId: number;
+  startTime: number;
+  endTime: number;
+  label: string | number;
+  audioFieldName?: string;
+  imageFieldName?: string;
+  miscInfoFieldName?: string;
+  generateAudio: boolean;
+  generateImage: boolean;
+}
+
+function trimToNonEmptyString(value: unknown): string | null {
+  if (typeof value !== 'string') {
+    return null;
+  }
+  const trimmed = value.trim();
+  return trimmed.length > 0 ? trimmed : null;
+}
+
+function getYoutubeVideoId(rawUrl: string): string | null {
+  try {
+    const parsed = new URL(rawUrl);
+    const host = parsed.hostname.toLowerCase();
+    if (host === 'youtu.be' || host.endsWith('.youtu.be')) {
+      return trimToNonEmptyString(parsed.pathname.replace(/^\/+/, '').split('/')[0]);
+    }
+    if (host === 'youtube.com' || host.endsWith('.youtube.com')) {
+      const watchId = trimToNonEmptyString(parsed.searchParams.get('v'));
+      if (watchId) {
+        return watchId;
+      }
+      const parts = parsed.pathname.split('/').filter(Boolean);
+      if (parts[0] === 'shorts' || parts[0] === 'embed' || parts[0] === 'live') {
+        return trimToNonEmptyString(parts[1]);
+      }
+    }
+  } catch {
+    return null;
+  }
+  return null;
+}
+
+export function youtubeMediaUrlsMatch(a: string, b: string): boolean {
+  if (a.trim() === b.trim()) {
+    return true;
+  }
+
+  const aVideoId = getYoutubeVideoId(a);
+  const bVideoId = getYoutubeVideoId(b);
+  return Boolean(aVideoId && bVideoId && aVideoId === bVideoId);
+}

src/config/config.test.ts

@@ -80,6 +80,8 @@ test('loads defaults when config is missing', () => {
   assert.equal('remoteControlDeviceName' in config.jellyfin, false);
   assert.equal('deviceId' in config.jellyfin, false);
   assert.equal('clientVersion' in config.jellyfin, false);
+  assert.equal(config.youtube.mediaCache.mode, 'direct');
+  assert.equal(config.youtube.mediaCache.maxHeight, 720);
   assert.equal(config.ai.enabled, false);
   assert.equal(config.ai.apiKeyCommand, '');
   assert.equal(config.texthooker.openBrowser, false);
@@ -90,6 +92,7 @@ test('loads defaults when config is missing', () => {
     model: '',
     systemPrompt: '',
   });
+  assert.equal(config.ankiConnect.media.normalizeAudio, true);
   assert.equal(config.startupWarmups.lowPowerMode, false);
   assert.equal(config.startupWarmups.mecab, true);
   assert.equal(config.startupWarmups.yomitanExtension, true);
@@ -1750,6 +1753,56 @@ test('parses global shortcuts and startup settings', () => {
   assert.equal(config.youtubeSubgen.fixWithAi, true);
 });
 
+test('parses YouTube media cache config and warns on invalid values', () => {
+  const validDir = makeTempDir();
+  fs.writeFileSync(
+    path.join(validDir, 'config.jsonc'),
+    `{
+      "youtube": {
+        "mediaCache": {
+          "mode": "background",
+          "maxHeight": 480
+        }
+      }
+    }`,
+    'utf-8',
+  );
+
+  const validService = new ConfigService(validDir);
+  assert.equal(validService.getConfig().youtube.mediaCache.mode, 'background');
+  assert.equal(validService.getConfig().youtube.mediaCache.maxHeight, 480);
+
+  const invalidDir = makeTempDir();
+  fs.writeFileSync(
+    path.join(invalidDir, 'config.jsonc'),
+    `{
+      "youtube": {
+        "mediaCache": {
+          "mode": "always",
+          "maxHeight": -1
+        }
+      }
+    }`,
+    'utf-8',
+  );
+
+  const invalidService = new ConfigService(invalidDir);
+  assert.equal(
+    invalidService.getConfig().youtube.mediaCache.mode,
+    DEFAULT_CONFIG.youtube.mediaCache.mode,
+  );
+  assert.equal(
+    invalidService.getConfig().youtube.mediaCache.maxHeight,
+    DEFAULT_CONFIG.youtube.mediaCache.maxHeight,
+  );
+  assert.ok(
+    invalidService.getWarnings().some((warning) => warning.path === 'youtube.mediaCache.mode'),
+  );
+  assert.ok(
+    invalidService.getWarnings().some((warning) => warning.path === 'youtube.mediaCache.maxHeight'),
+  );
+});
+
 test('parses controller settings with logical bindings and tuning knobs', () => {
   const dir = makeTempDir();
   fs.writeFileSync(

src/config/definitions/defaults-core.ts

@@ -111,6 +111,10 @@ export const CORE_DEFAULT_CONFIG: Pick<
   },
   youtube: {
     primarySubLanguages: ['ja', 'jpn'],
+    mediaCache: {
+      mode: 'direct',
+      maxHeight: 720,
+    },
   },
   subsync: {
     alass_path: '',

src/config/definitions/defaults-integrations.ts

@@ -51,6 +51,7 @@ export const INTEGRATIONS_DEFAULT_CONFIG: Pick<
       animatedMaxHeight: 0,
       animatedCrf: 35,
       syncAnimatedImageToWordAudio: true,
+      normalizeAudio: true,
       audioPadding: 0,
       fallbackDuration: 3.0,
       maxMediaDuration: 30,

src/config/definitions/domain-registry.test.ts

@@ -110,6 +110,7 @@ test('config option registry includes critical paths and has unique entries', ()
     'subtitleStyle.autoPauseVideoOnYomitanPopup',
     'ankiConnect.enabled',
     'subtitleStyle.nameMatchEnabled',
+    'ankiConnect.media.normalizeAudio',
     'anilist.characterDictionary.collapsibleSections.description',
     'mpv.executablePath',
     'mpv.launchMode',

src/config/definitions/options-core.ts

@@ -119,6 +119,24 @@ export function buildCoreConfigOptionRegistry(
       description:
         'Comma-separated primary subtitle language priority for managed subtitle auto-selection.',
     },
+    {
+      path: 'youtube.mediaCache.mode',
+      kind: 'enum',
+      enumValues: ['direct', 'background'],
+      enumLabels: {
+        direct: 'Direct stream extraction',
+        background: 'Background media cache',
+      },
+      defaultValue: defaultConfig.youtube.mediaCache.mode,
+      description: 'How YouTube card audio/images are extracted.',
+    },
+    {
+      path: 'youtube.mediaCache.maxHeight',
+      kind: 'number',
+      defaultValue: defaultConfig.youtube.mediaCache.maxHeight,
+      description:
+        'Maximum video height downloaded for the YouTube background media cache. Set to 0 for unlimited.',
+    },
     {
       path: 'controller.enabled',
       kind: 'boolean',

src/config/definitions/options-integrations.ts

@@ -181,6 +181,12 @@ export function buildIntegrationConfigOptionRegistry(
       defaultValue: defaultConfig.ankiConnect.media.generateAudio,
       description: 'Generate sentence audio for mined cards.',
     },
+    {
+      path: 'ankiConnect.media.normalizeAudio',
+      kind: 'boolean',
+      defaultValue: defaultConfig.ankiConnect.media.normalizeAudio,
+      description: 'Normalize generated sentence audio loudness during media extraction.',
+    },
     {
       path: 'ankiConnect.media.generateImage',
       kind: 'boolean',

src/config/definitions/shared.ts

@@ -32,6 +32,7 @@ export interface ConfigOptionRegistryEntry {
    * `osd` and `osd-system` in NOTIFICATION_TYPE_VALUES.
    */
   enumValues?: readonly string[];
+  enumLabels?: Record<string, string>;
   /**
    * Optional settings UI subset when legacy/runtime-valid enum options should remain
    * editable in config files but hidden from new UI choices, for example

src/config/resolve/core-domains.ts

@@ -297,6 +297,39 @@ export function applyCoreDomainConfig(context: ResolveContext): void {
         'Expected string array.',
       );
     }
+
+    if (isObject(src.youtube.mediaCache)) {
+      const mode = src.youtube.mediaCache.mode;
+      if (mode === 'direct' || mode === 'background') {
+        resolved.youtube.mediaCache.mode = mode;
+      } else if (mode !== undefined) {
+        warn(
+          'youtube.mediaCache.mode',
+          mode,
+          resolved.youtube.mediaCache.mode,
+          "Expected 'direct' or 'background'.",
+        );
+      }
+
+      const maxHeight = asNumber(src.youtube.mediaCache.maxHeight);
+      if (maxHeight !== undefined && Number.isInteger(maxHeight) && maxHeight >= 0) {
+        resolved.youtube.mediaCache.maxHeight = maxHeight;
+      } else if (src.youtube.mediaCache.maxHeight !== undefined) {
+        warn(
+          'youtube.mediaCache.maxHeight',
+          src.youtube.mediaCache.maxHeight,
+          resolved.youtube.mediaCache.maxHeight,
+          'Expected a whole number at least 0.',
+        );
+      }
+    } else if (src.youtube.mediaCache !== undefined) {
+      warn(
+        'youtube.mediaCache',
+        src.youtube.mediaCache,
+        resolved.youtube.mediaCache,
+        'Expected object.',
+      );
+    }
   }
 
   if (isObject(src.subsync)) {

src/config/settings/registry.test.ts

@@ -165,6 +165,20 @@ test('settings registry exposes specialized controls for config-assisted inputs'
   assert.equal(field('discordPresence.presenceStyle').control, 'select');
 });
 
+test('settings registry exposes YouTube media cache mode as a labeled select', () => {
+  const mediaCacheMode = field('youtube.mediaCache.mode');
+  const mediaCacheMaxHeight = field('youtube.mediaCache.maxHeight');
+
+  assert.equal(mediaCacheMode.control, 'select');
+  assert.deepEqual(mediaCacheMode.enumValues, ['direct', 'background']);
+  assert.deepEqual(mediaCacheMode.enumLabels, {
+    direct: 'Direct stream extraction',
+    background: 'Background media cache',
+  });
+  assert.equal(mediaCacheMaxHeight.control, 'number');
+  assert.equal(mediaCacheMaxHeight.defaultValue, 720);
+});
+
 test('settings registry exposes css declaration editor for primary and secondary subtitle appearance', () => {
   const primaryVisible = fields
     .filter(

src/config/settings/registry.ts

@@ -720,6 +720,7 @@ function fieldForLeaf(leaf: Leaf): ConfigSettingsField {
     ...(option?.settingsEnumValues || option?.enumValues
       ? { enumValues: option.settingsEnumValues ?? option.enumValues }
       : {}),
+    ...(option?.enumLabels ? { enumLabels: option.enumLabels } : {}),
     restartBehavior: restartBehaviorForPath(leaf.path),
     advanced:
       leaf.path.startsWith('controller.') ||

src/core/services/__tests__/stats-server.test.ts

@@ -2035,6 +2035,76 @@ Aligned English subtitle
     });
   });
 
+  it('POST /api/stats/mine-card marks Kiku word mining notes as word-and-sentence cards when enabled', async () => {
+    await withTempDir(async (dir) => {
+      const sourcePath = path.join(dir, 'episode.mkv');
+      fs.writeFileSync(sourcePath, 'fake media');
+
+      await withFakeAnkiConnect(
+        async (requests, url) => {
+          const app = createStatsApp(createMockTracker(), {
+            addYomitanNote: async () => 777,
+            createMediaGenerator: () => ({
+              generateAudio: async () => null,
+              generateScreenshot: async () => null,
+              generateAnimatedImage: async () => null,
+            }),
+            ankiConnectConfig: {
+              url,
+              deck: 'Mining',
+              fields: {
+                image: 'Picture',
+                sentence: 'Sentence',
+              },
+              media: {
+                generateAudio: false,
+                generateImage: false,
+              },
+              isKiku: {
+                enabled: true,
+                fieldGrouping: 'disabled',
+                deleteDuplicateInAuto: true,
+              },
+            },
+          });
+
+          const res = await app.request('/api/stats/mine-card?mode=word', {
+            method: 'POST',
+            headers: { 'Content-Type': 'application/json' },
+            body: JSON.stringify({
+              sourcePath,
+              startMs: 1_000,
+              endMs: 2_000,
+              sentence: '猫を見た',
+              word: '猫',
+              videoTitle: 'Episode 1',
+            }),
+          });
+
+          const body = await res.json();
+          assert.equal(res.status, 200, JSON.stringify(body));
+
+          const updateRequest = requests.find((request) => request.action === 'updateNoteFields');
+          const fields = updateRequest?.params?.note?.fields ?? {};
+          assert.equal(fields.Sentence, '<b>猫</b>を見た');
+          assert.equal(fields.IsWordAndSentenceCard, 'x');
+          assert.equal(fields.IsSentenceCard, '');
+          assert.equal(fields.IsAudioCard, '');
+        },
+        {
+          notesInfoFields: {
+            Expression: { value: '猫' },
+            Sentence: { value: '' },
+            Picture: { value: '' },
+            IsWordAndSentenceCard: { value: '' },
+            IsSentenceCard: { value: '' },
+            IsAudioCard: { value: '' },
+          },
+        },
+      );
+    });
+  });
+
   it('POST /api/stats/mine-card writes word mining sentence audio and image together', async () => {
     await withTempDir(async (dir) => {
       const sourcePath = path.join(dir, 'episode.mkv');

src/core/services/anki-jimaku.ts

@@ -40,6 +40,12 @@ export interface AnkiJimakuIpcRuntimeOptions {
   getAnkiIntegration: () => AnkiIntegration | null;
   setAnkiIntegration: (integration: AnkiIntegration | null) => void;
   getKnownWordCacheStatePath: () => string;
+  getCachedMediaPath?: (
+    currentVideoPath: string,
+    kind: 'audio' | 'video',
+  ) => Promise<string | null>;
+  shouldRequireRemoteMediaCache?: () => boolean;
+  getYoutubeMediaSourceUrl?: () => Promise<string | null | undefined> | string | null | undefined;
   showDesktopNotification: (title: string, options: { body?: string; icon?: string }) => void;
   showOverlayNotification?: (payload: OverlayNotificationPayload) => void;
   createFieldGroupingCallback: () => (
@@ -107,6 +113,9 @@ export function registerAnkiJimakuIpcRuntime(
           mergeAiConfig(config.ai, config.ankiConnect?.ai) as AiConfig,
           undefined,
           options.showOverlayNotification,
+          options.getCachedMediaPath,
+          options.shouldRequireRemoteMediaCache,
+          options.getYoutubeMediaSourceUrl,
         );
         integration.start();
         options.setAnkiIntegration(integration);

src/core/services/overlay-runtime-init.ts

@@ -25,6 +25,12 @@ type CreateAnkiIntegrationArgs = {
     data: KikuFieldGroupingRequestData,
   ) => Promise<KikuFieldGroupingChoice>;
   knownWordCacheStatePath: string;
+  getCachedMediaPath?: (
+    currentVideoPath: string,
+    kind: 'audio' | 'video',
+  ) => Promise<string | null>;
+  shouldRequireRemoteMediaCache?: () => boolean;
+  getYoutubeMediaSourceUrl?: () => Promise<string | null | undefined> | string | null | undefined;
 };
 
 export type OverlayWindowTrackerOptions = {
@@ -65,6 +71,9 @@ function createDefaultAnkiIntegration(args: CreateAnkiIntegrationArgs): AnkiInte
     args.aiConfig,
     undefined,
     args.showOverlayNotification,
+    args.getCachedMediaPath,
+    args.shouldRequireRemoteMediaCache,
+    args.getYoutubeMediaSourceUrl,
   );
 }
 
@@ -132,6 +141,12 @@ export function initializeOverlayRuntime(
       data: KikuFieldGroupingRequestData,
     ) => Promise<KikuFieldGroupingChoice>;
     getKnownWordCacheStatePath: () => string;
+    getCachedMediaPath?: (
+      currentVideoPath: string,
+      kind: 'audio' | 'video',
+    ) => Promise<string | null>;
+    shouldRequireRemoteMediaCache?: () => boolean;
+    getYoutubeMediaSourceUrl?: () => Promise<string | null | undefined> | string | null | undefined;
     shouldStartAnkiIntegration?: () => boolean;
     createAnkiIntegration?: (args: CreateAnkiIntegrationArgs) => AnkiIntegrationLike;
     backendOverride: string | null;
@@ -166,6 +181,12 @@ export function initializeOverlayAnkiIntegration(options: {
     data: KikuFieldGroupingRequestData,
   ) => Promise<KikuFieldGroupingChoice>;
   getKnownWordCacheStatePath: () => string;
+  getCachedMediaPath?: (
+    currentVideoPath: string,
+    kind: 'audio' | 'video',
+  ) => Promise<string | null>;
+  shouldRequireRemoteMediaCache?: () => boolean;
+  getYoutubeMediaSourceUrl?: () => Promise<string | null | undefined> | string | null | undefined;
   shouldStartAnkiIntegration?: () => boolean;
   createAnkiIntegration?: (args: CreateAnkiIntegrationArgs) => AnkiIntegrationLike;
 }): boolean {
@@ -200,6 +221,13 @@ export function initializeOverlayAnkiIntegration(options: {
     showOverlayNotification: options.showOverlayNotification,
     createFieldGroupingCallback: options.createFieldGroupingCallback,
     knownWordCacheStatePath: options.getKnownWordCacheStatePath(),
+    ...(options.getCachedMediaPath ? { getCachedMediaPath: options.getCachedMediaPath } : {}),
+    ...(options.shouldRequireRemoteMediaCache
+      ? { shouldRequireRemoteMediaCache: options.shouldRequireRemoteMediaCache }
+      : {}),
+    ...(options.getYoutubeMediaSourceUrl
+      ? { getYoutubeMediaSourceUrl: options.getYoutubeMediaSourceUrl }
+      : {}),
   });
   if (options.shouldStartAnkiIntegration?.() !== false) {
     integration.start();

src/core/services/stats-server.ts

@@ -204,6 +204,29 @@ function getStatsWordMiningAudioFieldName(
   );
 }
 
+function shouldUseStatsLapisKikuCardFields(ankiConfig: AnkiConnectConfig): boolean {
+  return ankiConfig.isLapis?.enabled === true || ankiConfig.isKiku?.enabled === true;
+}
+
+function applyStatsWordAndSentenceCardFields(
+  fields: Record<string, string>,
+  noteInfo: StatsServerNoteInfo | null,
+  ankiConfig: AnkiConnectConfig,
+): void {
+  if (!shouldUseStatsLapisKikuCardFields(ankiConfig) || !noteInfo) return;
+
+  const wordAndSentenceFlag = resolveStatsNoteFieldName(noteInfo, 'IsWordAndSentenceCard');
+  if (!wordAndSentenceFlag) return;
+
+  fields[wordAndSentenceFlag] = 'x';
+  for (const flagName of ['IsSentenceCard', 'IsAudioCard']) {
+    const resolved = resolveStatsNoteFieldName(noteInfo, flagName);
+    if (resolved && resolved !== wordAndSentenceFlag) {
+      fields[resolved] = '';
+    }
+  }
+}
+
 function getStatsDirectMiningAudioFieldNames(
   ankiConfig: AnkiConnectConfig,
   noteInfo: StatsServerNoteInfo | null,
@@ -1183,6 +1206,7 @@ export function createStatsApp(
     const mediaGen = options?.createMediaGenerator?.() ?? new MediaGenerator();
 
     const audioPadding = ankiConfig.media?.audioPadding ?? 0;
+    const normalizeAudio = ankiConfig.media?.normalizeAudio !== false;
     const maxMediaDuration = ankiConfig.media?.maxMediaDuration ?? 30;
 
     const startSec = startMs / 1000;
@@ -1205,7 +1229,14 @@ export function createStatsApp(
 
     const audioPromise = generateAudio
       ? timeMiningPhase(mode, 'generateAudio', () =>
-          mediaGen.generateAudio(sourcePath, startSec, clampedEndSec, audioPadding),
+          mediaGen.generateAudio(
+            sourcePath,
+            startSec,
+            clampedEndSec,
+            audioPadding,
+            null,
+            normalizeAudio,
+          ),
         )
       : Promise.resolve(null);
 
@@ -1299,7 +1330,11 @@ export function createStatsApp(
 
       let imageBuffer = imageResult.status === 'fulfilled' ? imageResult.value : null;
       let noteInfo: StatsServerNoteInfo | null = null;
-      if (audioBuffer || (syncAnimatedImageToWordAudio && generateImage)) {
+      if (
+        audioBuffer ||
+        (syncAnimatedImageToWordAudio && generateImage) ||
+        shouldUseStatsLapisKikuCardFields(ankiConfig)
+      ) {
         try {
           const noteInfoResult = (await client.notesInfo([noteId])) as StatsServerNoteInfo[];
           noteInfo = noteInfoResult[0] ?? null;
@@ -1339,6 +1374,7 @@ export function createStatsApp(
       const imageFieldName = ankiConfig.fields?.image ?? 'Picture';
 
       mediaFields[sentenceFieldName] = highlightedSentence;
+      applyStatsWordAndSentenceCardFields(mediaFields, noteInfo, ankiConfig);
 
       if (audioBuffer) {
         const audioFilename = `subminer_audio_${timestamp}.mp3`;

src/core/services/tokenizer/annotation-stage.test.ts

@@ -123,6 +123,48 @@ test('annotateTokens falls back to reading for known-word matches when headword
   assert.equal(result[0]?.frequencyRank, 1895);
 });
 
+test('annotateTokens ignores partial furigana readings for known-word fallback', () => {
+  const tokens = [
+    makeToken({
+      surface: '待ち合わせてる',
+      headword: '待ち合わせる',
+      reading: 'まあ',
+      partOfSpeech: PartOfSpeech.verb,
+      endPos: 7,
+    }),
+  ];
+
+  const result = annotateTokens(
+    tokens,
+    makeDeps({
+      isKnownWord: (text) => text === 'まあ',
+    }),
+  );
+
+  assert.equal(result[0]?.isKnown, false);
+});
+
+test('annotateTokens reading fallback still matches kana surfaces with complete readings', () => {
+  const tokens = [
+    makeToken({
+      surface: 'ください',
+      headword: '下さい',
+      reading: 'ください',
+      partOfSpeech: PartOfSpeech.verb,
+      endPos: 4,
+    }),
+  ];
+
+  const result = annotateTokens(
+    tokens,
+    makeDeps({
+      isKnownWord: (text) => text === 'ください',
+    }),
+  );
+
+  assert.equal(result[0]?.isKnown, true);
+});
+
 test('annotateTokens excludes frequency for particle/bound_auxiliary and pos1 exclusions', () => {
   const tokens = [
     makeToken({

src/core/services/tokenizer/annotation-stage.ts

@@ -635,6 +635,32 @@ export function stripSubtitleAnnotationMetadata(
   return sharedStripSubtitleAnnotationMetadata(token, options);
 }
 
+// Furigana-derived readings can be partial (kanji readings only, e.g. まあ for
+// 待ち合わせてる); matching those against known words produces false positives,
+// so the reading fallback requires a reading that plausibly covers the surface:
+// at least as many characters as the surface, with the surface's kana appearing
+// in order within the reading.
+function isCompleteReadingForSurface(surface: string, reading: string): boolean {
+  const surfaceChars = [...normalizeJlptTextForExclusion(surface)];
+  const readingChars = [...normalizeJlptTextForExclusion(reading)];
+  if (readingChars.length < surfaceChars.length) {
+    return false;
+  }
+
+  let cursor = 0;
+  for (const char of surfaceChars) {
+    if (!isKanaChar(char)) {
+      continue;
+    }
+    const foundAt = readingChars.indexOf(char, cursor);
+    if (foundAt === -1) {
+      return false;
+    }
+    cursor = foundAt + 1;
+  }
+  return true;
+}
+
 function computeTokenKnownStatus(
   token: MergedToken,
   isKnownWord: (text: string) => boolean,
@@ -650,6 +676,10 @@ function computeTokenKnownStatus(
     return false;
   }
 
+  if (!isCompleteReadingForSurface(token.surface, normalizedReading)) {
+    return false;
+  }
+
   return normalizedReading !== matchText.trim() && isKnownWord(normalizedReading);
 }
 

src/core/services/tokenizer/yomitan-parser-runtime.test.ts

@@ -964,7 +964,7 @@ test('requestYomitanScanTokens extracts best frequency rank from selected termsF
   assert.deepEqual(result, [
     {
       surface: '潜み',
-      reading: 'ひそ',
+      reading: 'ひそみ',
       headword: '潜む',
       startPos: 0,
       endPos: 2,
@@ -974,6 +974,72 @@ test('requestYomitanScanTokens extracts best frequency rank from selected termsF
   ]);
 });
 
+test('requestYomitanScanTokens emits complete readings for kanji-kana compounds', async () => {
+  let scannerScript = '';
+  const deps = createDeps(async (script) => {
+    if (script.includes('termsFind')) {
+      scannerScript = script;
+      return [];
+    }
+    if (script.includes('optionsGetFull')) {
+      return {
+        profileCurrent: 0,
+        profiles: [
+          {
+            options: {
+              scanning: { length: 40 },
+              dictionaries: [{ name: 'JPDBv2㋕', enabled: true, id: 0 }],
+            },
+          },
+        ],
+      };
+    }
+    return null;
+  });
+
+  await requestYomitanScanTokens('待ち合わせてる', deps, {
+    error: () => undefined,
+  });
+
+  const result = await runInjectedYomitanScript(scannerScript, (action, params) => {
+    if (action !== 'termsFind') {
+      throw new Error(`unexpected action: ${action}`);
+    }
+
+    const text = (params as { text?: string } | undefined)?.text ?? '';
+    if (!text.startsWith('待ち合わせてる')) {
+      return { originalTextLength: 0, dictionaryEntries: [] };
+    }
+
+    return {
+      originalTextLength: 7,
+      dictionaryEntries: [
+        {
+          headwords: [
+            {
+              term: '待ち合わせる',
+              reading: 'まちあわせる',
+              sources: [{ originalText: '待ち合わせてる', isPrimary: true, matchType: 'exact' }],
+            },
+          ],
+        },
+      ],
+    };
+  });
+
+  assert.deepEqual(result, [
+    {
+      surface: '待ち合わせてる',
+      reading: 'まちあわせてる',
+      headword: '待ち合わせる',
+      startPos: 0,
+      endPos: 7,
+      isNameMatch: false,
+      frequencyRank: undefined,
+    },
+  ]);
+});
+
 test('requestYomitanScanTokens uses frequency from later exact-match entry when first exact entry has none', async () => {
   let scannerScript = '';
   const deps = createDeps(async (script) => {

src/core/services/tokenizer/yomitan-parser-runtime.ts

@@ -817,6 +817,12 @@ const YOMITAN_SCANNING_HELPERS = String.raw`
       function isCodePointKana(codePoint) { return isCodePointInRanges(codePoint, KANA_RANGES); }
       function isCodePointJapanese(codePoint) { return isCodePointInRanges(codePoint, JAPANESE_RANGES); }
       function createFuriganaSegment(text, reading) { return {text, reading}; }
+      function getSegmentReadingContribution(segment) {
+        if (typeof segment.reading === "string" && segment.reading.length > 0) { return segment.reading; }
+        const segmentText = typeof segment.text === "string" ? segment.text : "";
+        const isKanaOnly = segmentText.length > 0 && [...segmentText].every((char) => isCodePointKana(char.codePointAt(0)));
+        return isKanaOnly ? segmentText : "";
+      }
       function getProlongedHiragana(previousCharacter) {
         switch (previousCharacter) {
           case "あ": case "か": case "が": case "さ": case "ざ": case "た": case "だ": case "な": case "は": case "ば": case "ぱ": case "ま": case "や": case "ら": case "わ": case "ぁ": case "ゃ": case "ゎ": return "あ";
@@ -1310,7 +1316,7 @@ ${YOMITAN_SCANNING_HELPERS}
             const segments = distributeFuriganaInflected(preferredHeadword.term, reading, source);
             const tokenPayload = {
               surface: segments.map((segment) => segment.text).join("") || source,
-              reading: segments.map((segment) => typeof segment.reading === "string" ? segment.reading : "").join(""),
+              reading: segments.map(getSegmentReadingContribution).join(""),
               headword: preferredHeadword.term,
               startPos: i,
               endPos: i + originalTextLength,

src/core/services/youtube/media-cache.test.ts

@@ -0,0 +1,540 @@
+import assert from 'node:assert/strict';
+import { EventEmitter } from 'node:events';
+import * as fs from 'node:fs';
+import * as os from 'node:os';
+import * as path from 'node:path';
+import test from 'node:test';
+
+import { createYoutubeMediaCacheService } from './media-cache';
+
+class FakeYtDlpProcess extends EventEmitter {
+  killed = false;
+  stdout = new EventEmitter();
+  stderr = new EventEmitter();
+
+  kill(): boolean {
+    this.killed = true;
+    return true;
+  }
+}
+
+type SpawnCall = {
+  command: string;
+  args: string[];
+  options?: { stdio?: Array<'ignore' | 'pipe'> };
+};
+
+function makeTempCacheRoot(): string {
+  return fs.mkdtempSync(path.join(os.tmpdir(), 'subminer-youtube-media-cache-test-'));
+}
+
+test('YouTube media cache does nothing in direct mode', async () => {
+  const cacheRoot = makeTempCacheRoot();
+  const spawnCalls: SpawnCall[] = [];
+
+  try {
+    const cache = createYoutubeMediaCacheService({
+      cacheRoot,
+      getYtDlpCommand: () => 'yt-dlp',
+      spawn: (command, args) => {
+        spawnCalls.push({ command, args });
+        return new FakeYtDlpProcess();
+      },
+    });
+
+    cache.start('https://youtu.be/demo', { mode: 'direct' });
+
+    assert.deepEqual(spawnCalls, []);
+    assert.equal(await cache.getCachedMediaPath('https://youtu.be/demo'), null);
+  } finally {
+    fs.rmSync(cacheRoot, { recursive: true, force: true });
+  }
+});
+
+test('YouTube media cache exposes the downloaded file after the background job completes', async () => {
+  const cacheRoot = makeTempCacheRoot();
+  const spawnedProcesses: FakeYtDlpProcess[] = [];
+  const spawnCalls: SpawnCall[] = [];
+  const readyEvents: Array<{ url: string; path: string }> = [];
+  const startedEvents: Array<{ url: string }> = [];
+
+  try {
+    const cache = createYoutubeMediaCacheService({
+      cacheRoot,
+      getYtDlpCommand: () => 'yt-dlp',
+      onDownloadStarted: (event) => {
+        startedEvents.push(event);
+      },
+      onReady: (event) => {
+        readyEvents.push(event);
+      },
+      spawn: (command, args, options) => {
+        spawnCalls.push({ command, args, options });
+        const proc = new FakeYtDlpProcess();
+        spawnedProcesses.push(proc);
+        return proc;
+      },
+    });
+
+    cache.start('https://youtu.be/demo', { mode: 'background' });
+
+    assert.deepEqual(startedEvents, [{ url: 'https://youtu.be/demo' }]);
+    assert.equal(spawnCalls.length, 1);
+    assert.equal(spawnCalls[0]?.command, 'yt-dlp');
+    assert.ok(spawnCalls[0]?.args.includes('--no-playlist'));
+    assert.ok(spawnCalls[0]?.args.includes('--force-ipv4'));
+    assert.equal(spawnCalls[0]?.args[spawnCalls[0].args.indexOf('--retries') + 1], '5');
+    assert.equal(spawnCalls[0]?.args[spawnCalls[0].args.indexOf('--fragment-retries') + 1], '5');
+    assert.equal(spawnCalls[0]?.args[spawnCalls[0].args.indexOf('--extractor-retries') + 1], '5');
+    assert.ok(spawnCalls[0]?.args.includes('--merge-output-format'));
+    assert.equal(
+      spawnCalls[0]?.args[spawnCalls[0].args.indexOf('-f') + 1],
+      'bestvideo*[height<=720]+bestaudio/best[height<=720]',
+    );
+    assert.deepEqual(spawnCalls[0]?.options?.stdio, ['ignore', 'ignore', 'ignore']);
+    assert.equal(await cache.getCachedMediaPath('https://youtu.be/demo'), null);
+
+    const outputTemplate = spawnCalls[0]?.args[spawnCalls[0].args.indexOf('-o') + 1];
+    assert.equal(typeof outputTemplate, 'string');
+    const outputDir = path.dirname(outputTemplate!);
+    const outputPath = path.join(outputDir, 'media.mkv');
+    fs.mkdirSync(outputDir, { recursive: true });
+    fs.writeFileSync(outputPath, 'cached media');
+    spawnedProcesses[0]?.emit('close', 0);
+
+    await new Promise((resolve) => setImmediate(resolve));
+
+    assert.equal(await cache.getCachedMediaPath('https://youtu.be/demo'), outputPath);
+    assert.deepEqual(readyEvents, [{ url: 'https://youtu.be/demo', path: outputPath }]);
+  } finally {
+    fs.rmSync(cacheRoot, { recursive: true, force: true });
+  }
+});
+
+test('YouTube media cache clears the active cache when direct mode starts', async () => {
+  const cacheRoot = makeTempCacheRoot();
+  const spawnedProcesses: FakeYtDlpProcess[] = [];
+  const spawnCalls: SpawnCall[] = [];
+
+  try {
+    const cache = createYoutubeMediaCacheService({
+      cacheRoot,
+      getYtDlpCommand: () => 'yt-dlp',
+      spawn: (command, args, options) => {
+        spawnCalls.push({ command, args, options });
+        const proc = new FakeYtDlpProcess();
+        spawnedProcesses.push(proc);
+        return proc;
+      },
+    });
+
+    cache.start('https://youtu.be/background', { mode: 'background' });
+    const outputTemplate = spawnCalls[0]?.args[spawnCalls[0].args.indexOf('-o') + 1];
+    assert.equal(typeof outputTemplate, 'string');
+    const outputDir = path.dirname(outputTemplate!);
+    const outputPath = path.join(outputDir, 'media.mkv');
+    fs.mkdirSync(outputDir, { recursive: true });
+    fs.writeFileSync(outputPath, 'cached media');
+    spawnedProcesses[0]?.emit('close', 0);
+
+    await new Promise((resolve) => setImmediate(resolve));
+    assert.equal(await cache.getActiveCachedMediaPath(), outputPath);
+
+    cache.start('https://youtu.be/direct', { mode: 'direct' });
+
+    assert.equal(await cache.getActiveCachedMediaPath(), null);
+    assert.equal(await cache.getCachedMediaPath('https://youtu.be/background'), outputPath);
+  } finally {
+    fs.rmSync(cacheRoot, { recursive: true, force: true });
+  }
+});
+
+test('YouTube media cache cancels the active download when direct mode starts', async () => {
+  const cacheRoot = makeTempCacheRoot();
+  const spawnedProcesses: FakeYtDlpProcess[] = [];
+  const spawnCalls: SpawnCall[] = [];
+  const readyEvents: Array<{ url: string; path: string }> = [];
+
+  try {
+    const cache = createYoutubeMediaCacheService({
+      cacheRoot,
+      getYtDlpCommand: () => 'yt-dlp',
+      onReady: (event) => {
+        readyEvents.push(event);
+      },
+      spawn: (command, args, options) => {
+        spawnCalls.push({ command, args, options });
+        const proc = new FakeYtDlpProcess();
+        spawnedProcesses.push(proc);
+        return proc;
+      },
+    });
+
+    cache.start('https://youtu.be/background', { mode: 'background' });
+    const outputTemplate = spawnCalls[0]?.args[spawnCalls[0].args.indexOf('-o') + 1];
+    assert.equal(typeof outputTemplate, 'string');
+    const outputDir = path.dirname(outputTemplate!);
+    const outputPath = path.join(outputDir, 'media.mkv');
+    fs.mkdirSync(outputDir, { recursive: true });
+    fs.writeFileSync(outputPath, 'cached media');
+
+    cache.start('https://youtu.be/direct', { mode: 'direct' });
+    spawnedProcesses[0]?.emit('close', 0);
+
+    await new Promise((resolve) => setImmediate(resolve));
+
+    assert.equal(spawnedProcesses[0]?.killed, true);
+    assert.deepEqual(readyEvents, []);
+    assert.equal(await cache.getActiveCachedMediaPath(), null);
+    assert.equal(await cache.getCachedMediaPath('https://youtu.be/background'), null);
+  } finally {
+    fs.rmSync(cacheRoot, { recursive: true, force: true });
+  }
+});
+
+test('YouTube media cache reports failed background downloads', async () => {
+  const cacheRoot = makeTempCacheRoot();
+  const spawnedProcesses: FakeYtDlpProcess[] = [];
+  const failedEvents: Array<{ url: string }> = [];
+
+  try {
+    const cache = createYoutubeMediaCacheService({
+      cacheRoot,
+      getYtDlpCommand: () => 'yt-dlp',
+      onFailed: (event) => {
+        failedEvents.push(event);
+      },
+      spawn: () => {
+        const proc = new FakeYtDlpProcess();
+        spawnedProcesses.push(proc);
+        return proc;
+      },
+    });
+
+    cache.start('https://youtu.be/demo', { mode: 'background' });
+    spawnedProcesses[0]?.emit('close', 1);
+
+    await new Promise((resolve) => setImmediate(resolve));
+
+    assert.deepEqual(failedEvents, [{ url: 'https://youtu.be/demo' }]);
+    assert.equal(await cache.getCachedMediaPath('https://youtu.be/demo'), null);
+  } finally {
+    fs.rmSync(cacheRoot, { recursive: true, force: true });
+  }
+});
+
+test('YouTube media cache only reports a failed download once when error is followed by close', async () => {
+  const cacheRoot = makeTempCacheRoot();
+  const spawnedProcesses: FakeYtDlpProcess[] = [];
+  const failedEvents: Array<{ url: string }> = [];
+
+  try {
+    const cache = createYoutubeMediaCacheService({
+      cacheRoot,
+      getYtDlpCommand: () => 'yt-dlp',
+      onFailed: (event) => {
+        failedEvents.push(event);
+      },
+      spawn: () => {
+        const proc = new FakeYtDlpProcess();
+        spawnedProcesses.push(proc);
+        return proc;
+      },
+    });
+
+    cache.start('https://youtu.be/demo', { mode: 'background' });
+    spawnedProcesses[0]?.emit('error', new Error('spawn failed'));
+    spawnedProcesses[0]?.emit('close', 1);
+
+    await new Promise((resolve) => setImmediate(resolve));
+
+    assert.deepEqual(failedEvents, [{ url: 'https://youtu.be/demo' }]);
+  } finally {
+    fs.rmSync(cacheRoot, { recursive: true, force: true });
+  }
+});
+
+test('YouTube media cache can disable the download height cap', () => {
+  const cacheRoot = makeTempCacheRoot();
+  const spawnCalls: SpawnCall[] = [];
+
+  try {
+    const cache = createYoutubeMediaCacheService({
+      cacheRoot,
+      getYtDlpCommand: () => 'yt-dlp',
+      spawn: (command, args, options) => {
+        spawnCalls.push({ command, args, options });
+        return new FakeYtDlpProcess();
+      },
+    });
+
+    cache.start('https://youtu.be/demo', { mode: 'background', maxHeight: 0 });
+
+    assert.equal(spawnCalls.length, 1);
+    assert.equal(
+      spawnCalls[0]?.args[spawnCalls[0].args.indexOf('-f') + 1],
+      'bestvideo*+bestaudio/best',
+    );
+  } finally {
+    fs.rmSync(cacheRoot, { recursive: true, force: true });
+  }
+});
+
+test('YouTube media cache applies the configured download height cap', () => {
+  const cacheRoot = makeTempCacheRoot();
+  const spawnCalls: SpawnCall[] = [];
+
+  try {
+    const cache = createYoutubeMediaCacheService({
+      cacheRoot,
+      getYtDlpCommand: () => 'yt-dlp',
+      spawn: (command, args, options) => {
+        spawnCalls.push({ command, args, options });
+        return new FakeYtDlpProcess();
+      },
+    });
+
+    cache.start('https://youtu.be/demo', { mode: 'background', maxHeight: 480 });
+
+    assert.equal(spawnCalls.length, 1);
+    assert.equal(
+      spawnCalls[0]?.args[spawnCalls[0].args.indexOf('-f') + 1],
+      'bestvideo*[height<=480]+bestaudio/best[height<=480]',
+    );
+  } finally {
+    fs.rmSync(cacheRoot, { recursive: true, force: true });
+  }
+});
+
+test('YouTube media cache restarts ready sessions when the height cap changes', async () => {
+  const cacheRoot = makeTempCacheRoot();
+  const spawnedProcesses: FakeYtDlpProcess[] = [];
+  const spawnCalls: SpawnCall[] = [];
+
+  try {
+    const cache = createYoutubeMediaCacheService({
+      cacheRoot,
+      getYtDlpCommand: () => 'yt-dlp',
+      spawn: (command, args, options) => {
+        spawnCalls.push({ command, args, options });
+        const proc = new FakeYtDlpProcess();
+        spawnedProcesses.push(proc);
+        return proc;
+      },
+    });
+
+    cache.start('https://youtu.be/demo', { mode: 'background', maxHeight: 720 });
+    const firstOutputTemplate = spawnCalls[0]?.args[spawnCalls[0].args.indexOf('-o') + 1];
+    assert.equal(typeof firstOutputTemplate, 'string');
+    const firstOutputDir = path.dirname(firstOutputTemplate!);
+    const firstOutputPath = path.join(firstOutputDir, 'media.mkv');
+    fs.mkdirSync(firstOutputDir, { recursive: true });
+    fs.writeFileSync(firstOutputPath, 'cached media');
+    spawnedProcesses[0]?.emit('close', 0);
+
+    await new Promise((resolve) => setImmediate(resolve));
+    assert.equal(await cache.getCachedMediaPath('https://youtu.be/demo'), firstOutputPath);
+
+    cache.start('https://youtu.be/demo', { mode: 'background', maxHeight: 480 });
+
+    assert.equal(spawnCalls.length, 2);
+    assert.equal(fs.existsSync(firstOutputPath), false);
+    assert.equal(await cache.getCachedMediaPath('https://youtu.be/demo'), null);
+    assert.equal(
+      spawnCalls[1]?.args[spawnCalls[1].args.indexOf('-f') + 1],
+      'bestvideo*[height<=480]+bestaudio/best[height<=480]',
+    );
+  } finally {
+    fs.rmSync(cacheRoot, { recursive: true, force: true });
+  }
+});
+
+test('YouTube media cache restarts running sessions when the height cap changes', () => {
+  const cacheRoot = makeTempCacheRoot();
+  const spawnedProcesses: FakeYtDlpProcess[] = [];
+  const spawnCalls: SpawnCall[] = [];
+
+  try {
+    const cache = createYoutubeMediaCacheService({
+      cacheRoot,
+      getYtDlpCommand: () => 'yt-dlp',
+      spawn: (command, args, options) => {
+        spawnCalls.push({ command, args, options });
+        const proc = new FakeYtDlpProcess();
+        spawnedProcesses.push(proc);
+        return proc;
+      },
+    });
+
+    cache.start('https://youtu.be/demo', { mode: 'background', maxHeight: 720 });
+    cache.start('https://youtu.be/demo', { mode: 'background', maxHeight: 480 });
+
+    assert.equal(spawnedProcesses[0]?.killed, true);
+    assert.equal(spawnCalls.length, 2);
+    assert.equal(
+      spawnCalls[1]?.args[spawnCalls[1].args.indexOf('-f') + 1],
+      'bestvideo*[height<=480]+bestaudio/best[height<=480]',
+    );
+  } finally {
+    fs.rmSync(cacheRoot, { recursive: true, force: true });
+  }
+});
+
+test('YouTube media cache removes stale files from previous runs on startup', () => {
+  const cacheRoot = makeTempCacheRoot();
+  const staleDir = path.join(cacheRoot, 'stale-session');
+  const spawnCalls: SpawnCall[] = [];
+
+  try {
+    fs.mkdirSync(staleDir, { recursive: true });
+    fs.writeFileSync(path.join(staleDir, 'media.mkv'), 'stale cached media');
+
+    createYoutubeMediaCacheService({
+      cacheRoot,
+      getYtDlpCommand: () => 'yt-dlp',
+      spawn: (command, args) => {
+        spawnCalls.push({ command, args });
+        return new FakeYtDlpProcess();
+      },
+    });
+
+    assert.equal(fs.existsSync(staleDir), false);
+    assert.deepEqual(spawnCalls, []);
+  } finally {
+    fs.rmSync(cacheRoot, { recursive: true, force: true });
+  }
+});
+
+test('YouTube media cache restarts when a ready cached file was deleted externally', async () => {
+  const cacheRoot = makeTempCacheRoot();
+  const spawnedProcesses: FakeYtDlpProcess[] = [];
+  const spawnCalls: Array<{ command: string; args: string[] }> = [];
+
+  try {
+    const cache = createYoutubeMediaCacheService({
+      cacheRoot,
+      getYtDlpCommand: () => 'yt-dlp',
+      spawn: (command, args) => {
+        spawnCalls.push({ command, args });
+        const proc = new FakeYtDlpProcess();
+        spawnedProcesses.push(proc);
+        return proc;
+      },
+    });
+
+    cache.start('https://youtu.be/demo', { mode: 'background' });
+    const outputTemplate = spawnCalls[0]?.args[spawnCalls[0].args.indexOf('-o') + 1];
+    assert.equal(typeof outputTemplate, 'string');
+    const outputDir = path.dirname(outputTemplate!);
+    const outputPath = path.join(outputDir, 'media.mkv');
+    fs.mkdirSync(outputDir, { recursive: true });
+    fs.writeFileSync(outputPath, 'cached media');
+    spawnedProcesses[0]?.emit('close', 0);
+
+    await new Promise((resolve) => setImmediate(resolve));
+    assert.equal(await cache.getCachedMediaPath('https://youtu.be/demo'), outputPath);
+
+    fs.rmSync(outputPath);
+    cache.start('https://youtu.be/demo', { mode: 'background' });
+
+    assert.equal(spawnCalls.length, 2);
+    assert.equal(await cache.getCachedMediaPath('https://youtu.be/demo'), null);
+  } finally {
+    fs.rmSync(cacheRoot, { recursive: true, force: true });
+  }
+});
+
+test('YouTube media cache removes stale disk siblings before starting a new cache', () => {
+  const cacheRoot = makeTempCacheRoot();
+  const staleDir = path.join(cacheRoot, 'stale-sibling');
+  const spawnCalls: Array<{ command: string; args: string[] }> = [];
+
+  try {
+    const cache = createYoutubeMediaCacheService({
+      cacheRoot,
+      getYtDlpCommand: () => 'yt-dlp',
+      spawn: (command, args) => {
+        spawnCalls.push({ command, args });
+        return new FakeYtDlpProcess();
+      },
+    });
+    fs.mkdirSync(staleDir, { recursive: true });
+    fs.writeFileSync(path.join(staleDir, 'media.mkv'), 'stale cached media');
+
+    cache.start('https://youtu.be/demo', { mode: 'background' });
+
+    assert.equal(fs.existsSync(staleDir), false);
+    assert.equal(spawnCalls.length, 1);
+  } finally {
+    fs.rmSync(cacheRoot, { recursive: true, force: true });
+  }
+});
+
+test('YouTube media cache drops old sessions when a new background cache starts', async () => {
+  const cacheRoot = makeTempCacheRoot();
+  const spawnedProcesses: FakeYtDlpProcess[] = [];
+  const spawnCalls: Array<{ command: string; args: string[] }> = [];
+
+  try {
+    const cache = createYoutubeMediaCacheService({
+      cacheRoot,
+      getYtDlpCommand: () => 'yt-dlp',
+      spawn: (command, args) => {
+        spawnCalls.push({ command, args });
+        const proc = new FakeYtDlpProcess();
+        spawnedProcesses.push(proc);
+        return proc;
+      },
+    });
+
+    cache.start('https://youtu.be/first', { mode: 'background' });
+    const firstOutputTemplate = spawnCalls[0]?.args[spawnCalls[0].args.indexOf('-o') + 1];
+    assert.equal(typeof firstOutputTemplate, 'string');
+    const firstOutputDir = path.dirname(firstOutputTemplate!);
+    fs.mkdirSync(firstOutputDir, { recursive: true });
+    fs.writeFileSync(path.join(firstOutputDir, 'media.mkv'), 'cached media');
+
+    cache.start('https://youtu.be/second', { mode: 'background' });
+
+    assert.equal(spawnedProcesses[0]?.killed, true);
+    assert.equal(fs.existsSync(firstOutputDir), false);
+    assert.equal(await cache.getCachedMediaPath('https://youtu.be/first'), null);
+    assert.equal(spawnCalls.length, 2);
+  } finally {
+    fs.rmSync(cacheRoot, { recursive: true, force: true });
+  }
+});
+
+test('YouTube media cache cleanup kills downloads and removes temp files', async () => {
+  const cacheRoot = makeTempCacheRoot();
+  const spawnedProcesses: FakeYtDlpProcess[] = [];
+  const spawnCalls: Array<{ command: string; args: string[] }> = [];
+
+  try {
+    const cache = createYoutubeMediaCacheService({
+      cacheRoot,
+      getYtDlpCommand: () => 'yt-dlp',
+      spawn: (command, args) => {
+        spawnCalls.push({ command, args });
+        const proc = new FakeYtDlpProcess();
+        spawnedProcesses.push(proc);
+        return proc;
+      },
+    });
+
+    cache.start('https://youtu.be/demo', { mode: 'background' });
+    const outputTemplate = spawnCalls[0]?.args[spawnCalls[0].args.indexOf('-o') + 1];
+    assert.equal(typeof outputTemplate, 'string');
+    const outputDir = path.dirname(outputTemplate!);
+    fs.mkdirSync(outputDir, { recursive: true });
+    fs.writeFileSync(path.join(outputDir, 'media.mkv'), 'cached media');
+
+    cache.cleanup();
+
+    assert.equal(spawnedProcesses[0]?.killed, true);
+    assert.equal(fs.existsSync(outputDir), false);
+    assert.equal(await cache.getCachedMediaPath('https://youtu.be/demo'), null);
+  } finally {
+    fs.rmSync(cacheRoot, { recursive: true, force: true });
+  }
+});

src/core/services/youtube/media-cache.ts

@@ -0,0 +1,307 @@
+import { spawn as spawnProcess } from 'node:child_process';
+import * as crypto from 'node:crypto';
+import { EventEmitter } from 'node:events';
+import * as fs from 'node:fs';
+import * as os from 'node:os';
+import * as path from 'node:path';
+
+import type { YoutubeMediaCacheMode } from '../../../types/integrations';
+import { getYoutubeYtDlpCommand } from './ytdlp-command';
+
+type MediaCacheSessionState = 'running' | 'ready' | 'failed';
+
+type SpawnedProcess = EventEmitter & {
+  killed?: boolean;
+  kill?: (signal?: NodeJS.Signals | number) => boolean;
+};
+
+type SpawnProcess = (
+  command: string,
+  args: string[],
+  options?: { stdio?: Array<'ignore' | 'pipe'> },
+) => SpawnedProcess;
+
+interface MediaCacheSession {
+  url: string;
+  dir: string;
+  maxHeight: number;
+  process: SpawnedProcess | null;
+  readyPath: string | null;
+  state: MediaCacheSessionState;
+}
+
+export interface YoutubeMediaCacheStartOptions {
+  mode: YoutubeMediaCacheMode;
+  maxHeight?: number;
+}
+
+export interface YoutubeMediaCacheServiceDeps {
+  cacheRoot?: string;
+  getYtDlpCommand?: () => string;
+  spawn?: SpawnProcess;
+  onDownloadStarted?: (event: { url: string }) => void;
+  onReady?: (event: { url: string; path: string }) => void;
+  onFailed?: (event: { url: string }) => void;
+  logInfo?: (message: string) => void;
+  logWarn?: (message: string) => void;
+}
+
+const MEDIA_FILE_EXTENSIONS = new Set(['.mkv', '.mp4', '.webm', '.m4a', '.mp3', '.opus']);
+const DEFAULT_MAX_HEIGHT = 720;
+
+function cacheKeyForUrl(url: string): string {
+  return crypto.createHash('sha256').update(url).digest('hex').slice(0, 24);
+}
+
+function isFinalMediaFile(fileName: string): boolean {
+  if (!fileName.startsWith('media.')) {
+    return false;
+  }
+  if (fileName.endsWith('.part') || fileName.endsWith('.ytdl') || fileName.endsWith('.tmp')) {
+    return false;
+  }
+  return MEDIA_FILE_EXTENSIONS.has(path.extname(fileName).toLowerCase());
+}
+
+function findReadyMediaPath(dir: string): string | null {
+  try {
+    const files = fs.readdirSync(dir);
+    const mediaFile = files.find(isFinalMediaFile);
+    return mediaFile ? path.join(dir, mediaFile) : null;
+  } catch {
+    return null;
+  }
+}
+
+function getFormatSelector(maxHeight: number): string {
+  return maxHeight > 0
+    ? `bestvideo*[height<=${maxHeight}]+bestaudio/best[height<=${maxHeight}]`
+    : 'bestvideo*+bestaudio/best';
+}
+
+function normalizeMaxHeight(maxHeight: number | undefined): number {
+  if (maxHeight === undefined) {
+    return DEFAULT_MAX_HEIGHT;
+  }
+  return Number.isInteger(maxHeight) && maxHeight >= 0 ? maxHeight : DEFAULT_MAX_HEIGHT;
+}
+
+function createYtDlpArgs(url: string, outputTemplate: string, maxHeight?: number): string[] {
+  return [
+    '--no-playlist',
+    '--no-warnings',
+    '--force-ipv4',
+    '--retries',
+    '5',
+    '--fragment-retries',
+    '5',
+    '--extractor-retries',
+    '5',
+    '-f',
+    getFormatSelector(normalizeMaxHeight(maxHeight)),
+    '--merge-output-format',
+    'mkv',
+    '-o',
+    outputTemplate,
+    url,
+  ];
+}
+
+export function createYoutubeMediaCacheService(deps: YoutubeMediaCacheServiceDeps = {}) {
+  const cacheRoot = deps.cacheRoot ?? path.join(os.tmpdir(), 'subminer-youtube-media-cache');
+  const getYtDlpCommand = deps.getYtDlpCommand ?? getYoutubeYtDlpCommand;
+  const spawn: SpawnProcess =
+    deps.spawn ??
+    ((command, args, options) =>
+      spawnProcess(command, args, options ?? {}) as unknown as SpawnedProcess);
+  const sessions = new Map<string, MediaCacheSession>();
+  let activeKey: string | null = null;
+
+  const getSessionDir = (url: string): string => path.join(cacheRoot, cacheKeyForUrl(url));
+  const removeCacheDir = (dir: string): void => {
+    try {
+      fs.rmSync(dir, { recursive: true, force: true });
+    } catch {
+      // Temp cache cleanup should not block shutdown or playback startup.
+    }
+  };
+  const removeCacheRootEntriesExcept = (dirsToKeep: string[]): void => {
+    let entries: fs.Dirent[];
+    try {
+      entries = fs.readdirSync(cacheRoot, { withFileTypes: true });
+    } catch {
+      return;
+    }
+    const keepDirs = new Set(dirsToKeep.map((dir) => path.resolve(dir)));
+    for (const entry of entries) {
+      const entryPath = path.join(cacheRoot, entry.name);
+      if (keepDirs.has(path.resolve(entryPath))) {
+        continue;
+      }
+      removeCacheDir(entryPath);
+    }
+  };
+  const removeSession = (key: string): void => {
+    const session = sessions.get(key);
+    if (!session) {
+      return;
+    }
+    if (session.state === 'running' && session.process?.kill && !session.process.killed) {
+      session.process.kill();
+    }
+    sessions.delete(key);
+    if (activeKey === key) {
+      activeKey = null;
+    }
+    removeCacheDir(session.dir);
+  };
+  const removeInactiveSessions = (keyToKeep: string): void => {
+    for (const key of [...sessions.keys()]) {
+      if (key !== keyToKeep) {
+        removeSession(key);
+      }
+    }
+  };
+  removeCacheRootEntriesExcept([]);
+
+  const getCachedMediaPath = async (url: string): Promise<string | null> => {
+    const key = cacheKeyForUrl(url);
+    const session = sessions.get(key);
+    if (session?.readyPath && fs.existsSync(session.readyPath)) {
+      return session.readyPath;
+    }
+
+    const readyPath = findReadyMediaPath(session?.dir ?? getSessionDir(url));
+    if (readyPath) {
+      sessions.set(key, {
+        url,
+        dir: path.dirname(readyPath),
+        maxHeight: session?.maxHeight ?? DEFAULT_MAX_HEIGHT,
+        process: null,
+        readyPath,
+        state: 'ready',
+      });
+      return readyPath;
+    }
+
+    return null;
+  };
+
+  const getActiveCachedMediaPath = async (): Promise<string | null> => {
+    if (!activeKey) {
+      return null;
+    }
+    const session = sessions.get(activeKey);
+    return session ? getCachedMediaPath(session.url) : null;
+  };
+
+  const start = (url: string, options: YoutubeMediaCacheStartOptions): void => {
+    if (options.mode !== 'background') {
+      if (activeKey) {
+        const activeSession = sessions.get(activeKey);
+        if (activeSession?.state === 'running') {
+          removeSession(activeKey);
+        }
+      }
+      activeKey = null;
+      return;
+    }
+
+    const key = cacheKeyForUrl(url);
+    const maxHeight = normalizeMaxHeight(options.maxHeight);
+    activeKey = key;
+    const dir = getSessionDir(url);
+    const existingSession = sessions.get(key);
+    const canReuseExistingSession = existingSession?.maxHeight === maxHeight;
+    if (existingSession?.state === 'running' && canReuseExistingSession) {
+      removeInactiveSessions(key);
+      removeCacheRootEntriesExcept([existingSession.dir]);
+      return;
+    }
+    if (existingSession) {
+      if (
+        canReuseExistingSession &&
+        existingSession.state === 'ready' &&
+        ((existingSession.readyPath && fs.existsSync(existingSession.readyPath)) ||
+          findReadyMediaPath(existingSession.dir))
+      ) {
+        removeInactiveSessions(key);
+        removeCacheRootEntriesExcept([existingSession.dir]);
+        return;
+      }
+      removeSession(key);
+      activeKey = key;
+    }
+    removeInactiveSessions(key);
+    removeCacheRootEntriesExcept([dir]);
+
+    fs.mkdirSync(dir, { recursive: true });
+    const outputTemplate = path.join(dir, 'media.%(ext)s');
+    const args = createYtDlpArgs(url, outputTemplate, maxHeight);
+    const child = spawn(getYtDlpCommand(), args, { stdio: ['ignore', 'ignore', 'ignore'] });
+    const session: MediaCacheSession = {
+      url,
+      dir,
+      maxHeight,
+      process: child,
+      readyPath: null,
+      state: 'running',
+    };
+    sessions.set(key, session);
+    deps.logInfo?.(`Started YouTube media cache download for ${url}`);
+    deps.onDownloadStarted?.({ url });
+
+    child.once('error', (error) => {
+      const currentSession = sessions.get(key);
+      if (currentSession !== session) {
+        return;
+      }
+      session.state = 'failed';
+      session.process = null;
+      deps.logWarn?.(
+        `YouTube media cache download failed: ${
+          error instanceof Error ? error.message : String(error)
+        }`,
+      );
+      deps.onFailed?.({ url });
+    });
+
+    child.once('close', (code) => {
+      const currentSession = sessions.get(key);
+      if (currentSession !== session) {
+        return;
+      }
+      if (session.state === 'failed') {
+        return;
+      }
+      session.process = null;
+      if (code === 0) {
+        const readyPath = findReadyMediaPath(dir);
+        if (readyPath) {
+          session.state = 'ready';
+          session.readyPath = readyPath;
+          deps.logInfo?.(`YouTube media cache ready at ${readyPath}`);
+          deps.onReady?.({ url, path: readyPath });
+          return;
+        }
+      }
+      session.state = 'failed';
+      deps.logWarn?.(`YouTube media cache download exited without a usable media file.`);
+      deps.onFailed?.({ url });
+    });
+  };
+
+  const cleanup = (): void => {
+    for (const key of [...sessions.keys()]) {
+      removeSession(key);
+    }
+    activeKey = null;
+  };
+
+  return {
+    cleanup,
+    getActiveCachedMediaPath,
+    getCachedMediaPath,
+    start,
+  };
+}

src/logger.test.ts

@@ -2,9 +2,10 @@ import test from 'node:test';
 import assert from 'node:assert/strict';
 import path from 'node:path';
 import { resolveDefaultLogFilePath, setLogRotation } from './logger';
+import { localDateKey } from './shared/log-files';
 
 test('resolveDefaultLogFilePath uses APPDATA on windows', () => {
-  const today = new Date().toISOString().slice(0, 10);
+  const today = localDateKey(new Date());
   const resolved = resolveDefaultLogFilePath({
     platform: 'win32',
     homeDir: 'C:\\Users\\tester',
@@ -20,7 +21,7 @@ test('resolveDefaultLogFilePath uses APPDATA on windows', () => {
 });
 
 test('resolveDefaultLogFilePath uses .config on linux', () => {
-  const today = new Date().toISOString().slice(0, 10);
+  const today = localDateKey(new Date());
   const resolved = resolveDefaultLogFilePath({
     platform: 'linux',
     homeDir: '/home/tester',
@@ -34,7 +35,7 @@ test('resolveDefaultLogFilePath uses .config on linux', () => {
 
 test('setLogRotation accepts numeric retention days', () => {
   const previous = process.env.SUBMINER_LOG_ROTATION;
-  const today = new Date().toISOString().slice(0, 10);
+  const today = localDateKey(new Date());
   setLogRotation(14);
   try {
     const resolved = resolveDefaultLogFilePath({

src/main.ts

@@ -332,6 +332,7 @@ import {
   acquireYoutubeSubtitleTrack,
   acquireYoutubeSubtitleTracks,
 } from './core/services/youtube/generate';
+import { createYoutubeMediaCacheService } from './core/services/youtube/media-cache';
 import { resolveYoutubePlaybackUrl } from './core/services/youtube/playback-resolve';
 import { probeYoutubeTracks } from './core/services/youtube/track-probe';
 import {
@@ -347,6 +348,7 @@ import {
   shouldAutoOpenFirstRunSetup,
 } from './main/runtime/first-run-setup-service';
 import { createYoutubeFlowRuntime } from './main/runtime/youtube-flow';
+import { createYoutubeMediaCachePlaybackRuntime } from './main/runtime/youtube-media-cache-playback';
 import { createYoutubePlaybackRuntime } from './main/runtime/youtube-playback-runtime';
 import {
   clearYoutubePrimarySubtitleNotificationTimer,
@@ -532,7 +534,6 @@ import { shouldSuppressVisibleOverlayRaiseForSeparateWindow } from './main/runti
 import {
   isSameYoutubeMediaPath,
   isYoutubeMediaPath,
-  isYoutubePlaybackActive,
   shouldUseCachedYoutubeParsedCues,
 } from './main/runtime/youtube-playback';
 import { createYomitanProfilePolicy } from './main/runtime/yomitan-profile-policy';
@@ -1172,6 +1173,65 @@ const prepareYoutubePlaybackInMpv = createPrepareYoutubePlaybackInMpvHandler({
   },
   wait: (ms) => new Promise((resolve) => setTimeout(resolve, ms)),
 });
+const youtubeMediaCache = createYoutubeMediaCacheService({
+  onDownloadStarted: (event) => {
+    showConfiguredStatusNotification('YouTube media cache is downloading.', {
+      id: 'youtube-media-cache-status',
+      title: 'YouTube media cache',
+      variant: 'progress',
+      persistent: true,
+    });
+    logger.info(`YouTube media cache download notification shown for ${event.url}`);
+  },
+  onReady: (event) => {
+    showConfiguredStatusNotification('YouTube media cache ready.', {
+      id: 'youtube-media-cache-status',
+      title: 'YouTube media cache',
+      variant: 'success',
+      persistent: false,
+    });
+    void appState.ankiIntegration
+      ?.handleYoutubeMediaCacheReady(event.url, event.path, { notifyNoQueued: false })
+      .catch((error) => {
+        logger.warn(
+          `Failed to apply queued YouTube media updates: ${
+            error instanceof Error ? error.message : String(error)
+          }`,
+        );
+      });
+  },
+  onFailed: (event) => {
+    showConfiguredStatusNotification('YouTube media cache failed.', {
+      id: 'youtube-media-cache-status',
+      title: 'YouTube media cache',
+      variant: 'error',
+      persistent: false,
+    });
+    void appState.ankiIntegration
+      ?.handleYoutubeMediaCacheFailed(event.url, { notifyStatus: false })
+      .catch((error) => {
+        logger.warn(
+          `Failed to drain queued YouTube media updates after cache failure: ${
+            error instanceof Error ? error.message : String(error)
+          }`,
+        );
+      });
+  },
+  logInfo: (message) => logger.info(message),
+  logWarn: (message) => logger.warn(message),
+});
+const youtubeMediaCachePlaybackRuntime = createYoutubeMediaCachePlaybackRuntime({
+  getMediaCacheConfig: () => getResolvedConfig().youtube.mediaCache,
+  requestMpvProperty: async (name) => {
+    const client = appState.mpvClient;
+    if (!client) return null;
+    return await client.requestProperty(name);
+  },
+  startYoutubeMediaCache: (url, options) => {
+    youtubeMediaCache.start(url, options);
+  },
+  logWarn: (message) => logger.warn(message),
+});
 const waitForYoutubeMpvConnected = createWaitForMpvConnectedHandler({
   getMpvClient: () => appState.mpvClient,
   now: () => Date.now(),
@@ -1316,6 +1376,12 @@ const youtubePlaybackRuntime = createYoutubePlaybackRuntime({
   },
   waitForYoutubeMpvConnected: (timeoutMs) => waitForYoutubeMpvConnected(timeoutMs),
   prepareYoutubePlaybackInMpv: (request) => prepareYoutubePlaybackInMpv(request),
+  startYoutubeMediaCache: (url) => {
+    youtubeMediaCache.start(url, {
+      mode: getResolvedConfig().youtube.mediaCache.mode,
+      maxHeight: getResolvedConfig().youtube.mediaCache.maxHeight,
+    });
+  },
   runYoutubePlaybackFlow: (request) => youtubeFlowRuntime.runYoutubePlaybackFlow(request),
   logInfo: (message) => logger.info(message),
   logWarn: (message) => logger.warn(message),
@@ -1629,9 +1695,61 @@ const youtubePrimarySubtitleNotificationRuntime = createYoutubePrimarySubtitleNo
 });
 
 function isYoutubePlaybackActiveNow(): boolean {
-  return isYoutubePlaybackActive(
-    appState.currentMediaPath,
-    appState.mpvClient?.currentVideoPath ?? null,
+  return Boolean(getCurrentYoutubeMediaCacheSourceUrlSnapshot());
+}
+
+function getCurrentYoutubeMediaCacheSourceUrlSnapshot(): string | null {
+  const currentMediaPath = appState.currentMediaPath?.trim() || null;
+  if (isYoutubeMediaPath(currentMediaPath)) {
+    return currentMediaPath;
+  }
+
+  const currentVideoPath = appState.mpvClient?.currentVideoPath?.trim() || null;
+  if (isYoutubeMediaPath(currentVideoPath)) {
+    return currentVideoPath;
+  }
+
+  return youtubeMediaCachePlaybackRuntime.getActiveYoutubeSourceUrlSnapshot();
+}
+
+async function getCurrentYoutubeMediaCacheSourceUrl(): Promise<string | null> {
+  const currentMediaPath = appState.currentMediaPath?.trim() || null;
+  if (isYoutubeMediaPath(currentMediaPath)) {
+    return currentMediaPath;
+  }
+
+  const currentVideoPath = appState.mpvClient?.currentVideoPath?.trim() || null;
+  if (isYoutubeMediaPath(currentVideoPath)) {
+    return currentVideoPath;
+  }
+
+  return await youtubeMediaCachePlaybackRuntime.getActiveYoutubeSourceUrl();
+}
+
+function shouldRequireYoutubeMediaCacheForCurrentPlayback(): boolean {
+  return (
+    getResolvedConfig().youtube.mediaCache.mode === 'background' && isYoutubePlaybackActiveNow()
+  );
+}
+
+async function getCachedYoutubeMediaPathForCurrentPlayback(
+  currentVideoPath: string,
+  _kind: 'audio' | 'video',
+): Promise<string | null> {
+  if (getResolvedConfig().youtube.mediaCache.mode !== 'background') {
+    return null;
+  }
+  const cacheSourceUrl = isYoutubeMediaPath(currentVideoPath)
+    ? currentVideoPath
+    : await getCurrentYoutubeMediaCacheSourceUrl();
+  if (!cacheSourceUrl) {
+    return null;
+  }
+  // mpv can expose the resolved stream URL here while the cache key uses the original page URL.
+  // Keep the active-cache fallback so current playback can still resolve the ready cached file.
+  return (
+    (await youtubeMediaCache.getCachedMediaPath(cacheSourceUrl)) ??
+    (await youtubeMediaCache.getActiveCachedMediaPath())
   );
 }
 
@@ -2244,6 +2362,7 @@ const mediaRuntime = createMediaRuntimeService(
 const characterDictionaryRuntime = createCharacterDictionaryRuntimeService({
   userDataPath: USER_DATA_PATH,
   getCurrentMediaPath: () => appState.currentMediaPath,
+  getCurrentVideoPath: () => appState.mpvClient?.currentVideoPath,
   getCurrentMediaTitle: () => appState.currentMediaTitle,
   resolveMediaPathForJimaku: (mediaPath) => mediaRuntime.resolveMediaPathForJimaku(mediaPath),
   guessAnilistMediaInfo: (mediaPath, mediaTitle) => guessAnilistMediaInfo(mediaPath, mediaTitle),
@@ -2561,6 +2680,10 @@ function clearWindowsVisibleOverlayForegroundPollLoop(): void {
   visibleOverlayInteractionRuntime.clearWindowsVisibleOverlayForegroundPollLoop();
 }
 
+function tickWindowsOverlayPointerInteractionNow(): void {
+  visibleOverlayInteractionRuntime.tickWindowsOverlayPointerInteractionNow();
+}
+
 function scheduleVisibleOverlayBlurRefresh(): void {
   visibleOverlayInteractionRuntime.scheduleVisibleOverlayBlurRefresh();
 }
@@ -2631,6 +2754,7 @@ const overlayNotificationsRuntime = createOverlayNotificationsRuntime({
 });
 const {
   flushQueuedOverlayNotifications,
+  flushQueuedMpvOsdNotifications,
   openAnkiCardFromNotification,
   toggleNotificationHistoryPanel,
   showConfiguredPlaybackFeedback,
@@ -3796,6 +3920,7 @@ const {
     },
     stopJellyfinRemoteSession: () => stopJellyfinRemoteSession(),
     cleanupYoutubeSubtitleTempDirs: () => youtubeFlowRuntime.cleanupSubtitleTempDirs(),
+    cleanupYoutubeMediaCache: () => youtubeMediaCache.cleanup(),
     cleanupJellyfinSubtitleCache: () => cleanupJellyfinSubtitleCache(),
     stopDiscordPresenceService: () => {
       void appState.discordPresenceService?.stop();
@@ -4281,6 +4406,7 @@ const {
     },
     onMpvConnected: () => {
       maybeStartOverlayLoadingOsd();
+      flushQueuedMpvOsdNotifications();
       if (appState.sessionBindingsInitialized) {
         sendMpvCommandRuntime(appState.mpvClient, [
           'script-message',
@@ -4356,6 +4482,7 @@ const {
         subtitlePrefetchRuntime.cancelPendingInit();
       }
       youtubePrimarySubtitleNotificationRuntime.handleMediaPathChange(path);
+      void youtubeMediaCachePlaybackRuntime.handleMediaPathChange(path);
       if (path) {
         ensureImmersionTrackerStarted();
         void subtitlePrefetchRuntime.refreshSubtitlePrefetchFromActiveTrack();
@@ -5408,13 +5535,15 @@ const { registerIpcRuntimeHandlers } = composeIpcRuntimeHandlers({
         if (!mainWindow || senderWindow !== mainWindow) {
           return;
         }
-        if (visibleOverlayInteractionRuntime.getVisibleOverlayInteractionActive() === active) {
+        const previousActive =
+          visibleOverlayInteractionRuntime.getVisibleOverlayInteractionActive();
+        visibleOverlayInteractionRuntime.setVisibleOverlayInteractionActive(active);
+        if (previousActive === active) {
           if (active && process.platform === 'darwin' && !mainWindow.isFocused()) {
             overlayVisibilityRuntime.updateVisibleOverlayVisibility();
           }
           return;
         }
-        visibleOverlayInteractionRuntime.setVisibleOverlayInteractionActive(active);
         overlayVisibilityRuntime.updateVisibleOverlayVisibility();
       },
       onOverlayInteractiveHint: (interactive, senderWindow) => {
@@ -5614,6 +5743,7 @@ const { registerIpcRuntimeHandlers } = composeIpcRuntimeHandlers({
       reportOverlayContentBounds: (payload: unknown) => {
         if (overlayContentMeasurementStore.report(payload)) {
           tickLinuxOverlayPointerInteractionNow();
+          tickWindowsOverlayPointerInteractionNow();
           primeLinuxOverlayPointerInteractionAfterFirstMeasurement();
           autoplayReadyGate.flushPendingAutoplayReadySignal();
           scheduleVisibleOverlaySubtitleRefreshAfterFirstPaint();
@@ -5723,6 +5853,10 @@ const { registerIpcRuntimeHandlers } = composeIpcRuntimeHandlers({
         );
       },
       getKnownWordCacheStatePath: () => path.join(USER_DATA_PATH, 'known-words-cache.json'),
+      getCachedMediaPath: (currentVideoPath, kind) =>
+        getCachedYoutubeMediaPathForCurrentPlayback(currentVideoPath, kind),
+      shouldRequireRemoteMediaCache: () => shouldRequireYoutubeMediaCacheForCurrentPlayback(),
+      getYoutubeMediaSourceUrl: () => getCurrentYoutubeMediaCacheSourceUrl(),
       showDesktopNotification,
       showOverlayNotification,
       createFieldGroupingCallback: () => createFieldGroupingCallback(),
@@ -6199,6 +6333,10 @@ const { initializeOverlayRuntime: initializeOverlayRuntimeHandler } =
       showOverlayNotification,
       createFieldGroupingCallback: () => createFieldGroupingCallback(),
       getKnownWordCacheStatePath: () => path.join(USER_DATA_PATH, 'known-words-cache.json'),
+      getCachedMediaPath: (currentVideoPath, kind) =>
+        getCachedYoutubeMediaPathForCurrentPlayback(currentVideoPath, kind),
+      shouldRequireRemoteMediaCache: () => shouldRequireYoutubeMediaCacheForCurrentPlayback(),
+      getYoutubeMediaSourceUrl: () => getCurrentYoutubeMediaCacheSourceUrl(),
       shouldStartAnkiIntegration: () =>
         !(appState.initialArgs && isHeadlessInitialCommand(appState.initialArgs)),
     },

src/main/character-dictionary-runtime.test.ts

@@ -2386,6 +2386,36 @@ test('buildMergedDictionary rebuilds snapshots written with an older format vers
   }
 });
 
+test('getManualSelectionSnapshot falls back to mpv current video path when app media path is not ready', async () => {
+  const userDataPath = makeTempDir();
+  const mpvPath =
+    'C:\\Videos\\KonoSuba - God’s blessing on this wonderful world!! (2016) - S02E05.mkv';
+  const calls: Array<{ mediaPath: string | null; mediaTitle: string | null }> = [];
+  const runtime = createCharacterDictionaryRuntimeService({
+    userDataPath,
+    getCurrentMediaPath: () => null,
+    getCurrentVideoPath: () => mpvPath,
+    getCurrentMediaTitle: () => null,
+    resolveMediaPathForJimaku: (mediaPath) => mediaPath,
+    guessAnilistMediaInfo: async (mediaPath, mediaTitle) => {
+      calls.push({ mediaPath, mediaTitle });
+      return {
+        title: 'KonoSuba - God’s blessing on this wonderful world!!',
+        season: 2,
+        episode: 5,
+        source: 'fallback',
+      };
+    },
+    now: () => 1_700_000_000_000,
+  });
+
+  const snapshot = await runtime.getManualSelectionSnapshot(undefined, '');
+
+  assert.deepEqual(calls, [{ mediaPath: mpvPath, mediaTitle: null }]);
+  assert.equal(snapshot.guessTitle, 'KonoSuba - God’s blessing on this wonderful world!!');
+  assert.equal(snapshot.candidates.length, 0);
+});
+
 test('buildMergedDictionary reapplies collapsible open states from current config', async () => {
   const userDataPath = makeTempDir();
   const originalFetch = globalThis.fetch;

src/main/character-dictionary-runtime.ts

@@ -76,6 +76,11 @@ function expandUserPath(input: string): string {
   return input;
 }
 
+function trimToNull(input: string | null | undefined): string | null {
+  const trimmed = typeof input === 'string' ? input.trim() : '';
+  return trimmed.length > 0 ? trimmed : null;
+}
+
 function isVideoFile(filePath: string): boolean {
   return hasVideoExtension(path.extname(filePath));
 }
@@ -195,8 +200,9 @@ export function createCharacterDictionaryRuntimeService(deps: CharacterDictionar
     return dictionaryTarget.length > 0
       ? resolveDictionaryGuessInputs(dictionaryTarget)
       : {
-          mediaPath: deps.getCurrentMediaPath(),
-          mediaTitle: deps.getCurrentMediaTitle(),
+          mediaPath:
+            trimToNull(deps.getCurrentMediaPath()) ?? trimToNull(deps.getCurrentVideoPath?.()),
+          mediaTitle: trimToNull(deps.getCurrentMediaTitle()),
         };
   };
 

src/main/character-dictionary-runtime/types.ts

@@ -137,6 +137,7 @@ export type CharacterDictionaryManualSelectionResult = {
 export interface CharacterDictionaryRuntimeDeps {
   userDataPath: string;
   getCurrentMediaPath: () => string | null;
+  getCurrentVideoPath?: () => string | null | undefined;
   getCurrentMediaTitle: () => string | null;
   resolveMediaPathForJimaku: (mediaPath: string | null) => string | null;
   guessAnilistMediaInfo: (

src/main/dependencies.ts

@@ -126,6 +126,9 @@ export interface AnkiJimakuIpcRuntimeServiceDepsParams {
   getAnkiIntegration: AnkiJimakuIpcRuntimeOptions['getAnkiIntegration'];
   setAnkiIntegration: AnkiJimakuIpcRuntimeOptions['setAnkiIntegration'];
   getKnownWordCacheStatePath: AnkiJimakuIpcRuntimeOptions['getKnownWordCacheStatePath'];
+  getCachedMediaPath?: AnkiJimakuIpcRuntimeOptions['getCachedMediaPath'];
+  shouldRequireRemoteMediaCache?: AnkiJimakuIpcRuntimeOptions['shouldRequireRemoteMediaCache'];
+  getYoutubeMediaSourceUrl?: AnkiJimakuIpcRuntimeOptions['getYoutubeMediaSourceUrl'];
   showDesktopNotification: AnkiJimakuIpcRuntimeOptions['showDesktopNotification'];
   showOverlayNotification?: (payload: OverlayNotificationPayload) => void;
   createFieldGroupingCallback: AnkiJimakuIpcRuntimeOptions['createFieldGroupingCallback'];
@@ -317,6 +320,13 @@ export function createAnkiJimakuIpcRuntimeServiceDeps(
     getAnkiIntegration: params.getAnkiIntegration,
     setAnkiIntegration: params.setAnkiIntegration,
     getKnownWordCacheStatePath: params.getKnownWordCacheStatePath,
+    ...(params.getCachedMediaPath ? { getCachedMediaPath: params.getCachedMediaPath } : {}),
+    ...(params.shouldRequireRemoteMediaCache
+      ? { shouldRequireRemoteMediaCache: params.shouldRequireRemoteMediaCache }
+      : {}),
+    ...(params.getYoutubeMediaSourceUrl
+      ? { getYoutubeMediaSourceUrl: params.getYoutubeMediaSourceUrl }
+      : {}),
     showDesktopNotification: params.showDesktopNotification,
     showOverlayNotification: params.showOverlayNotification,
     createFieldGroupingCallback: params.createFieldGroupingCallback,

src/main/main-wiring.test.ts

@@ -50,6 +50,16 @@ test('media path changes clear rendered subtitle state without clearing same-you
   );
 });
 
+test('media path changes start the YouTube media cache coordinator', () => {
+  const source = readMainSource();
+  const actionBlock = source.match(
+    /updateCurrentMediaPath:\s*\(path\)\s*=>\s*\{(?<body>[\s\S]*?)\n    restoreMpvSubVisibility:/,
+  )?.groups?.body;
+
+  assert.ok(actionBlock);
+  assert.match(actionBlock, /youtubeMediaCachePlaybackRuntime\.handleMediaPathChange\(path\);/);
+});
+
 test('same media path updates do not reset autoplay ready fallback state', () => {
   const source = readMainSource();
   const actionBlock = source.match(
@@ -308,7 +318,7 @@ test('visible overlay content-ready does not tokenize before first measurement',
   );
 });
 
-test('accepted visible overlay measurement immediately refreshes Linux pointer interaction', () => {
+test('accepted visible overlay measurement immediately refreshes pointer interaction', () => {
   const source = readMainSource();
   const measurementBlock = source.match(
     /reportOverlayContentBounds:\s*\(payload: unknown\)\s*=>\s*\{(?<body>[\s\S]*?)\n      \},/,
@@ -317,6 +327,7 @@ test('accepted visible overlay measurement immediately refreshes Linux pointer i
   assert.ok(measurementBlock);
   assert.match(measurementBlock, /overlayContentMeasurementStore\.report\(payload\)/);
   assert.match(measurementBlock, /tickLinuxOverlayPointerInteractionNow\(\)/);
+  assert.match(measurementBlock, /tickWindowsOverlayPointerInteractionNow\(\)/);
   assert.match(measurementBlock, /primeLinuxOverlayPointerInteractionAfterFirstMeasurement\(\)/);
   assert.ok(
     measurementBlock.indexOf('overlayContentMeasurementStore.report(payload)') <
@@ -324,6 +335,10 @@ test('accepted visible overlay measurement immediately refreshes Linux pointer i
   );
   assert.ok(
     measurementBlock.indexOf('tickLinuxOverlayPointerInteractionNow();') <
+      measurementBlock.indexOf('tickWindowsOverlayPointerInteractionNow();'),
+  );
+  assert.ok(
+    measurementBlock.indexOf('tickWindowsOverlayPointerInteractionNow();') <
       measurementBlock.indexOf('primeLinuxOverlayPointerInteractionAfterFirstMeasurement();'),
   );
 });
@@ -509,6 +524,48 @@ test('configured overlay notifications require visible ready overlay window', ()
   assert.match(statusBlock, /isOverlayReady: \(\) => isVisibleOverlayContentReady\(\)/);
 });
 
+test('YouTube media cache lifecycle routes through configured status notifications', () => {
+  const source = readMainSource();
+  const cacheBlock = source.match(
+    /const youtubeMediaCache = createYoutubeMediaCacheService\(\{(?<body>[\s\S]*?)\n\}\);\nconst waitForYoutubeMpvConnected/,
+  )?.groups?.body;
+  const startCacheBlock = source.match(
+    /startYoutubeMediaCache:\s*\(url\)\s*=>\s*\{(?<body>[\s\S]*?)\n  \},\n  runYoutubePlaybackFlow/,
+  )?.groups?.body;
+
+  assert.ok(cacheBlock);
+  assert.ok(startCacheBlock);
+  assert.match(
+    cacheBlock,
+    /onDownloadStarted:\s*\(event\)\s*=>\s*\{[\s\S]*showConfiguredStatusNotification\(\s*'YouTube media cache is downloading\.'/,
+  );
+  assert.match(cacheBlock, /id:\s*'youtube-media-cache-status'/);
+  assert.match(cacheBlock, /variant:\s*'progress'/);
+  assert.match(cacheBlock, /persistent:\s*true/);
+  assert.match(
+    cacheBlock,
+    /onReady:\s*\(event\)\s*=>\s*\{[\s\S]*showConfiguredStatusNotification\(\s*'YouTube media cache ready\.'/,
+  );
+  assert.match(cacheBlock, /variant:\s*'success'/);
+  assert.match(cacheBlock, /notifyNoQueued:\s*false/);
+  assert.match(startCacheBlock, /mode:\s*getResolvedConfig\(\)\.youtube\.mediaCache\.mode/);
+  assert.match(
+    startCacheBlock,
+    /maxHeight:\s*getResolvedConfig\(\)\.youtube\.mediaCache\.maxHeight/,
+  );
+});
+
+test('mpv connection flushes queued configured OSD notifications', () => {
+  const source = readMainSource();
+  const connectedBlock = source.match(
+    /onMpvConnected:\s*\(\)\s*=>\s*\{(?<body>[\s\S]*?)\n    \},\n    maybeRunAnilistPostWatchUpdate:/,
+  )?.groups?.body;
+
+  assert.ok(connectedBlock);
+  assert.match(source, /flushQueuedMpvOsdNotifications/);
+  assert.match(connectedBlock, /flushQueuedMpvOsdNotifications\(\);/);
+});
+
 test('manual visible overlay show primes current subtitle from mpv before relying on live events', () => {
   const source = readMainSource();
   const setBlock = source.match(

src/main/runtime/app-lifecycle-actions.test.ts

@@ -41,18 +41,20 @@ test('on will quit cleanup handler runs all cleanup steps', () => {
     clearYomitanSettingsWindow: () => calls.push('clear-yomitan-settings-window'),
     stopJellyfinRemoteSession: () => calls.push('stop-jellyfin-remote'),
     cleanupYoutubeSubtitleTempDirs: () => calls.push('cleanup-youtube-subtitles'),
+    cleanupYoutubeMediaCache: () => calls.push('cleanup-youtube-media'),
     cleanupJellyfinSubtitleCache: () => calls.push('cleanup-jellyfin-subtitles'),
     stopDiscordPresenceService: () => calls.push('stop-discord-presence'),
   });
 
   cleanup();
-  assert.equal(calls.length, 32);
+  assert.equal(calls.length, 33);
   assert.equal(calls[0], 'destroy-tray');
   assert.equal(calls[calls.length - 1], 'stop-discord-presence');
   assert.ok(calls.includes('cleanup-jellyfin-subtitles'));
   assert.ok(calls.includes('clear-windows-visible-overlay-poll'));
   assert.ok(calls.includes('clear-linux-mpv-fullscreen-overlay-refresh-timeouts'));
   assert.ok(calls.includes('cleanup-youtube-subtitles'));
+  assert.ok(calls.includes('cleanup-youtube-media'));
   assert.ok(calls.indexOf('flush-mpv-log') < calls.indexOf('destroy-socket'));
 });
 
@@ -92,6 +94,7 @@ test('on will quit cleanup handler cleans jellyfin subtitle cache when stopping
       throw new Error('stop failed');
     },
     cleanupYoutubeSubtitleTempDirs: () => calls.push('cleanup-youtube-subtitles'),
+    cleanupYoutubeMediaCache: () => calls.push('cleanup-youtube-media'),
     cleanupJellyfinSubtitleCache: () => calls.push('cleanup-jellyfin-subtitles'),
     stopDiscordPresenceService: () => calls.push('stop-discord-presence'),
   });

src/main/runtime/app-lifecycle-actions.ts

@@ -29,6 +29,7 @@ export function createOnWillQuitCleanupHandler(deps: {
   clearYomitanSettingsWindow: () => void;
   stopJellyfinRemoteSession: () => void;
   cleanupYoutubeSubtitleTempDirs: () => void;
+  cleanupYoutubeMediaCache: () => void;
   cleanupJellyfinSubtitleCache: () => void;
   stopDiscordPresenceService: () => void;
 }) {
@@ -67,6 +68,7 @@ export function createOnWillQuitCleanupHandler(deps: {
       deps.cleanupJellyfinSubtitleCache();
     }
     deps.cleanupYoutubeSubtitleTempDirs();
+    deps.cleanupYoutubeMediaCache();
     deps.stopDiscordPresenceService();
   };
 }

src/main/runtime/app-lifecycle-main-cleanup.test.ts

@@ -70,6 +70,7 @@ test('cleanup deps builder returns handlers that guard optional runtime objects'
 
     stopJellyfinRemoteSession: () => calls.push('stop-jellyfin-remote'),
     cleanupYoutubeSubtitleTempDirs: () => calls.push('cleanup-youtube-subtitles'),
+    cleanupYoutubeMediaCache: () => calls.push('cleanup-youtube-media'),
     cleanupJellyfinSubtitleCache: () => calls.push('cleanup-jellyfin-subtitles'),
     stopDiscordPresenceService: () => calls.push('stop-discord-presence'),
   });
@@ -92,6 +93,7 @@ test('cleanup deps builder returns handlers that guard optional runtime objects'
   assert.ok(calls.includes('destroy-yomitan-settings-window'));
   assert.ok(calls.includes('stop-jellyfin-remote'));
   assert.ok(calls.includes('cleanup-youtube-subtitles'));
+  assert.ok(calls.includes('cleanup-youtube-media'));
   assert.ok(calls.includes('cleanup-jellyfin-subtitles'));
   assert.ok(calls.includes('stop-discord-presence'));
   assert.ok(calls.includes('clear-windows-visible-overlay-foreground-poll-loop'));
@@ -147,6 +149,7 @@ test('cleanup deps builder skips destroyed yomitan window', () => {
     clearYomitanSettingsWindow: () => {},
     stopJellyfinRemoteSession: () => {},
     cleanupYoutubeSubtitleTempDirs: () => {},
+    cleanupYoutubeMediaCache: () => {},
     cleanupJellyfinSubtitleCache: () => {},
     stopDiscordPresenceService: () => {},
   });
@@ -197,6 +200,7 @@ test('cleanup deps builder skips global shortcut cleanup before app ready', () =
     clearYomitanSettingsWindow: () => {},
     stopJellyfinRemoteSession: () => {},
     cleanupYoutubeSubtitleTempDirs: () => {},
+    cleanupYoutubeMediaCache: () => {},
     cleanupJellyfinSubtitleCache: () => {},
     stopDiscordPresenceService: () => {},
   });

src/main/runtime/app-lifecycle-main-cleanup.ts

@@ -58,6 +58,7 @@ export function createBuildOnWillQuitCleanupDepsHandler(deps: {
 
   stopJellyfinRemoteSession: () => void;
   cleanupYoutubeSubtitleTempDirs: () => void;
+  cleanupYoutubeMediaCache: () => void;
   cleanupJellyfinSubtitleCache: () => void;
   stopDiscordPresenceService: () => void;
 }) {
@@ -142,6 +143,7 @@ export function createBuildOnWillQuitCleanupDepsHandler(deps: {
     clearYomitanSettingsWindow: () => deps.clearYomitanSettingsWindow(),
     stopJellyfinRemoteSession: () => deps.stopJellyfinRemoteSession(),
     cleanupYoutubeSubtitleTempDirs: () => deps.cleanupYoutubeSubtitleTempDirs(),
+    cleanupYoutubeMediaCache: () => deps.cleanupYoutubeMediaCache(),
     cleanupJellyfinSubtitleCache: () => deps.cleanupJellyfinSubtitleCache(),
     stopDiscordPresenceService: () => deps.stopDiscordPresenceService(),
   });