fix(review): address latest CodeRabbit comments

Replace every user-facing "token(s)" label, tooltip, and message in the
stats UI with "words" so the terminology is consistent and friendlier
(e.g. "Words Seen", "word occurrences", "3.4 / 100 words", "Words Today").

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

.github/workflows/release.yml

@@ -334,14 +334,6 @@ jobs:
         id: version
         run: echo "VERSION=${GITHUB_REF#refs/tags/}" >> "$GITHUB_OUTPUT"
 
-      - name: Build changelog artifacts for release
-        run: |
-          if find changes -maxdepth 1 -name '*.md' -not -name README.md -print -quit | grep -q .; then
-            bun run changelog:build --version "${{ steps.version.outputs.VERSION }}"
-          else
-            echo "No pending changelog fragments found."
-          fi
-
       - name: Verify changelog is ready for tagged release
         run: bun run changelog:check --version "${{ steps.version.outputs.VERSION }}"
 

CHANGELOG.md

@@ -1,66 +1,5 @@
 # Changelog
 
-## v0.9.2 (2026-03-25)
-
-### Fixed
-- Overlay: Fixed overlay pointer tracking so Windows click-through toggles immediately when the cursor enters or leaves subtitle regions, without waiting for a later hover resync.
-- Overlay: Fixed Windows overlay window tracking on scaled displays by converting native tracked window bounds to Electron DIP coordinates before applying overlay bounds.
-- Launcher: Fixed Windows direct `--youtube-play` startup so MPV boots reliably, stays paused until the app-owned subtitle flow is ready, and reuses an already-running SubMiner instance when available.
-- Launcher: Fixed standalone Windows `--youtube-play` sessions so closing MPV fully exits SubMiner instead of leaving hidden overlay windows or a background process behind.
-- Overlay: Fixed `subminer <youtube-url>` on Linux so the YouTube playback flow waits for Yomitan to load before creating the overlay window, avoiding the broken lookup popup state that previously required a manual overlay refresh.
-
-## v0.9.1 (2026-03-24)
-
-### Changed
-- Release: Reduced packaged release size by excluding duplicate `extraResources` payload and pruning docs, tests, sourcemaps, and other source-only files from Electron bundles.
-
-### Fixed
-- Overlay: Restored controller navigation and lookup/mining controls while the subtitle sidebar is open, while keeping true modal dialogs blocking controller actions.
-- Tokenizer: Fixed subtitle annotation clearing so explanatory contrast endings like `んですけど` are excluded consistently across the shared tokenizer filter and annotation stage.
-
-## v0.9.0 (2026-03-23)
-
-### Added
-- Docs: Added a new WebSocket / Texthooker API and integration guide covering WebSocket payloads, custom client patterns, mpv plugin automation, and webhook-style relay examples. Linked from configuration and mining workflow docs for easier discovery.
-
-### Changed
-- Launcher: Added an app-owned YouTube subtitle flow that pauses mpv, uses absPlayer-style YouTube timedtext parsing/conversion to download subtitle tracks, and injects them as external files before playback resumes.
-- Launcher: Changed YouTube subtitle startup to auto-load the best-available primary and secondary subtitle tracks at launch instead of forcing the picker modal first. Secondary subtitle failures no longer block playback resume.
-- Launcher: Added `Ctrl+Alt+C` as the default keybinding to manually open the YouTube subtitle picker during active YouTube playback.
-- Launcher: Added yt-dlp metadata probing so YouTube playback and immersion tracking record canonical video title and channel metadata.
-- Launcher: Stopped forcing `--ytdl-raw-options=` before user-provided mpv options so existing YouTube cookie integrations in user `--args` are no longer clobbered.
-- Launcher: Disabled mpv native YouTube subtitle auto-loading for the app-owned flow so injected external subtitle files remain authoritative.
-- Launcher: Added OSD status messages for YouTube playback startup, subtitle acquisition, and subtitle loading so the flow stays visible before and during the picker.
-- Subtitle Sidebar: Added startup-auto-open controls and resume positioning improvements so the sidebar jumps directly to the first resolved active cue.
-- Subtitle Sidebar: Improved subtitle prefetch and embedded overlay passthrough sync so sidebar and overlay subtitle states stay consistent across media transitions.
-- Subtitle Sidebar: Updated scroll handling, embedded layout styling, and active-cue visual behavior.
-- Stats: Stats Library tab now displays YouTube video title, channel name, and channel thumbnail for YouTube media entries, with retry logic to fill in metadata that arrives after initial load.
-
-### Fixed
-- Launcher: Fixed Anki media mining for mpv YouTube streams by unwrapping the stream URL so audio and screenshot capture work correctly for YouTube playback sessions.
-- Immersion: Fixed YouTube media path handling in the immersion runtime and tracking so YouTube sessions record correct media references, AniList guessing skips YouTube URLs, and post-watch state transitions do not fire for YouTube media.
-- Launcher: Fixed startup-launched YouTube playback so primary subtitle overlay updates continue after auto-load completes.
-- Launcher: Fixed auto-loaded YouTube primary subtitles so parsed cues appear in the subtitle sidebar without needing a manual picker retry.
-- Launcher: Fixed the YouTube picker to guard against duplicate subtitle submissions and tightened YouTube URL detection so follow-up runtime flows only treat real YouTube hosts as YouTube playback.
-- Launcher: Fixed primary subtitle failure notifications being shown while app-owned YouTube subtitle probing and downloads are still in flight.
-- Launcher: Preserved existing authoritative YouTube subtitle tracks when available; downloaded tracks are used only to fill missing sides, and native mpv secondary subtitle rendering is hidden so the overlay remains the sole secondary display.
-
-## v0.8.0 (2026-03-22)
-
-### Added
-- Overlay: Added the subtitle sidebar feature with a new `subtitleSidebar` configuration surface and rendered sidebar modal with cue list rendering, click-to-seek, active-cue highlighting, and embedded layout support.
-- IPC: Added sidebar snapshot plumbing between renderer and main process for overlay/sidebar synchronization.
-
-### Changed
-- Config: Added hot-reloadable sidebar options for enablement, layout, visibility, typography, opacity, sizing, and interaction behavior (`autoOpen`, `pauseOnHover`, `autoScroll`, toggle key).
-- Docs: Added full `subtitleSidebar` documentation coverage, including sample config, option table, and toggle shortcut notes.
-- Runtime: Improved subtitle prefetch/rendering flow so sidebar and overlay subtitle states stay in sync across media transitions.
-
-### Fixed
-- Overlay: Kept sidebar cue tracking stable across playback transitions and timing edge cases.
-- Overlay: Improved sidebar resume/start behavior to jump directly to the first resolved active cue.
-- Overlay: Stopped stale subtitle refreshes from regressing active-cue and text state.
-
 ## v0.7.0 (2026-03-19)
 
 ### Added

README.md

@@ -1,163 +1,60 @@
 <div align="center">
-
-<img src="assets/SubMiner.png" width="160" alt="SubMiner logo">
+  <img src="assets/SubMiner.png" width="140" alt="SubMiner logo">
 
 # SubMiner
 
-## Turn mpv into a sentence-mining workstation.
+**Sentence-mine from mpv — look up words, one-key Anki export, immersion tracking.**
 
-Look up words with Yomitan, export to Anki in one key, track your immersion — all without leaving mpv.
-
-[![License: GPL v3](https://img.shields.io/badge/license-GPLv3-1a1a2e?style=flat-square)](https://www.gnu.org/licenses/gpl-3.0)
-[![Platform](https://img.shields.io/badge/platform-Linux%20·%20macOS%20·%20Windows-1a1a2e?style=flat-square)](https://github.com/ksyasuda/SubMiner)
-[![Docs](https://img.shields.io/badge/docs-docs.subminer.moe-e6a817?style=flat-square)](https://docs.subminer.moe)
-[![AUR](https://img.shields.io/aur/version/subminer-bin?style=flat-square&color=1a1a2e)](https://aur.archlinux.org/packages/subminer-bin)
-
-[![SubMiner demo](./assets/minecard.webp)](./assets/minecard.mp4)
+[![License: GPL v3](https://img.shields.io/badge/License-GPLv3-blue.svg)](https://www.gnu.org/licenses/gpl-3.0)
+[![Linux](https://img.shields.io/badge/platform-Linux%20%7C%20macOS%20%7C%20Windows-informational)](https://github.com/ksyasuda/SubMiner)
+[![Docs](https://img.shields.io/badge/docs-docs.subminer.moe-blueviolet)](https://docs.subminer.moe)
+[![AUR](https://img.shields.io/aur/version/subminer-bin)](https://aur.archlinux.org/packages/subminer-bin)
 
 </div>
 
-## How It Works
-
-SubMiner runs as an invisible Electron overlay on top of mpv. Subtitles render as an interactive layer. Move your cursor over any word and trigger a [Yomitan](https://github.com/yomidevs/yomitan) lookup. Press one key to snapshot the sentence, audio, and screenshot into Anki via AnkiConnect.
-
-## Features
-
-### Dictionary Lookups
-
-Yomitan runs inside the overlay. Trigger a lookup on any word for full dictionary popups — definitions, pitch accent, frequency data — without ever leaving mpv.
-
-<div align="center">
-  <img src="docs-site/public/screenshots/yomitan-lookup.png" width="800" alt="Yomitan dictionary popup over annotated subtitles in mpv">
-</div>
-
-<br>
-
-### Instant Anki Mining
-
-Create an Anki card with the sentence, audio clip, screenshot, and machine translation from the exact playback moment with one key press, click, or controller input.
-
-<div align="center">
-  <img src="docs-site/public/screenshots/one-key-mining.png" width="800" alt="Anki card created from SubMiner with sentence, audio, and screenshot">
-</div>
-
-<br>
-
-### Reading Annotations
-
-Real-time subtitle annotations with frequency highlighting, JLPT tags, N+1 targeting, and a character name dictionary. Known words fade back; new words stand out. Grammar-only tokens render as plain text so you focus on what matters.
-
-<div align="center">
-  <img src="docs-site/public/screenshots/annotations.png" width="800" alt="Annotated subtitles with frequency coloring, JLPT underlines, and N+1 targets">
-</div>
-
-<br>
-
-### Immersion Dashboard
-
-Local stats dashboard — watch time, anime library, vocabulary growth, mining throughput, session history, and trends. All stored locally, no third-party tracking.
-
-<div align="center">
-  <img src="docs-site/public/screenshots/stats-overview.png" width="800" alt="Stats dashboard showing watch time, cards mined, streaks, and tracking data">
-</div>
-
-<br>
-
-### Integrations
-
-<table>
-  <tr>
-    <td><b>YouTube</b></td>
-    <td>Auto-loaded yt-dlp subtitle tracks at startup with a manual overlay picker on demand (<code>Ctrl+Alt+C</code>)</td>
-  </tr>
-  <tr>
-    <td><b>AniList</b></td>
-    <td>Automatic episode tracking and progress sync</td>
-  </tr>
-  <tr>
-    <td><b>Jellyfin</b></td>
-    <td>Browse and launch media from your Jellyfin server</td>
-  </tr>
-  <tr>
-    <td><b>Jimaku</b></td>
-    <td>Search and download Japanese subtitles</td>
-  </tr>
-  <tr>
-    <td><b>alass / ffsubsync</b></td>
-    <td>Automatic subtitle retiming</td>
-  </tr>
-  <tr>
-    <td><b>WebSocket</b></td>
-    <td>Annotated subtitle feed for external clients (texthooker pages, custom tools)</td>
-  </tr>
-</table>
-
-<div align="center">
-  <img src="docs-site/public/screenshots/texthooker.png" width="800" alt="Texthooker page receiving annotated subtitle lines via WebSocket">
-</div>
-
-<br>
-
 ---
 
-## Requirements
+SubMiner is an Electron overlay for [mpv](https://mpv.io) that turns video into a sentence-mining workstation. Look up any word with [Yomitan](https://github.com/yomidevs/yomitan), mine it to Anki with one key, and track your immersion over time.
 
-|                | Required                                | Optional                               |
-| -------------- | --------------------------------------- | -------------------------------------- |
-| **Player**     | [`mpv`](https://mpv.io) with IPC socket | —                                      |
-| **Processing** | `ffmpeg`, `mecab` + `mecab-ipadic`      | `guessit` (AniSkip)                    |
-| **Media**      | —                                       | `yt-dlp`, `chafa`, `ffmpegthumbnailer` |
-| **Selection**  | —                                       | `fzf` / `rofi`                         |
+<div align="center">
 
-> [!NOTE]
-> [`bun`](https://bun.sh) is required if building from source or using the CLI wrapper: `subminer`. Pre-built releases (AppImage, DMG, installer) do not require it.
+[![SubMiner demo (Animated preview)](./assets/minecard.webp)](./assets/minecard.mp4)
 
-**Platform-specific:**
+</div>
 
-| Linux                               | macOS                    | Windows       |
-| ----------------------------------- | ------------------------ | ------------- |
-| `hyprctl` or `xdotool` + `xwininfo` | Accessibility permission | No extra deps |
+## Features
 
-<details>
-<summary><b>Arch Linux</b></summary>
+**Dictionary lookups** — Yomitan runs inside the overlay. Hover or navigate to any word for full dictionary popups without leaving mpv.
 
-```bash
-paru -S --needed mpv ffmpeg mecab-git mecab-ipadic
-# Optional
-paru -S --needed yt-dlp fzf rofi chafa ffmpegthumbnailer xdotool xorg-xwininfo
-# X11 / XWAYLAND
-paru -S --needed xdotool xorg-xwininfo
-```
+**One-key Anki mining** — Press one key to create a card with the sentence, audio clip, screenshot, and machine translation from the exact playback moment.
 
-</details>
+<div align="center">
+  <img src="docs-site/public/screenshots/yomitan-lookup.png" width="800" alt="Yomitan popup with dictionary entry and mine button over annotated subtitles in mpv">
+</div>
 
-<details>
-<summary><b>macOS</b></summary>
+**Reading annotations** — Real-time subtitle annotations with N+1 targeting, frequency highlighting, JLPT tags, and a character name dictionary. Grammar-only tokens render as plain text.
 
-```bash
-brew install mpv ffmpeg mecab mecab-ipadic
-# Optional
-brew install yt-dlp fzf rofi chafa ffmpegthumbnailer
-```
+<div align="center">
+  <img src="docs-site/public/screenshots/annotations.png" width="800" alt="Annotated subtitles with frequency highlighting, JLPT underlines, known words, and N+1 targets">
+</div>
 
-Grant Accessibility permission to SubMiner in **System Settings > Privacy & Security > Accessibility**.
+**Immersion dashboard** — Local stats dashboard with watch time, anime progress, vocabulary growth, mining throughput, and session history.
 
-</details>
+<div align="center">
+  <img src="docs-site/public/screenshots/stats-overview.png" width="800" alt="Stats dashboard with watch time, cards mined, streaks, and tracking snapshot">
+</div>
 
-<details>
-<summary><b>Windows</b></summary>
+**Integrations** — AniList episode tracking, Jellyfin remote playback, Jimaku subtitle downloads, alass/ffsubsync, and an annotated websocket feed for external clients.
 
-Install [`mpv`](https://mpv.io/installation/) and [`ffmpeg`](https://ffmpeg.org/download.html) and ensure both are on your `PATH`.
-
-For MeCab, install [MeCab for Windows](https://taku910.github.io/mecab/#download) with the UTF-8 dictionary.
-
-</details>
+<div align="center">
+  <img src="docs-site/public/screenshots/texthooker.png" width="800" alt="Texthooker page with annotated subtitle lines and frequency highlighting">
+</div>
 
 ---
 
 ## Quick Start
 
-### 1. Install
+### Install
 
 <details>
 <summary><b>Arch Linux (AUR)</b></summary>
@@ -191,63 +88,53 @@ wget https://github.com/ksyasuda/SubMiner/releases/latest/download/subminer -O ~
 </details>
 
 <details>
-<summary><b>macOS</b></summary>
+<summary><b>macOS / Windows / From source</b></summary>
 
-Download the latest DMG or ZIP from [GitHub Releases](https://github.com/ksyasuda/SubMiner/releases/latest) and drag `SubMiner.app` into `/Applications`.
+**macOS** — Download the latest DMG/ZIP from [GitHub Releases](https://github.com/ksyasuda/SubMiner/releases/latest) and drag `SubMiner.app` into `/Applications`.
+
+**Windows** — Download the latest installer or portable `.zip` from [GitHub Releases](https://github.com/ksyasuda/SubMiner/releases/latest). Keep `mpv` on `PATH`.
+
+**From source** — See [docs.subminer.moe/installation#from-source](https://docs.subminer.moe/installation#from-source).
 
 </details>
 
-<details>
-<summary><b>Windows</b></summary>
+### First Launch
 
-Download the latest installer or portable `.zip` from [GitHub Releases](https://github.com/ksyasuda/SubMiner/releases/latest). Make sure `mpv` is on your `PATH`.
+Run `SubMiner.AppImage` (Linux), `SubMiner.app` (macOS), or `SubMiner.exe` (Windows). On first launch, SubMiner starts in the tray, creates a default config, and opens a setup popup where you can install the mpv plugin and configure Yomitan dictionaries.
 
-</details>
-
-<details>
-<summary><b>From source</b></summary>
-
-See the [build-from-source guide](https://docs.subminer.moe/installation#from-source).
-
-</details>
-
-### 2. First Launch
-
-Run the app. On first launch SubMiner starts in the system tray, creates a default config, and opens a setup popup to install the mpv plugin and configure Yomitan dictionaries.
-
-### 3. Mine
+### Mine
 
 ```bash
-subminer video.mkv          # play video with overlay
-subminer --start video.mkv  # explicit overlay start
-subminer stats              # open immersion dashboard
-subminer stats -b           # stats daemon in background
-subminer stats -s           # stop background stats daemon
+subminer video.mkv          # auto-starts overlay + resumes playback
+subminer --start video.mkv  # explicit overlay start (if plugin auto_start=no)
+subminer stats              # open the immersion dashboard
+subminer stats -b           # keep the stats daemon running in background
+subminer stats -s           # stop the dedicated stats daemon
+subminer stats cleanup      # repair/prune stored stats vocabulary rows
 ```
 
 ---
 
+## Requirements
+
+| Required                                               | Optional                      |
+| ------------------------------------------------------ | ----------------------------- |
+| [`mpv`](https://mpv.io) with IPC socket                | `yt-dlp`                      |
+| `ffmpeg`                                               | `guessit` (AniSkip detection) |
+| `mecab` + `mecab-ipadic`                               | `fzf` / `rofi`                |
+| [`bun`](https://bun.sh) (source builds, Linux wrapper) | `chafa`, `ffmpegthumbnailer`  |
+| Linux: `hyprctl` or `xdotool` + `xwininfo`             |                               |
+| macOS: Accessibility permission                        |                               |
+
+Windows uses native window tracking and does not need the Linux compositor tools.
+
 ## Documentation
 
-Full guides on configuration, Anki setup, Jellyfin, immersion tracking, and more: **[docs.subminer.moe](https://docs.subminer.moe)**
-
----
+Full guides on configuration, Anki, Jellyfin, immersion tracking, and more at **[docs.subminer.moe](https://docs.subminer.moe)**.
 
 ## Acknowledgments
 
-SubMiner builds on the work of these open-source projects:
-
-| Project                                                                                     | Role                                                                    |
-| ------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------- |
-| [Anacreon-Script](https://github.com/friedrich-de/Anacreon-Script)                          | Inspiration for the mining workflow                                     |
-| [asbplayer](https://github.com/killergerbah/asbplayer)                                      | Inspiration for subtitle sidebar and logic for YouTube subtitle parsing |
-| [Bee's Character Dictionary](https://github.com/bee-san/Japanese_Character_Name_Dictionary) | Character name recognition in subtitles                                 |
-| [GameSentenceMiner](https://github.com/bpwhelan/GameSentenceMiner)                          | Inspiration for Electron overlay with Yomitan integration               |
-| [jellyfin-mpv-shim](https://github.com/jellyfin/jellyfin-mpv-shim)                          | Jellyfin integration                                                    |
-| [Jimaku.cc](https://jimaku.cc)                                                              | Japanese subtitle search and downloads                                  |
-| [Renji's Texthooker Page](https://github.com/Renji-XD/texthooker-ui)                        | Base for the WebSocket texthooker integration                           |
-| [Yomitan](https://github.com/yomidevs/yomitan)                                              | Dictionary engine powering all lookups and the morphological parser     |
-| [yomitan-jlpt-vocab](https://github.com/stephenmk/yomitan-jlpt-vocab)                       | JLPT level tags for vocabulary                                          |
+Built on [GameSentenceMiner](https://github.com/bpwhelan/GameSentenceMiner), [Renji's Texthooker Page](https://github.com/Renji-XD/texthooker-ui), [Anacreon-Script](https://github.com/friedrich-de/Anacreon-Script), and [Bee's Character Dictionary](https://github.com/bee-san/Japanese_Character_Name_Dictionary). Subtitles from [Jimaku.cc](https://jimaku.cc). Lookups via [Yomitan](https://github.com/yomidevs/yomitan). JLPT tags from [yomitan-jlpt-vocab](https://github.com/stephenmk/yomitan-jlpt-vocab).
 
 ## License
 

backlog/tasks/task-143 - Keep-character-dictionary-auto-sync-non-blocking-during-startup.md

@@ -2,10 +2,9 @@
 id: TASK-143
 title: Keep character dictionary auto-sync non-blocking during startup
 status: Done
-assignee:
-  - codex
+assignee: []
 created_date: '2026-03-09 01:45'
-updated_date: '2026-03-23 03:22'
+updated_date: '2026-03-18 05:28'
 labels:
   - dictionary
   - startup
@@ -18,7 +17,7 @@ references:
   - >-
     /home/sudacode/projects/japanese/SubMiner/src/main/runtime/current-media-tokenization-gate.ts
 priority: high
-ordinal: 144500
+ordinal: 38500
 ---
 
 ## Description
@@ -34,20 +33,8 @@ Keep character dictionary auto-sync running in parallel during startup without d
 - [x] #3 Regression coverage verifies auto-sync builds before the gate and only mutates Yomitan after the gate resolves.
 <!-- AC:END -->
 
-## Implementation Plan
-
-<!-- SECTION:PLAN:BEGIN -->
-1. Add a regression test for startup autoplay release surviving delayed mpv readiness or late subtitle refresh after dictionary sync.
-2. Harden the autoplay-ready release path so paused startup keeps retrying until mpv is actually released or media changes, without resuming user-paused playback later.
-3. Keep the existing character-dictionary revisit fixes and paused-startup OSD fixes aligned with the autoplay change, then run targeted runtime tests and typecheck.
-<!-- SECTION:PLAN:END -->
-
 ## Implementation Notes
 
 <!-- SECTION:NOTES:BEGIN -->
 Added a small current-media tokenization gate in main runtime. Media changes reset the gate, the first tokenization-ready event marks it ready, and auto-sync now waits on that gate only before Yomitan dictionary inspection/import/settings updates. Snapshot generation and merged ZIP build still run immediately in parallel.
-
-2026-03-20: User reports startup remains paused after annotations/tokenization are visible and only resumes after character-dictionary generation/import finishes. Investigating autoplay-ready release regression vs dictionary sync completion refresh.
-
-2026-03-20: Added startup autoplay retry-budget helper so paused startup retries cover the full plugin gate window instead of only ~2.8s. Verification: bun test src/main/runtime/startup-autoplay-release-policy.test.ts src/main/runtime/character-dictionary-auto-sync.test.ts src/main/runtime/startup-osd-sequencer.test.ts src/main/runtime/character-dictionary-auto-sync-completion.test.ts; bun run typecheck; bun run test:fast; bun run test:env; bun run build; bun run test:smoke:dist; runtime-compat verifier passed at .tmp/skill-verification/subminer-verify-20260320-022106-nM28Nk. Pending real installed-app/mpv validation.
 <!-- SECTION:NOTES:END -->

backlog/tasks/task-177.1 - Fix-overview-lookup-rate-metric.md

@@ -5,7 +5,7 @@ status: Done
 assignee:
   - '@codex'
 created_date: '2026-03-19 17:46'
-updated_date: '2026-03-23 03:22'
+updated_date: '2026-03-19 17:54'
 labels:
   - stats
   - immersion-tracking
@@ -19,7 +19,6 @@ references:
   - src/core/services/stats-server.ts
 parent_task_id: TASK-177
 priority: medium
-ordinal: 132500
 ---
 
 ## Description

backlog/tasks/task-177.2 - Count-homepage-new-words-by-headword.md

@@ -5,7 +5,7 @@ status: Done
 assignee:
   - '@codex'
 created_date: '2026-03-19 19:38'
-updated_date: '2026-03-23 03:22'
+updated_date: '2026-03-19 19:40'
 labels:
   - stats
   - immersion-tracking
@@ -17,7 +17,6 @@ references:
   - stats/src/lib/dashboard-data.ts
 parent_task_id: TASK-177
 priority: medium
-ordinal: 130500
 ---
 
 ## Description

backlog/tasks/task-177.3 - Fix-attached-stats-command-flow-and-browser-config.md

@@ -5,7 +5,7 @@ status: Done
 assignee:
   - '@codex'
 created_date: '2026-03-19 20:15'
-updated_date: '2026-03-23 03:22'
+updated_date: '2026-03-19 20:17'
 labels:
   - launcher
   - stats
@@ -19,7 +19,6 @@ references:
   - src/main/runtime/stats-cli-command.test.ts
 parent_task_id: TASK-177
 priority: medium
-ordinal: 129500
 ---
 
 ## Description

backlog/tasks/task-182.2 - Improve-session-detail-known-word-chart-scaling.md

@@ -5,7 +5,7 @@ status: Done
 assignee:
   - codex
 created_date: '2026-03-19 20:31'
-updated_date: '2026-03-23 03:22'
+updated_date: '2026-03-19 20:52'
 labels:
   - bug
   - stats
@@ -17,7 +17,6 @@ references:
   - >-
     /Users/sudacode/projects/japanese/SubMiner/stats/src/lib/session-detail.test.tsx
 parent_task_id: TASK-182
-ordinal: 128500
 ---
 
 ## Description

backlog/tasks/task-188 - Refactor-stats-chart-data-pipeline-to-use-backend-aggregated-series.md

@@ -5,7 +5,7 @@ status: Done
 assignee:
   - codex
 created_date: '2026-03-18 00:29'
-updated_date: '2026-03-23 03:22'
+updated_date: '2026-03-18 00:55'
 labels:
   - stats
   - performance
@@ -22,7 +22,6 @@ references:
   - stats/src/types/stats.ts
   - stats/src/lib/dashboard-data.ts
 priority: medium
-ordinal: 138500
 ---
 
 ## Description

backlog/tasks/task-191 - Assess-PR-19-CodeRabbit-review-follow-ups.md

@@ -5,7 +5,7 @@ status: Done
 assignee:
   - codex
 created_date: '2026-03-17 23:15'
-updated_date: '2026-03-23 03:22'
+updated_date: '2026-03-17 23:18'
 labels:
   - pr-review
   - stats
@@ -16,7 +16,6 @@ references:
   - src/core/services/immersion-tracker-service.ts
   - src/core/services/immersion-tracker-service.test.ts
 priority: medium
-ordinal: 139500
 ---
 
 ## Description

backlog/tasks/task-192 - Assess-remaining-PR-19-review-batch.md

@@ -0,0 +1,76 @@
+---
+id: TASK-192
+title: 'Assess remaining PR #19 review batch'
+status: Done
+assignee:
+  - codex
+created_date: '2026-03-17 23:24'
+updated_date: '2026-03-17 23:42'
+labels:
+  - pr-review
+  - stats
+  - docs
+milestone: m-1
+dependencies: []
+references:
+  - docs/superpowers/plans/2026-03-12-immersion-stats-page.md
+  - src/core/services/immersion-tracker/__tests__/query.test.ts
+  - src/core/services/ipc.ts
+  - src/core/services/stats-server.ts
+  - src/main.ts
+  - src/renderer/handlers/keyboard.ts
+  - stats/src
+priority: medium
+---
+
+## Description
+
+<!-- SECTION:DESCRIPTION:BEGIN -->
+Validate the remaining PR #19 automated review findings against the current branch, implement only the technically correct fixes, and document which comments are stale, already addressed, or not warranted.
+<!-- SECTION:DESCRIPTION:END -->
+
+## Acceptance Criteria
+<!-- AC:BEGIN -->
+- [x] #1 Each remaining review comment is classified as actionable, already fixed, stale, or not warranted
+- [x] #2 Confirmed bugs or correctness issues are fixed with focused regression coverage where it fits
+- [x] #3 Final notes record which comments were intentionally not applied and why
+<!-- AC:END -->
+
+## Implementation Plan
+
+<!-- SECTION:PLAN:BEGIN -->
+1. Inspect the referenced files in batches and compare each comment against current branch behavior.
+2. Separate correctness/security regressions from stylistic nitpicks and already-fixed items.
+3. Add tests first for confirmed behavior bugs where practical, apply the smallest safe fixes, and rerun targeted verification.
+<!-- SECTION:PLAN:END -->
+
+## Implementation Notes
+
+<!-- SECTION:NOTES:BEGIN -->
+Swept the pasted PR #19 review batch against the current branch.
+
+Classification:
+- Already fixed on current branch: `src/core/services/immersion-tracker/__tests__/query.test.ts` cleanup rethrow, `src/core/services/ipc.ts` limit validation, `src/core/services/stats-server.ts` max-limit parsing and CORS removal, `src/main.ts` quit-path TDZ issue, `src/renderer/handlers/keyboard.ts` stats-toggle shortcut ordering/config usage, `stats/src/components/vocabulary/WordList.tsx`, `stats/src/hooks/useSessions.ts`, `stats/src/hooks/useTrends.ts` stale-error reset, `src/core/services/__tests__/stats-server.test.ts` kanji endpoint/readability notes, `src/core/services/stats-window.ts`, `stats/src/App.tsx`, `stats/src/components/layout/TabBar.tsx`, `stats/src/components/overview/QuickStats.tsx`, `stats/src/components/overview/WatchTimeChart.tsx`, `stats/src/components/sessions/SessionDetail.tsx`, `stats/src/components/sessions/SessionRow.tsx`, `stats/src/components/trends/DateRangeSelector.tsx`, `stats/src/components/vocabulary/KanjiBreakdown.tsx`, `stats/src/components/vocabulary/VocabularyTab.tsx`, `stats/src/hooks/useVocabulary.ts`, `stats/src/lib/api-client.ts`, `stats/src/types/stats.ts`.
+- Stale / obsolete against current architecture: `docs/superpowers/plans/2026-03-12-immersion-stats-page.md` path does not exist on this branch; `stats/src/components/trends/TrendsTab.tsx` / monthly-range comments describe older client-side aggregation code that is no longer present because trends now come from `getTrendsDashboard`.
+- Not warranted as written: `stats/src/lib/formatters.ts` no longer emits negative `Xd ago`; current code short-circuits future timestamps to `just now`, so the reported bug condition is gone even though the suggested wording differs.
+- Actionable and fixed now: `src/core/services/ipc.ts` no-tracker `statsGetOverview` fallback omitted required hint fields (`totalLookupCount`, `totalLookupHits`, `newWordsToday`, `newWordsThisWeek`). Added the missing fields in the fallback object and updated IPC tests to assert the full shape.
+
+Verification:
+- `bun test src/core/services/ipc.test.ts`
+- `bun test src/core/services/ipc.test.ts --test-name-pattern "empty stats overview shape without a tracker|validates and clamps stats request limits"`
+- `bash .agents/skills/subminer-change-verification/scripts/classify_subminer_diff.sh src/core/services/ipc.ts src/core/services/ipc.test.ts`
+
+Repo verifier note:
+- `bash .agents/skills/subminer-change-verification/scripts/verify_subminer_change.sh --lane core src/core/services/ipc.ts src/core/services/ipc.test.ts`
+- That verifier run captured a temporary `bun run typecheck` failure in `src/anki-integration.test.ts` and `src/core/services/__tests__/stats-server.test.ts`, but a fresh rerun after the follow-up validation no longer reproduces those diagnostics.
+- Fresh verification: `bun run typecheck` passes locally.
+- artifact dir from the earlier failed verifier snapshot: `.tmp/skill-verification/subminer-verify-20260317-234027-i6QJ3n`
+<!-- SECTION:NOTES:END -->
+
+## Final Summary
+
+<!-- SECTION:FINAL_SUMMARY:BEGIN -->
+The larger pasted PR #19 review batch was not mostly new work on the current branch. After verifying each item against the live code, almost all were already fixed or stale. One additional item was still actionable: the no-tracker fallback returned by `statsGetOverview` in `src/core/services/ipc.ts` omitted required hint fields, which made the fallback shape inconsistent with the normal overview payload. That fallback is now fixed and covered by IPC tests.
+
+Count-wise: the earlier open CodeRabbit service comments contributed 2 actionable fixes, and this larger pasted batch contributed 1 additional actionable fix on top of those.
+<!-- SECTION:FINAL_SUMMARY:END -->

backlog/tasks/task-192 - Fix-stale-anime-cover-art-after-AniList-reassignment.md

@@ -1,68 +0,0 @@
----
-id: TASK-192
-title: Fix stale anime cover art after AniList reassignment
-status: Done
-assignee:
-  - codex
-created_date: '2026-03-20 00:12'
-updated_date: '2026-03-23 03:22'
-labels:
-  - stats
-  - immersion-tracker
-  - anilist
-milestone: m-1
-dependencies: []
-references:
-  - src/core/services/immersion-tracker-service.ts
-  - src/core/services/immersion-tracker/query.ts
-  - src/core/services/immersion-tracker-service.test.ts
-priority: medium
-ordinal: 127500
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-Fix the stats anime-detail cover image path so reassigning an anime to a different AniList entry replaces the stored cover art bytes instead of keeping the previous image blob under updated metadata.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 Reassigning an anime to a different AniList entry stores the new cover art bytes for that anime's videos
-- [x] #2 Shared blob deduplication still works when multiple videos in the anime use the same new cover image
-- [x] #3 Focused regression coverage proves stale cover blobs are replaced on reassignment
-<!-- AC:END -->
-
-## Implementation Plan
-
-<!-- SECTION:PLAN:BEGIN -->
-1. Add a failing regression test that reassigns an anime twice with different downloaded cover bytes and asserts the resolved cover updates.
-2. Update cover-art upsert logic so new blob bytes generate a new shared hash instead of reusing an existing hash for the row.
-3. Run the focused immersion tracker service test file and record the result.
-<!-- SECTION:PLAN:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-2026-03-20: Created during live debugging of a user-reported stale anime profile picture after changing the AniList entry from the stats UI.
-2026-03-20: Root cause was in `upsertCoverArt(...)`. When a row already had `cover_blob_hash`, a later AniList reassignment with a freshly downloaded cover reused the existing hash instead of hashing the new bytes, so the blob store kept serving the old image while metadata changed.
-2026-03-20: Added a regression in `src/core/services/immersion-tracker-service.test.ts` that reassigns the same anime twice with different fetched image bytes and asserts the resolved anime cover changes to the second blob while both videos still deduplicate to one shared hash.
-2026-03-20: Fixed `src/core/services/immersion-tracker/query.ts` so incoming cover blob bytes compute a fresh hash before falling back to an existing row hash. Existing hashes are now reused only when no new bytes were fetched.
-2026-03-20: Verification commands run:
-  - `bun test src/core/services/immersion-tracker-service.test.ts`
-  - `bash .agents/skills/subminer-change-verification/scripts/classify_subminer_diff.sh src/core/services/immersion-tracker/query.ts src/core/services/immersion-tracker-service.test.ts`
-  - `bash .agents/skills/subminer-change-verification/scripts/verify_subminer_change.sh --lane core src/core/services/immersion-tracker/query.ts src/core/services/immersion-tracker-service.test.ts`
-2026-03-20: Verification results:
-  - focused service test: passed
-  - verifier lane selection: `core`
-  - verifier result: passed (`bun run typecheck`, `bun run test:fast`)
-  - verifier artifacts: `.tmp/skill-verification/subminer-verify-20260320-001433-IZLFqs/`
-<!-- SECTION:NOTES:END -->
-
-## Final Summary
-
-<!-- SECTION:FINAL_SUMMARY:BEGIN -->
-Fixed stale anime cover art after AniList reassignment by correcting cover-blob hash replacement in the immersion tracker storage layer. Reassignments now store the new fetched image bytes instead of reusing the previous blob hash from the row, while still deduplicating the updated image across videos in the same anime.
-
-Added focused regression coverage that reproduces the exact failure mode: same anime reassigned twice with different cover downloads, with the second image expected to replace the first. Verified with the touched service test file plus the SubMiner `core` verification lane.
-<!-- SECTION:FINAL_SUMMARY:END -->

backlog/tasks/task-194 - App-owned-YouTube-subtitle-picker-flow.md

@@ -1,35 +0,0 @@
----
-id: TASK-194
-title: App-owned YouTube subtitle picker flow
-status: Done
-assignee: []
-created_date: '2026-03-18 07:52'
-updated_date: '2026-03-23 03:22'
-labels: []
-dependencies: []
-references:
-  - /home/sudacode/projects/japanese/SubMiner/launcher/youtube/orchestrator.ts
-  - /home/sudacode/projects/japanese/SubMiner/launcher/youtube/manual-subs.ts
-  - /home/sudacode/projects/japanese/SubMiner/src/core/services/tokenizer.ts
-documentation:
-  - /home/sudacode/projects/japanese/SubMiner/youtube.md
-priority: medium
-ordinal: 137500
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-Replace the YouTube subtitle-generation-first flow with an app-owned picker flow that boots mpv paused, opens an overlay track picker, downloads selected subtitles into external subtitle files, and preserves generation as an explicit mode. Keep the existing SubMiner tokenization and annotation pipeline as the downstream consumer of downloaded subtitle files.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 Launcher and app expose YouTube subtitle acquisition modes `download` and `generate`, with `download` as the default.
-- [x] #2 YouTube playback boots mpv paused and presents an overlay selection UI for primary and secondary subtitle choices.
-- [x] #3 Selected YouTube subtitle tracks are downloaded to external subtitle files and loaded into mpv before playback resumes.
-- [x] #4 `generate` mode preserves the existing subtitle generation path as an explicit opt-in behavior.
-- [x] #5 Downloaded YouTube subtitle files integrate with the existing SubMiner subtitle/tokenization/annotation pipeline without regressing current overlay behavior.
-- [x] #6 Tests cover mode selection, subtitle-track enumeration/selection flow, and the paused bootstrap plus app handoff path.
-- [x] #7 User-facing config and launcher docs are updated to describe the new modes and default behavior.
-<!-- AC:END -->

backlog/tasks/task-194 - Redesign-YouTube-subtitle-acquisition-around-download-first-track-selection.md

@@ -0,0 +1,34 @@
+---
+id: TASK-194
+title: Redesign YouTube subtitle acquisition around download-first track selection
+status: To Do
+assignee: []
+created_date: '2026-03-18 07:52'
+labels: []
+dependencies: []
+references:
+  - /home/sudacode/projects/japanese/SubMiner/launcher/youtube/orchestrator.ts
+  - /home/sudacode/projects/japanese/SubMiner/launcher/youtube/manual-subs.ts
+  - /home/sudacode/projects/japanese/SubMiner/src/core/services/tokenizer.ts
+documentation:
+  - /home/sudacode/projects/japanese/SubMiner/youtube.md
+priority: medium
+---
+
+## Description
+
+<!-- SECTION:DESCRIPTION:BEGIN -->
+Replace the current YouTube subtitle-generation-first flow with a download-first flow that enumerates available YouTube subtitle tracks, prompts for primary and secondary track selection before playback, downloads selected tracks into external subtitle files for mpv, and preserves generation as an explicit mode and as fallback behavior in auto mode. Keep the existing SubMiner tokenization and annotation pipeline as the downstream consumer of downloaded subtitle files.
+<!-- SECTION:DESCRIPTION:END -->
+
+## Acceptance Criteria
+<!-- AC:BEGIN -->
+- [ ] #1 Launcher and config expose YouTube subtitle acquisition modes `download`, `generate`, and `auto`, with `download` as the default for launcher YouTube playback.
+- [ ] #2 YouTube playback enumerates available subtitle tracks before mpv launch and presents a selection UI that supports primary and secondary subtitle choices.
+- [ ] #3 Selected YouTube subtitle tracks are downloaded to external subtitle files and loaded into mpv before playback starts when download mode succeeds.
+- [ ] #4 `auto` mode attempts download-first for the selected tracks and falls back to generation only when required tracks cannot be downloaded or download fails.
+- [ ] #5 `generate` mode preserves the existing whisper/AI generation path as an explicit opt-in behavior.
+- [ ] #6 Downloaded YouTube subtitle files integrate with the existing SubMiner subtitle/tokenization/annotation pipeline without regressing current overlay behavior.
+- [ ] #7 Tests cover mode selection, subtitle-track enumeration/selection flow, download-first success path, and fallback behavior for auto mode.
+- [ ] #8 User-facing config and launcher docs are updated to describe the new modes and default behavior.
+<!-- AC:END -->

backlog/tasks/task-196 - Fix-subtitle-prefetch-cache-key-mismatch-and-active-cue-window.md

@@ -4,16 +4,13 @@ title: Fix subtitle prefetch cache-key mismatch and active-cue window
 status: Done
 assignee: []
 created_date: '2026-03-18 16:05'
-updated_date: '2026-03-23 03:22'
 labels: []
 dependencies: []
 references:
-  - >-
-    /home/sudacode/projects/japanese/SubMiner/src/core/services/subtitle-processing-controller.ts
-  - >-
-    /home/sudacode/projects/japanese/SubMiner/src/core/services/subtitle-prefetch.ts
+  - /home/sudacode/projects/japanese/SubMiner/src/core/services/subtitle-processing-controller.ts
+  - /home/sudacode/projects/japanese/SubMiner/src/core/services/subtitle-prefetch.ts
+documentation: []
 priority: high
-ordinal: 136500
 ---
 
 ## Description

backlog/tasks/task-197 - Eliminate-per-line-plain-subtitle-flash-on-prefetch-cache-hit.md

@@ -4,19 +4,15 @@ title: Eliminate per-line plain subtitle flash on prefetch cache hit
 status: Done
 assignee: []
 created_date: '2026-03-18 16:28'
-updated_date: '2026-03-23 03:22'
 labels: []
 dependencies:
   - TASK-196
 references:
-  - >-
-    /home/sudacode/projects/japanese/SubMiner/src/core/services/subtitle-processing-controller.ts
-  - >-
-    /home/sudacode/projects/japanese/SubMiner/src/main/runtime/mpv-main-event-actions.ts
-  - >-
-    /home/sudacode/projects/japanese/SubMiner/src/main/runtime/mpv-main-event-main-deps.ts
+  - /home/sudacode/projects/japanese/SubMiner/src/core/services/subtitle-processing-controller.ts
+  - /home/sudacode/projects/japanese/SubMiner/src/main/runtime/mpv-main-event-actions.ts
+  - /home/sudacode/projects/japanese/SubMiner/src/main/runtime/mpv-main-event-main-deps.ts
+documentation: []
 priority: high
-ordinal: 135500
 ---
 
 ## Description

backlog/tasks/task-199 - Forward-launcher-log-level-into-mpv-plugin-script-opts.md

@@ -4,7 +4,6 @@ title: Forward launcher log level into mpv plugin script opts
 status: Done
 assignee: []
 created_date: '2026-03-18 21:16'
-updated_date: '2026-03-23 03:22'
 labels: []
 dependencies:
   - TASK-198
@@ -13,8 +12,8 @@ references:
   - /home/sudacode/projects/japanese/SubMiner/launcher/mpv.ts
   - /home/sudacode/projects/japanese/SubMiner/launcher/main.test.ts
   - /home/sudacode/projects/japanese/SubMiner/launcher/aniskip-metadata.test.ts
+documentation: []
 priority: medium
-ordinal: 134500
 ---
 
 ## Description

backlog/tasks/task-200 - Address-latest-PR-19-CodeRabbit-follow-ups.md

@@ -5,7 +5,7 @@ status: Done
 assignee:
   - '@codex'
 created_date: '2026-03-19 07:18'
-updated_date: '2026-03-23 03:22'
+updated_date: '2026-03-19 07:28'
 labels:
   - pr-review
   - anki-integration
@@ -19,7 +19,6 @@ references:
   - src/anki-integration/runtime.ts
   - src/anki-integration/known-word-cache.ts
 priority: medium
-ordinal: 133500
 ---
 
 ## Description

backlog/tasks/task-201 - Suppress-repeated-macOS-overlay-loading-OSD-during-fullscreen-tracker-flaps.md

@@ -5,7 +5,7 @@ status: Done
 assignee:
   - '@codex'
 created_date: '2026-03-19 18:47'
-updated_date: '2026-03-23 03:22'
+updated_date: '2026-03-19 19:01'
 labels:
   - bug
   - macos
@@ -20,7 +20,6 @@ references:
   - >-
     /Users/sudacode/projects/japanese/SubMiner/src/core/services/overlay-visibility.test.ts
 priority: high
-ordinal: 131500
 ---
 
 ## Description

backlog/tasks/task-204.1 - Restore-stale-only-startup-known-word-cache-refresh.md

@@ -5,7 +5,7 @@ status: Done
 assignee:
   - '@Codex'
 created_date: '2026-03-20 02:52'
-updated_date: '2026-03-23 03:22'
+updated_date: '2026-03-20 03:02'
 labels:
   - anki
   - cache
@@ -17,7 +17,6 @@ references:
   - docs/plans/2026-03-19-known-word-cache-incremental-sync-design.md
 parent_task_id: TASK-204
 priority: high
-ordinal: 124500
 ---
 
 ## Description

backlog/tasks/task-205 - Address-PR-19-Claude-frontend-review-follow-ups.md

@@ -5,7 +5,7 @@ status: Done
 assignee:
   - codex
 created_date: '2026-03-20 02:41'
-updated_date: '2026-03-23 03:22'
+updated_date: '2026-03-20 02:46'
 labels: []
 milestone: m-1
 dependencies: []
@@ -14,7 +14,6 @@ references:
   - stats/src/hooks/useSessions.ts
   - stats/src/hooks/useTrends.ts
 priority: medium
-ordinal: 126500
 ---
 
 ## Description

backlog/tasks/task-206 - Assess-latest-PR-19-CodeRabbit-review-comments.md

@@ -5,7 +5,7 @@ status: Done
 assignee:
   - '@codex'
 created_date: '2026-03-20 02:51'
-updated_date: '2026-03-23 03:22'
+updated_date: '2026-03-20 02:59'
 labels:
   - pr-review
   - launcher
@@ -22,7 +22,6 @@ references:
   - src/anki-integration.ts
   - src/anki-integration/known-word-cache.ts
 priority: medium
-ordinal: 125500
 ---
 
 ## Description

backlog/tasks/task-207 - Verify-PR-19-follow-up-typecheck-blocker-is-cleared.md

@@ -5,7 +5,7 @@ status: Done
 assignee:
   - '@codex'
 created_date: '2026-03-20 03:03'
-updated_date: '2026-03-23 03:22'
+updated_date: '2026-03-20 03:04'
 labels:
   - pr-review
   - anki-integration
@@ -15,7 +15,6 @@ dependencies: []
 references:
   - src/anki-integration/anki-connect-proxy.test.ts
 priority: medium
-ordinal: 123500
 ---
 
 ## Description

backlog/tasks/task-208 - Assess-newest-PR-19-CodeRabbit-round-after-1227706.md

@@ -5,7 +5,7 @@ status: Done
 assignee:
   - '@codex'
 created_date: '2026-03-20 03:37'
-updated_date: '2026-03-23 03:22'
+updated_date: '2026-03-20 03:47'
 labels:
   - pr-review
   - launcher
@@ -17,7 +17,6 @@ references:
   - launcher/mpv.ts
   - src/anki-integration.ts
 priority: medium
-ordinal: 122500
 ---
 
 ## Description

backlog/tasks/task-209 - Exclude-grammar-tail-そうだ-from-subtitle-annotations.md

@@ -5,7 +5,7 @@ status: Done
 assignee:
   - codex
 created_date: '2026-03-20 04:06'
-updated_date: '2026-03-23 03:22'
+updated_date: '2026-03-20 04:33'
 labels:
   - bug
   - tokenizer
@@ -18,7 +18,6 @@ references:
   - >-
     /Users/sudacode/projects/japanese/SubMiner/src/core/services/tokenizer.test.ts
 priority: high
-ordinal: 120500
 ---
 
 ## Description

backlog/tasks/task-210 - Show-latest-session-position-in-anime-episode-progress.md

@@ -5,7 +5,7 @@ status: Done
 assignee:
   - '@Codex'
 created_date: '2026-03-20 04:09'
-updated_date: '2026-03-23 03:22'
+updated_date: '2026-03-20 04:25'
 labels:
   - stats
   - bug
@@ -17,7 +17,6 @@ references:
   - src/core/services/immersion-tracker/query.ts
   - src/core/services/immersion-tracker/session.ts
   - src/core/services/immersion-tracker-service.ts
-ordinal: 121500
 ---
 
 ## Description

backlog/tasks/task-211 - Recover-anime-episode-progress-from-subtitle-timing-when-checkpoints-are-missing.md

@@ -1,41 +0,0 @@
----
-id: TASK-211
-title: >-
-  Recover anime episode progress from subtitle timing when checkpoints are
-  missing
-status: Done
-assignee:
-  - '@Codex'
-created_date: '2026-03-20 10:15'
-updated_date: '2026-03-23 03:22'
-labels:
-  - stats
-  - bug
-milestone: m-1
-dependencies: []
-references:
-  - src/core/services/immersion-tracker/query.ts
-  - src/core/services/immersion-tracker/__tests__/query.test.ts
-ordinal: 119500
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-Anime episode progress can still show `0%` for older sessions that have watch-time and subtitle timing but no persisted `ended_media_ms` checkpoint. Recover progress from the latest retained subtitle/event segment end so already-recorded sessions render a useful progress percentage.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 `getAnimeEpisodes` returns the latest known session position even when `ended_media_ms` is null but subtitle/event timing exists.
-- [x] #2 Existing ended-session metrics and aggregation totals do not regress.
-- [x] #3 Regression coverage locks the fallback behavior.
-<!-- AC:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-Added a query-side fallback for anime episode progress: when the newest session for a video has no persisted `ended_media_ms`, `getAnimeEpisodes` now uses the latest retained subtitle-line or session-event `segment_end_ms` from that same session. This recovers useful progress for already-recorded sessions that have timing data but predate or missed checkpoint persistence.
-
-Verification: `bun test src/core/services/immersion-tracker/__tests__/query.test.ts` passed. `bun run typecheck` passed.
-<!-- SECTION:NOTES:END -->

backlog/tasks/task-212 - Fix-mac-texthooker-helper-startup-blocking-mpv-launch.md

@@ -1,44 +0,0 @@
----
-id: TASK-212
-title: Fix mac texthooker helper startup blocking mpv launch
-status: Done
-assignee: []
-created_date: '2026-03-20 08:27'
-updated_date: '2026-03-23 03:22'
-labels:
-  - bug
-  - macos
-  - startup
-dependencies: []
-references:
-  - /Users/sudacode/projects/japanese/SubMiner/src/core/services/startup.ts
-  - /Users/sudacode/projects/japanese/SubMiner/src/main.ts
-  - /Users/sudacode/projects/japanese/SubMiner/plugin/subminer/process.lua
-priority: high
-ordinal: 140500
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-`subminer` mpv auto-start on mac can stall before the video is usable because the helper process launched with `--texthooker` still runs heavy app-ready startup. Recent logs show the helper loading the Yomitan Chromium extension, emitting `Permission 'contextMenus' is unknown` warnings, then hitting Chromium runtime errors before SubMiner signals readiness back to the mpv plugin. The texthooker helper should take the minimal startup path needed to serve texthooker traffic without loading overlay/window-only startup work that can crash or delay readiness.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 Launching SubMiner with `--texthooker` avoids heavy app-ready startup work that is not required for texthooker helper mode.
-- [x] #2 A regression test covers texthooker helper startup so it fails if Yomitan extension loading is reintroduced on that path.
-- [x] #3 The change preserves existing startup behavior for non-texthooker app launches.
-<!-- AC:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-Follow-up: user confirmed the root issue is the plugin auto-start ordering. Adjust mpv plugin sequencing so `--start` launches before any separate `--texthooker` helper, then verify plugin regressions still pass.
-<!-- SECTION:NOTES:END -->
-
-## Final Summary
-
-<!-- SECTION:FINAL_SUMMARY:BEGIN -->
-Fixed the mac mpv startup hang caused by the `--texthooker` helper taking the full app-ready path. `runAppReadyRuntime` now fast-paths texthooker-only mode through minimal startup (`reloadConfig` plus CLI handling) so it no longer loads Yomitan or first-run setup work before serving texthooker traffic. Added regression coverage in `src/core/services/app-ready.test.ts`, then verified with `bun test src/core/services/app-ready.test.ts src/core/services/startup.test.ts`, `bun test src/cli/args.test.ts src/main/early-single-instance.test.ts src/main/runtime/stats-cli-command.test.ts`, and `bun run typecheck`.
-<!-- SECTION:FINAL_SUMMARY:END -->

backlog/tasks/task-213 - Show-character-dictionary-progress-during-paused-startup-waits.md

@@ -1,43 +0,0 @@
----
-id: TASK-213
-title: Show character dictionary progress during paused startup waits
-status: Done
-assignee: []
-created_date: '2026-03-20 08:59'
-updated_date: '2026-03-23 03:22'
-labels:
-  - bug
-  - ux
-  - dictionary
-  - startup
-dependencies: []
-references:
-  - >-
-    /Users/sudacode/projects/japanese/SubMiner/src/main/runtime/startup-osd-sequencer.ts
-  - >-
-    /Users/sudacode/projects/japanese/SubMiner/src/main/runtime/character-dictionary-auto-sync-notifications.ts
-  - /Users/sudacode/projects/japanese/SubMiner/src/main.ts
-priority: medium
-ordinal: 141500
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-During startup on mpv auto-start, character dictionary regeneration/update can be active while playback remains paused. The current startup OSD sequencer buffers dictionary progress behind annotation-loading OSD, which leaves the user with no visible dictionary-specific progress while the pause is active. Adjust the startup OSD sequencing so dictionary progress can surface once tokenization is ready during the paused startup window, without regressing later ready/failure handling.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [ ] #1 When tokenization is ready during startup, later character dictionary progress updates are shown on OSD even if annotation-loading state is still active.
-- [ ] #2 Startup OSD completion/failure behavior for character dictionary sync remains coherent after the new progress ordering.
-- [ ] #3 Regression coverage exercises the paused startup sequencing for dictionary progress.
-<!-- AC:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-2026-03-20: Confirmed issue is broader than OSD-only. Paused-startup OSD fixes remain relevant, but current user report also points at a regression in non-blocking startup playback release (tracked in TASK-143).
-
-2026-03-20: OSD sequencing fix remains in local patch alongside TASK-143 regression fix. Covered by startup-osd-sequencer tests; pending installed-app/mpv validation before task finalization.
-<!-- SECTION:NOTES:END -->

backlog/tasks/task-214 - Jump-subtitle-sidebar-directly-to-resume-position-on-first-resolved-cue.md

@@ -1,40 +0,0 @@
----
-id: TASK-214
-title: Jump subtitle sidebar directly to resume position on first resolved cue
-status: Done
-assignee: []
-created_date: '2026-03-21 11:15'
-updated_date: '2026-03-23 03:22'
-labels:
-  - bug
-  - ux
-  - overlay
-  - subtitles
-dependencies: []
-references:
-  - >-
-    /Users/sudacode/projects/japanese/SubMiner/src/renderer/modals/subtitle-sidebar.ts
-  - >-
-    /Users/sudacode/projects/japanese/SubMiner/src/renderer/modals/subtitle-sidebar.test.ts
-priority: medium
-ordinal: 142500
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-When playback starts from a resumed timestamp while the subtitle sidebar is open, the sidebar currently smooth-scrolls from the top of the cue list to the resumed cue. Change the first resolved active-cue positioning to jump immediately to the resume location while preserving smooth auto-follow for later playback-driven cue advances.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 The first active cue resolved after open/resume uses an instant jump instead of smooth-scrolling through the list.
-- [x] #2 Normal subtitle-sidebar auto-follow remains smooth after the first active cue has been positioned.
-- [x] #3 Regression coverage distinguishes the initial jump behavior from later smooth auto-follow updates.
-<!-- AC:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-2026-03-21: Fixed by treating the first auto-scroll from `previousActiveCueIndex < 0` as `behavior: 'auto'` in the subtitle sidebar scroll helper. Added renderer regression coverage for initial jump plus later smooth follow.
-<!-- SECTION:NOTES:END -->

backlog/tasks/task-215 - Add-startup-auto-open-option-for-subtitle-sidebar.md

@@ -1,44 +0,0 @@
----
-id: TASK-215
-title: Add startup auto-open option for subtitle sidebar
-status: Done
-assignee: []
-created_date: '2026-03-21 11:35'
-updated_date: '2026-03-23 03:22'
-labels:
-  - feature
-  - ux
-  - overlay
-  - subtitles
-dependencies: []
-references:
-  - /Users/sudacode/projects/japanese/SubMiner/src/types.ts
-  - >-
-    /Users/sudacode/projects/japanese/SubMiner/src/config/definitions/defaults-subtitle.ts
-  - >-
-    /Users/sudacode/projects/japanese/SubMiner/src/config/resolve/subtitle-domains.ts
-  - >-
-    /Users/sudacode/projects/japanese/SubMiner/src/renderer/modals/subtitle-sidebar.ts
-  - /Users/sudacode/projects/japanese/SubMiner/src/renderer/renderer.ts
-priority: medium
-ordinal: 143500
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-Add a subtitle sidebar config option that auto-opens the sidebar once during overlay startup. The option should default to `false`, only apply when the sidebar feature is enabled, and should not force the sidebar back open later in the same session after manual close or later visibility changes.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 `subtitleSidebar.autoOpen` is available in config with default `false`.
-- [x] #2 When enabled, overlay startup opens the subtitle sidebar once after initial sidebar config/snapshot load.
-- [x] #3 Regression coverage covers config resolution and startup-only auto-open behavior.
-<!-- AC:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-2026-03-21: Added `subtitleSidebar.autoOpen` to types/defaults/config registry and resolver. Renderer bootstrap now calls a startup-only subtitle sidebar helper after the initial snapshot refresh. Modal regression coverage verifies startup auto-open requires both `enabled` and `autoOpen`.
-<!-- SECTION:NOTES:END -->

backlog/tasks/task-216 - Address-PR-28-CodeRabbit-follow-ups-on-subtitle-sidebar.md

@@ -1,79 +0,0 @@
----
-id: TASK-216
-title: 'Address PR #28 CodeRabbit follow-ups on subtitle sidebar'
-status: Completed
-assignee:
-  - '@codex'
-created_date: '2026-03-21 00:00'
-updated_date: '2026-03-21 00:00'
-labels:
-  - pr-review
-  - subtitle-sidebar
-  - renderer
-dependencies: []
-references:
-  - src/main/runtime/subtitle-prefetch-init.ts
-  - src/main/runtime/subtitle-prefetch-init.test.ts
-  - src/renderer/handlers/mouse.ts
-  - src/renderer/handlers/mouse.test.ts
-  - src/renderer/modals/subtitle-sidebar.ts
-  - src/renderer/modals/subtitle-sidebar.test.ts
-  - src/renderer/style.css
-priority: medium
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-Validate the CodeRabbit follow-ups on PR #28 for the subtitle sidebar workstream, implement the confirmed fixes, and verify the touched runtime and renderer paths.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 Review comments that described real regressions are fixed in code
-- [x] #2 Focused regression coverage exists for the fixed behaviors
-- [x] #3 Targeted typecheck and runtime-compat verification pass
-<!-- AC:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-Completed follow-up fixes for PR #28:
-- Cleared parsed subtitle cues on subtitle prefetch init failure so stale snapshot cache entries do not survive a failed refresh.
-- Treated primary and secondary subtitle containers as one hover region so moving between them does not resume playback mid-transition.
-- Kept the subtitle sidebar closed when disabled, serialized snapshot polling with timeouts, made cue rows keyboard-activatable, resolved stale cue selection fallback, and resumed hover-paused playback when the modal closes.
-
-Regression coverage added:
-- `src/main/runtime/subtitle-prefetch-init.test.ts`
-- `src/renderer/handlers/mouse.test.ts`
-- `src/renderer/modals/subtitle-sidebar.test.ts`
-
-Verification:
-- `bun test src/main/runtime/subtitle-prefetch-init.test.ts`
-- `bun test src/renderer/handlers/mouse.test.ts`
-- `bun test src/renderer/modals/subtitle-sidebar.test.ts`
-- `bun run typecheck`
-- `bun run test:runtime:compat`
-
-2026-03-21: Reopened to assess a newer CodeRabbit review pass on PR #28 and address any remaining valid action items before push/reply.
-
-2026-03-21: Addressed the latest CodeRabbit follow-up pass in commit d70c6448 after rebasing onto the updated remote branch tip.
-
-2026-03-21: Reopened for the latest CodeRabbit round on commit d70c6448; current actionable item is the invalid ctx.state.isOverSubtitleSidebar assignment in subtitle-sidebar.ts.
-
-2026-03-22: Addressed the live hover-state and startup mouse-ignore follow-ups from the latest CodeRabbit pass. `handleMouseLeave()` now clears `isOverSubtitle` and drops `secondary-sub-hover-active` when leaving the secondary subtitle container toward the primary container, and renderer startup now calls `syncOverlayMouseIgnoreState(ctx)` instead of forcing `setIgnoreMouseEvents(true, { forward: true })`. The sidebar IPC hover catch and CSS spacing comments were already satisfied in the current tree.
-
-2026-03-22: Regenerated `bun.lock` from a clean install so the `electron-builder-squirrel-windows` override now resolves at `26.8.2` in the lockfile alongside `app-builder-lib@26.8.2`.
-
-2026-03-21: Finished the remaining cleanup pass from the latest review. `subtitleSidebar.layout` now uses enum validation, `SubtitleCue` is re-exported from `src/types.ts` as the single public type path, and the subtitle sidebar resize listener now has unload cleanup wired through the renderer.
-<!-- SECTION:NOTES:END -->
-
-## Final Summary
-
-<!-- SECTION:FINAL_SUMMARY:BEGIN -->
-Implemented the confirmed PR #28 CodeRabbit follow-ups for subtitle sidebar behavior and added regression coverage plus verification for the touched renderer and runtime paths.
-
-Handled the latest CodeRabbit review pass for PR #28: accepted zero sidebar opacity, closed/inerted the sidebar when refresh sees config disabled, moved poll rescheduling out of finally, caught hover pause IPC failures, and fixed the stylelint spacing issue.
-
-Verification: bun test src/config/resolve/subtitle-sidebar.test.ts; bun test src/renderer/modals/subtitle-sidebar.test.ts; bun test src/renderer/handlers/mouse.test.ts; bun run typecheck; bun run test:fast; bun run test:env; bun run build; SubMiner verifier lanes config + runtime-compat (including test:runtime:compat and test:smoke:dist).
-<!-- SECTION:FINAL_SUMMARY:END -->

backlog/tasks/task-217 - Fix-embedded-overlay-passthrough-sync-between-subtitle-and-sidebar.md

@@ -1,70 +0,0 @@
----
-id: TASK-217
-title: Fix embedded overlay passthrough sync between subtitle and sidebar
-status: Done
-assignee:
-  - codex
-created_date: '2026-03-21 23:16'
-updated_date: '2026-03-23 03:22'
-labels:
-  - bug
-  - overlay
-  - macos
-dependencies: []
-references:
-  - src/renderer/handlers/mouse.ts
-  - src/renderer/modals/subtitle-sidebar.ts
-  - src/renderer/renderer.ts
-documentation:
-  - docs/workflow/verification.md
-priority: high
-ordinal: 118500
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-On macOS, when both the subtitle overlay and embedded subtitle sidebar are visible, mouse passthrough to mpv can remain stale until the user hovers the sidebar. After closing the sidebar, passthrough can likewise remain stale until the user hovers the subtitle again. Fix the overlay input-state synchronization so passthrough reflects the current hover/open state immediately instead of relying on the last hover target.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 When the embedded subtitle sidebar is open and the pointer is not over subtitle or sidebar content, the overlay returns to mouse passthrough immediately without requiring a sidebar hover cycle.
-- [x] #2 When transitioning between subtitle hover and sidebar hover states on macOS embedded sidebar mode, mouse ignore state stays in sync with the currently interactive region.
-- [x] #3 Closing the embedded subtitle sidebar restores the correct passthrough state based on remaining subtitle hover/modal state without requiring an additional hover.
-- [x] #4 Regression tests cover the passthrough synchronization behavior.
-<!-- AC:END -->
-
-## Implementation Plan
-
-<!-- SECTION:PLAN:BEGIN -->
-1. Add a shared renderer-side passthrough sync helper that derives whether the overlay should ignore mouse events from subtitle hover, embedded sidebar visibility/hover, popup visibility, and modal state.
-2. Replace direct embedded-sidebar passthrough toggles in subtitle hover/sidebar handlers with calls to the shared sync helper so state is recomputed on every transition.
-3. Add regression tests for macOS embedded sidebar mode covering sidebar-open idle passthrough, subtitle-to-sidebar transitions, and sidebar-close restore behavior.
-4. Run targeted renderer tests for mouse/sidebar passthrough coverage, then summarize any residual risk.
-<!-- SECTION:PLAN:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-Added shared renderer overlay mouse-ignore recompute so subtitle hover, embedded sidebar hover/open/close, and popup idle transitions all derive passthrough from current state instead of last hover target.
-
-Added regression coverage for embedded sidebar idle passthrough on subtitle leave and for sidebar-close recompute behavior.
-
-Verification: `bun run typecheck` passed; `bun test src/renderer/handlers/mouse.test.ts` passed; `bun test src/renderer/modals/subtitle-sidebar.test.ts` passed; core verification wrapper artifact at `.tmp/skill-verification/subminer-verify-20260321-162743-XhSBxw` hit an unrelated `bun run test:fast` failure in `scripts/update-aur-package.test.ts` because macOS system bash lacks `mapfile`.
-<!-- SECTION:NOTES:END -->
-
-## Final Summary
-
-<!-- SECTION:FINAL_SUMMARY:BEGIN -->
-Fixed stale embedded-sidebar passthrough sync on macOS by introducing a shared renderer mouse-ignore recompute path and tracking sidebar-hover state separately from subtitle hover. Subtitle hover leave, sidebar hover enter/leave, sidebar open, and sidebar close now all recompute passthrough from the current overlay state instead of waiting for a later hover event to repair it. Added regression tests covering subtitle-leave passthrough while the embedded sidebar is open but idle, plus sidebar-close restore behavior based on remaining subtitle hover state.
-
-Tests run:
-- `bun run typecheck`
-- `bun test src/renderer/handlers/mouse.test.ts`
-- `bun test src/renderer/modals/subtitle-sidebar.test.ts`
-- `bash .agents/skills/subminer-change-verification/scripts/classify_subminer_diff.sh src/renderer/state.ts src/renderer/overlay-mouse-ignore.ts src/renderer/handlers/mouse.ts src/renderer/handlers/mouse.test.ts src/renderer/modals/subtitle-sidebar.ts src/renderer/modals/subtitle-sidebar.test.ts`
-- `bash .agents/skills/subminer-change-verification/scripts/verify_subminer_change.sh --lane core src/renderer/state.ts src/renderer/overlay-mouse-ignore.ts src/renderer/handlers/mouse.ts src/renderer/handlers/mouse.test.ts src/renderer/modals/subtitle-sidebar.ts src/renderer/modals/subtitle-sidebar.test.ts` (typecheck passed; `test:fast` blocked by unrelated `scripts/update-aur-package.test.ts` failure on macOS Bash 3.2 lacking `mapfile`)
-
-Risk: the classifier flagged this as a real-runtime candidate, so actual Electron/mpv macOS pointer behavior was not exercised in a live runtime during this turn.
-<!-- SECTION:FINAL_SUMMARY:END -->

backlog/tasks/task-218 - Delete-zero-session-media-from-stats-library-and-trends.md

@@ -1,69 +0,0 @@
----
-id: TASK-218
-title: Delete zero-session media from stats library and trends
-status: Done
-assignee:
-  - codex
-created_date: '2026-03-22 16:20'
-updated_date: '2026-03-24 06:41'
-labels:
-  - stats
-  - immersion-tracker
-dependencies: []
-references:
-  - >-
-    /Users/sudacode/projects/japanese/SubMiner/src/core/services/immersion-tracker/query.ts
-  - >-
-    /Users/sudacode/projects/japanese/SubMiner/src/core/services/immersion-tracker/lifetime.ts
-  - >-
-    /Users/sudacode/projects/japanese/SubMiner/src/core/services/immersion-tracker/maintenance.ts
-  - >-
-    /Users/sudacode/projects/japanese/SubMiner/src/core/services/immersion-tracker/__tests__/query.test.ts
-priority: medium
-ordinal: 153500
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-Deleting the last retained session for a video still left stale lifetime media rows and trend rollups behind, so the stats dashboard could continue showing ghost entries in Library and Trends after all sessions were gone.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 Deleting the final session for a video removes that media from Library queries and detail reads
-- [x] #2 Deleting the final session for a video removes stale daily/monthly trend rollups for that media
-- [x] #3 Regression coverage proves zero-session media disappears from affected stats surfaces after deletion
-<!-- AC:END -->
-
-## Implementation Plan
-
-<!-- SECTION:PLAN:BEGIN -->
-1. Add a failing regression around deleting the only retained session for a video while preexisting lifetime and rollup rows exist.
-2. Patch the deletion path to rebuild lifetime and rollup state from retained sessions inside the same transaction.
-3. Run focused immersion-tracker tests plus the repo-native verifier core lane and record results.
-<!-- SECTION:PLAN:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-Added a query regression that seeds a finished session plus stale lifetime media/anime rows and daily/monthly rollups, deletes that only session, and asserts Library, Anime detail, and Trends all drop the media immediately.
-
-Refactored lifetime rebuild logic so it can run inside an existing delete transaction, then reused that helper from `deleteSession`, `deleteSessions`, and `deleteVideo`.
-
-Added a rollup rebuild helper that clears existing daily/monthly rollups and reconstructs them from retained telemetry inside the current transaction so deleted sessions cannot leave ghost trend points behind.
-
-Verification passed:
-- `bun test src/core/services/immersion-tracker/__tests__/query.test.ts`
-- `bun test src/core/services/immersion-tracker-service.test.ts`
-- `bash .agents/skills/subminer-change-verification/scripts/classify_subminer_diff.sh src/core/services/immersion-tracker/query.ts src/core/services/immersion-tracker/lifetime.ts src/core/services/immersion-tracker/maintenance.ts src/core/services/immersion-tracker/__tests__/query.test.ts`
-- `bash .agents/skills/subminer-change-verification/scripts/verify_subminer_change.sh --lane core src/core/services/immersion-tracker/query.ts src/core/services/immersion-tracker/lifetime.ts src/core/services/immersion-tracker/maintenance.ts src/core/services/immersion-tracker/__tests__/query.test.ts`
-
-Verifier artifact dir: `.tmp/skill-verification/subminer-verify-20260322-210718-n6sGL8`
-<!-- SECTION:NOTES:END -->
-
-## Final Summary
-
-<!-- SECTION:FINAL_SUMMARY:BEGIN -->
-Delete paths now rebuild lifetime summaries and trend rollups after removing sessions, so when the last session for a video disappears the stats database also drops that media from Library, related detail reads, and chart data. Added a regression proving a video with only stale lifetime/rollup rows vanishes after its final session is deleted, and verified the change with focused immersion-tracker tests plus the SubMiner core verification lane.
-<!-- SECTION:FINAL_SUMMARY:END -->

backlog/tasks/task-219 - Restore-streamed-video-progress-in-anime-episodes.md

@@ -1,43 +0,0 @@
----
-id: TASK-219
-title: Restore streamed video progress in anime episodes
-status: Done
-assignee:
-  - codex
-created_date: '2026-03-22 21:25'
-updated_date: '2026-03-24 06:44'
-labels:
-  - stats
-  - immersion-tracker
-  - youtube
-dependencies: []
-references:
-  - >-
-    /Users/sudacode/projects/japanese/SubMiner/src/core/services/immersion-tracker/query.ts
-  - >-
-    /Users/sudacode/projects/japanese/SubMiner/src/core/services/immersion-tracker-service.ts
-  - >-
-    /Users/sudacode/projects/japanese/SubMiner/src/core/services/immersion-tracker/__tests__/query.test.ts
-  - >-
-    /Users/sudacode/projects/japanese/SubMiner/src/core/services/immersion-tracker-service.test.ts
-priority: medium
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-Episode progress for streamed media can stay at `0%` because some remote sessions persist `ended_media_ms = 0` even when subtitle timing and watch activity clearly advanced, and the anime episode query currently treats `0` as a valid progress checkpoint.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 Anime episode progress ignores zero-valued session checkpoints and falls back to subtitle/event timing
-- [x] #2 New streamed sessions persist meaningful progress even when playback-position updates are missing or sparse
-- [x] #3 Regression tests cover the zero-checkpoint remote-session case
-<!-- AC:END -->
-
-## Final Summary
-
-<!-- SECTION:FINAL_SUMMARY:BEGIN -->
-Restored anime episode progress handling for streamed sessions by ignoring zero-valued `ended_media_ms` checkpoints and falling back to subtitle/event timing, with regression coverage for the remote-session zero-checkpoint case.
-<!-- SECTION:FINAL_SUMMARY:END -->

backlog/tasks/task-220 - Restore-YouTube-overlay-mpv-keybindings-after-picker-routing.md

@@ -1,66 +0,0 @@
----
-id: TASK-220
-title: Restore YouTube overlay mpv keybindings after picker routing
-status: Done
-assignee:
-  - codex
-created_date: '2026-03-22 00:00'
-updated_date: '2026-03-22 23:49'
-labels:
-  - bug
-  - overlay
-  - youtube
-  - keyboard
-dependencies: []
-references:
-  - src/renderer/handlers/keyboard.ts
-  - src/renderer/modals/youtube-track-picker.ts
-  - src/renderer/handlers/keyboard.test.ts
-  - src/renderer/modals/youtube-track-picker.test.ts
-documentation:
-  - docs/workflow/verification.md
-priority: high
-ordinal: 118800
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-Regression: after adding the YouTube subtitle picker modal path, visible-overlay keydown handling can stop before reaching the shared mpv keybinding dispatch path. Result: default overlay mpv bindings like `Space` pause/play and `q` quit stop working while the overlay owns focus during YouTube playback.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 Unhandled keys while the YouTube track picker state is active still fall through to the shared overlay mpv keybinding dispatcher.
-- [x] #2 The YouTube picker continues to consume `Enter` and `Escape` for its own actions.
-- [x] #3 Renderer regression tests cover both the picker modal key contract and the shared keyboard dispatch fallback.
-<!-- AC:END -->
-
-## Implementation Plan
-
-<!-- SECTION:PLAN:BEGIN -->
-1. Add a failing renderer keyboard regression test covering YouTube picker state plus shared mpv keybinding fallback.
-2. Update the global keyboard handler to return early only when the YouTube picker actually handles the key event.
-3. Update the picker modal handler to return false for unhandled keys while preserving `Enter`/`Escape`.
-4. Run the cheap renderer verification lane and record results.
-<!-- SECTION:PLAN:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-Fixed the regression by making the global renderer keyboard handler stop early for the YouTube picker only when the picker actually consumes the key. The picker modal now returns `false` for unrelated keys, so shared overlay mpv bindings like `Space` and `KeyQ` still dispatch while the visible overlay has focus.
-
-Added regression coverage in the keyboard handler suite for mpv keybinding fallback during YouTube picker state, plus a picker-modal contract test that keeps `Escape` handled but leaves unrelated keys unclaimed.
-
-Verification:
-- `bun test src/renderer/handlers/keyboard.test.ts src/renderer/modals/youtube-track-picker.test.ts`
-- `bash .agents/skills/subminer-change-verification/scripts/classify_subminer_diff.sh src/renderer/handlers/keyboard.ts src/renderer/handlers/keyboard.test.ts src/renderer/modals/youtube-track-picker.ts src/renderer/modals/youtube-track-picker.test.ts`
-- `bash .agents/skills/subminer-change-verification/scripts/verify_subminer_change.sh --lane core src/renderer/handlers/keyboard.ts src/renderer/handlers/keyboard.test.ts src/renderer/modals/youtube-track-picker.ts src/renderer/modals/youtube-track-picker.test.ts`
-- verifier artifact: `.tmp/skill-verification/subminer-verify-20260322-234831-b2m6nJ`
-<!-- SECTION:NOTES:END -->
-
-## Final Summary
-
-<!-- SECTION:FINAL_SUMMARY:BEGIN -->
-Restored YouTube-session overlay mpv keybindings by removing an unconditional early return added to the renderer keyboard path for the YouTube subtitle picker modal. Unhandled keys now fall through to the shared mpv keybinding dispatcher, while handled picker keys (`Enter`, `Escape`) still stay local to the picker. Added renderer regression tests for both the keyboard fallback path and the picker modal key-consumption contract.
-<!-- SECTION:FINAL_SUMMARY:END -->

backlog/tasks/task-221 - Assess-and-address-PR-31-latest-CodeRabbit-review.md

@@ -1,58 +0,0 @@
----
-id: TASK-221
-title: 'Assess and address PR #31 latest CodeRabbit review'
-status: Done
-assignee: []
-created_date: '2026-03-23 07:53'
-updated_date: '2026-03-24 06:41'
-labels:
-  - pr-review
-  - coderabbit
-dependencies: []
-references:
-  - >-
-    PR #31 feat: add app-owned YouTube subtitle flow with absPlayer-style
-    parsing
-priority: medium
-ordinal: 152500
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-Inspect the latest CodeRabbit review on PR #31, evaluate each actionable comment against the current branch, implement valid fixes, verify the changes, and prepare PR thread updates.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Implementation Plan
-
-<!-- SECTION:PLAN:BEGIN -->
-1. Inspect latest CodeRabbit review on PR #31 and separate valid action items from non-blocking suggestions.
-2. Add regression coverage for any real bugs before changing production code.
-3. Implement the minimal fixes for confirmed issues in runtime, renderer modal flow, and test fixtures.
-4. Run targeted tests plus repo-native verification lanes.
-5. Update PR threads with fix status and rationale for any comments not actioned yet.
-<!-- SECTION:PLAN:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-Implemented and pushed commit 207151db to PR #31.
-
-Replied in-thread to the CodeRabbit comments for YouTube host matching, duplicate picker submissions, and missing MediaDetailView test fixture videoId fields.
-
-Follow-up scope added: update release-facing docs/changelog for the YouTube subtitle picker work and run a release-readiness gate before handoff.
-
-Added release-facing docs/changelog updates in commit b7e0026d and pushed them to PR #31.
-
-Ran the release-readiness gate: changelog:lint, changelog:pr-check, verify:config-example, typecheck, test:fast, test:env, build, test:smoke:dist, docs:test, docs:build.
-<!-- SECTION:NOTES:END -->
-
-## Final Summary
-
-<!-- SECTION:FINAL_SUMMARY:BEGIN -->
-Assessed the latest CodeRabbit review on PR #31 and applied the confirmed fixes. Tightened `isYoutubeMediaPath()` to match only exact YouTube hosts or subdomains with a regression test for `notyoutube.com`, added an in-flight guard plus temporary control disabling to the YouTube track picker with a duplicate-submit regression test, replaced the picker empty-state `innerHTML` fallback with explicit DOM construction, and added the missing `videoId` fields to the `MediaDetailView` test fixtures. Verified with targeted Bun tests and the `runtime-compat` verification lane (`build`, `test:runtime:compat`, `test:smoke:dist`).
-
-Updated `README.md`, `docs-site/usage.md`, and `changes/2026-03-23-immersion-youtube.md` so the PR is release-facing and user-visible surfaces describe the YouTube subtitle picker flow plus its latest hardening.
-
-Release-readiness checks passed locally: `bun run changelog:lint`, `bun run changelog:pr-check`, `bun run verify:config-example`, `bun run typecheck`, `bun run test:fast`, `bun run test:env`, `bun run build`, `bun run test:smoke:dist`, `bun run docs:test`, and `bun run docs:build`.
-<!-- SECTION:FINAL_SUMMARY:END -->

backlog/tasks/task-222 - Fix-YouTube-overlay-keybindings-in-subtitle-path.md

@@ -1,56 +0,0 @@
----
-id: TASK-222
-title: Fix YouTube overlay keybindings in subtitle path
-status: Done
-assignee:
-  - codex
-created_date: '2026-03-23 08:32'
-updated_date: '2026-03-24 06:41'
-labels:
-  - bug
-dependencies: []
-references:
-  - /Users/sudacode/projects/japanese/SubMiner/src/main/runtime
-  - /Users/sudacode/projects/japanese/SubMiner/src/core/services
-priority: high
-ordinal: 151500
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-Users watching video through the YouTube subtitle path cannot use some overlay keyboard controls such as quit and pause/play. Restore expected overlay keybinding behavior for that playback path without regressing other overlay input handling.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 Overlay quit and pause/play keybindings work while using the YouTube subtitle path.
-- [x] #2 Existing overlay keybinding behavior for non-YouTube playback remains unchanged.
-- [x] #3 Regression coverage exercises the YouTube subtitle path keyboard handling.
-<!-- AC:END -->
-
-## Implementation Plan
-
-<!-- SECTION:PLAN:BEGIN -->
-1. Add a regression test around YouTube track-picker close to verify it requests main-process main-window focus restoration before returning overlay focus locally.
-2. Update the YouTube track-picker close flow to call `window.electronAPI.focusMainWindow()` alongside the existing `window.focus()` and `overlay.focus()` restoration.
-3. Run targeted tests for the picker/keyboard paths to verify YouTube playback regains overlay keybindings without regressing existing overlay behavior.
-<!-- SECTION:PLAN:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-Investigated overlay input path. Renderer already maps Space/KeyQ to mpv commands, but YouTube track-picker close only restores DOM focus (`window.focus` + `overlay.focus`) and does not invoke main-process window focus recovery, unlike the keyboard-mode focus reclaim path. Suspected root cause: overlay BrowserWindow focus is not restored after the YouTube picker closes, so playback keybindings stop reaching renderer keydown handlers.
-
-User approved implementation plan on 2026-03-23. Proceeding with TDD: add failing regression first, then minimal fix, then targeted verification.
-
-Implemented fix in the YouTube track-picker close path: request main-process `focusMainWindow()` before restoring renderer window/overlay focus so overlay keydown handlers regain input after YouTube subtitle selection.
-
-Verification: `bun test src/renderer/modals/youtube-track-picker.test.ts` and `bun test src/renderer/handlers/keyboard.test.ts` both pass.
-<!-- SECTION:NOTES:END -->
-
-## Final Summary
-
-<!-- SECTION:FINAL_SUMMARY:BEGIN -->
-Restored overlay keyboard focus after closing the YouTube subtitle picker by invoking the main-process `focusMainWindow()` recovery path before local window/overlay focus restoration. Added regression coverage to the YouTube picker modal test and verified existing keyboard handler coverage for YouTube picker passthrough keys (`Space`, `KeyQ`) remains green.
-<!-- SECTION:FINAL_SUMMARY:END -->

backlog/tasks/task-223 - Fix-YouTube-overlay-Anki-initialization-regression.md

@@ -1,109 +0,0 @@
----
-id: TASK-223
-title: Fix YouTube overlay Anki initialization regression
-status: Done
-assignee:
-  - codex
-created_date: '2026-03-23 08:41'
-updated_date: '2026-03-24 06:41'
-labels:
-  - bug
-  - youtube
-  - anki
-dependencies: []
-references:
-  - /Users/sudacode/projects/japanese/SubMiner/src/main.ts
-  - >-
-    /Users/sudacode/projects/japanese/SubMiner/src/core/services/overlay-runtime-init.ts
-  - >-
-    /Users/sudacode/projects/japanese/SubMiner/src/main/runtime/cli-command-runtime-handler.ts
-documentation:
-  - /Users/sudacode/projects/japanese/SubMiner/docs/workflow/verification.md
-priority: high
-ordinal: 154500
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-Restore Anki-backed lookup and known-word behavior during YouTube playback. Recent startup changes appear to let the YouTube flow initialize the overlay before runtime prerequisites exist, leaving the Anki integration unavailable for popup Mine actions and known-word highlighting.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 YouTube playback initializes Anki integration once overlay startup prerequisites are available so lookup can offer card-add actions again
-- [x] #2 Known-word / N+1 state is available during YouTube playback when the user has Anki-backed known-word highlighting enabled
-- [x] #3 Regression coverage fails before the fix and passes after it for the YouTube startup path
-<!-- AC:END -->
-
-## Implementation Plan
-
-<!-- SECTION:PLAN:BEGIN -->
-1. Add a regression test covering the YouTube playback command path and assert overlay startup prerequisites are established before the flow runs.
-2. Reuse the overlay startup prerequisite bootstrap for the YouTube playback path so Anki integration sees subtitle tracker, mpv client, and runtime options manager before initialization.
-3. Verify with focused runtime/CLI tests, then run the cheapest sufficient verification lane for the touched files.
-<!-- SECTION:PLAN:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-Identified regression path in CLI command runtime: YouTube playback commands could reach overlay initialization without first materializing overlay startup prerequisites, leaving Anki integration unavailable during the initial startup attempt.
-
-Added a regression test at src/main/runtime/cli-command-runtime-handler.test.ts covering youtubePlay command dispatch outside texthooker-only mode.
-
-Verified with bun test src/main/runtime/cli-command-runtime-handler.test.ts src/main/runtime/cli-command-prechecks.test.ts src/main/runtime/cli-command-prechecks-main-deps.test.ts and bun test src/core/services/cli-command.test.ts src/main/runtime/cli-command-context-main-deps.test.ts src/main/runtime/cli-command-prechecks-main-deps.test.ts src/main/runtime/cli-command-runtime-handler.test.ts.
-
-Removed the YouTube-only ensureOverlayRuntimeReady path from src/main.ts after confirming regular app startup already loads Yomitan and the shared CLI overlay pre-dispatch bootstrap now covers overlay prerequisites.
-
-Moved overlay bootstrap into the generic initial-args startup path for any initial command that needs overlay runtime, so overlay prerequisites and overlay initialization happen before CLI dispatch instead of inside a YouTube-only or last-moment command path.
-
-Additional verification passed: bun test src/main/runtime/initial-args-runtime-handler.test.ts src/main/runtime/initial-args-handler.test.ts src/main/runtime/initial-args-main-deps.test.ts, bun run typecheck, bun run test:runtime:compat
-
-Follow-up regression: subtitle picker was still auto-submitting the default selection during YouTube startup. Investigating renderer-side immediate Enter key bleed-through on picker open.
-
-Root cause for remaining picker regression: the YouTube track picker accepted Enter immediately on open, so the launch keypress could auto-submit the default track selection before the modal was visible to the user.
-
-Added renderer regression coverage in src/renderer/modals/youtube-track-picker.test.ts proving immediate Enter after open is ignored and a later Enter still submits normally.
-
-Implemented a 200ms open-key guard in src/renderer/modals/youtube-track-picker.ts for Enter-based submission only; Escape/click behavior unchanged.
-
-New follow-up regression report: YouTube subtitle picker can open before the mpv playback window is ready, leaving the picker behind the overlay after geometry snaps into place. Investigating picker-open gating and modal-targeting timing.
-
-Identified likely cause of picker-behind-overlay regression: YouTube picker open logic mixed overlay targets. First attempt preferred the visible main overlay, timeout retry switched to the dedicated modal window, allowing a late first open to cover the modal.
-
-Extracted picker-open policy into src/main/runtime/youtube-picker-open.ts and changed YouTube picker startup to always target the dedicated modal window, including retries. This keeps the picker on a single window path and lets overlay-runtime hide/click-through the main overlay while the modal is active.
-
-Added regression tests in src/main/runtime/youtube-picker-open.test.ts covering dedicated modal first-open, dedicated-modal retry, and failure when no modal target is available.
-
-User reports overlay flow still feels wrong: YouTube path appears to preload subtitles before mandatory selection and may open the picker before mpv window readiness. Re-evaluating flow design against regular video startup before further implementation.
-
-New follow-up regression report: duplicate overlay windows appear during YouTube playback and only one window shows subtitles. Investigating main-overlay versus dedicated modal-window handoff/cleanup.
-<!-- SECTION:NOTES:END -->
-
-## Final Summary
-
-<!-- SECTION:FINAL_SUMMARY:BEGIN -->
-Fixed the YouTube Anki initialization regression by making CLI commands that require overlay runtime bootstrap overlay startup prerequisites before command dispatch when not in texthooker-only mode. This ensures the YouTube playback flow has the mpv client, runtime options manager, and subtitle timing tracker ready before overlay/Anki initialization runs, restoring Mine actions and known-word-backed behavior.
-
-Added a regression test covering youtubePlay command dispatch in src/main/runtime/cli-command-runtime-handler.test.ts.
-
-Verification:
-- bun test src/main/runtime/cli-command-runtime-handler.test.ts src/main/runtime/cli-command-prechecks.test.ts src/main/runtime/cli-command-prechecks-main-deps.test.ts
-- bun test src/core/services/cli-command.test.ts src/main/runtime/cli-command-context-main-deps.test.ts src/main/runtime/cli-command-prechecks-main-deps.test.ts src/main/runtime/cli-command-runtime-handler.test.ts
-
-Updated the fix to avoid a YouTube-specific startup path: removed the dedicated ensureOverlayRuntimeReady helper from src/main.ts and relied on the shared CLI overlay prerequisite bootstrap instead.
-
-Additional verification: bun run typecheck
-
-Follow-up adjustment: initial overlay-runtime commands now bootstrap overlay prerequisites and initialize overlay during the shared initial-args startup path, rather than waiting for command dispatch. This keeps YouTube on the regular startup path while preserving earlier overlay availability.
-
-Additional verification: bun test src/main/runtime/initial-args-runtime-handler.test.ts src/main/runtime/initial-args-handler.test.ts src/main/runtime/initial-args-main-deps.test.ts; bun run test:runtime:compat
-
-Follow-up fix: the YouTube subtitle picker now ignores immediate Enter key bleed-through right after opening, preventing the startup keypress from auto-submitting the default track selection before the modal is visible.
-
-Added renderer regression coverage for immediate Enter suppression and verified with bun test src/renderer/modals/youtube-track-picker.test.ts plus the runtime-compat verification lane for the touched files.
-
-Follow-up fix: YouTube subtitle picker startup now uses a dedicated modal-window path consistently instead of mixing main-overlay first-open with modal-window retry. That prevents late overlay opens from covering the interactive picker while mpv/window tracking settles.
-
-Verified with bun test src/main/runtime/youtube-picker-open.test.ts, bun test src/renderer/modals/youtube-track-picker.test.ts, and the runtime-compat verification lane for src/main.ts plus the touched picker files.
-<!-- SECTION:FINAL_SUMMARY:END -->

backlog/tasks/task-224 - Auto-load-default-YouTube-subtitles-at-playback-start-and-make-picker-manual-only.md

@@ -1,61 +0,0 @@
----
-id: TASK-224
-title: >-
-  Auto-load default YouTube subtitles at playback start and make picker
-  manual-only
-status: Done
-assignee:
-  - Codex
-created_date: '2026-03-23 18:51'
-updated_date: '2026-03-24 06:41'
-labels:
-  - youtube
-  - mpv
-  - overlay
-  - keybindings
-dependencies: []
-references:
-  - /Users/sudacode/projects/japanese/SubMiner/src/main/runtime/youtube-flow.ts
-  - >-
-    /Users/sudacode/projects/japanese/SubMiner/src/renderer/modals/youtube-track-picker.ts
-  - /Users/sudacode/projects/japanese/SubMiner/src/config/definitions/shared.ts
-priority: high
-ordinal: 150500
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-Replace the mandatory YouTube subtitle picker startup flow with automatic default-track loading. On YouTube playback start, attempt to load the default primary subtitle and best-effort secondary subtitle without prompting. Gate playback only on primary subtitle load/tokenization readiness. If primary subtitle probing/download/loading fails, resume playback and report the failure through the configured notification/output path. Keep the YouTube subtitle picker as a regular overlay modal opened by a new default keybinding during active YouTube playback.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 Opening a YouTube URL auto-selects and attempts to load the default primary subtitle without opening the picker modal.
-- [x] #2 Opening a YouTube URL also attempts to load the default secondary subtitle when available, but playback never waits on secondary success.
-- [x] #3 Playback remains gated only until the primary subtitle is loaded and tokenization is ready; primary failure resumes playback immediately.
-- [x] #4 Primary auto-load failures report through the existing configured notification/output path and keep playback running.
-- [x] #5 The YouTube subtitle picker can be opened manually during active YouTube playback via a new default keybinding.
-- [x] #6 Regression tests cover startup auto-load success, primary failure fallback, and the manual picker keybinding flow.
-<!-- AC:END -->
-
-## Implementation Plan
-
-<!-- SECTION:PLAN:BEGIN -->
-1. Add failing tests for YouTube startup auto-load success, primary failure fallback, and manual picker keybinding flow.
-2. Refactor the YouTube runtime to auto-select default tracks on startup, gate playback only on primary subtitle/tokenization readiness, and route failures through the configured notification/output path.
-3. Add a new default keybinding and command path to open the YouTube picker manually during active YouTube playback.
-4. Run targeted tests, then SubMiner verification lanes for launcher/runtime changes; update docs/changelog if required by the final behavior change.
-<!-- SECTION:PLAN:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-Verification blocker outside this change: `bun run test:fast` still fails at `scripts/update-aur-package.test.ts` on macOS because `scripts/update-aur-package.sh` uses `mapfile`, which is unavailable in the system Bash 3.x environment used here.
-<!-- SECTION:NOTES:END -->
-
-## Final Summary
-
-<!-- SECTION:FINAL_SUMMARY:BEGIN -->
-Reworked app-owned YouTube playback to auto-load the default primary subtitle plus a best-effort secondary subtitle at startup instead of forcing the picker modal first. Playback now waits only on primary subtitle load/tokenization readiness, routes startup primary-failure messaging through the configured notification output path, and keeps the YouTube subtitle picker available on demand via a new default `Ctrl+Shift+J` keybinding during active YouTube playback. Updated the runtime/IPC/config plumbing, user-facing help/docs, and added regression coverage for startup auto-load, primary-failure fallback, and manual picker invocation.
-<!-- SECTION:FINAL_SUMMARY:END -->

backlog/tasks/task-225 - Fix-frozen-primary-YouTube-subtitle-display-after-auto-load-startup.md

@@ -1,41 +0,0 @@
----
-id: TASK-225
-title: Fix frozen primary YouTube subtitle display after auto-load startup
-status: Done
-assignee: []
-created_date: '2026-03-23 20:07'
-updated_date: '2026-03-24 06:41'
-labels:
-  - bug
-  - youtube
-  - subtitles
-dependencies:
-  - TASK-224
-priority: high
-ordinal: 149500
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-After the new YouTube auto-load startup flow, the primary subtitle overlay can stay stuck on an older line while the subtitle sidebar continues advancing. Investigate startup suppression / subtitle refresh timing and restore live primary overlay updates after auto-loaded subtitles are injected.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 When YouTube auto-load succeeds, the visible primary subtitle continues advancing after playback resumes.
-- [x] #2 Startup suppression does not leave the primary subtitle display stuck on a stale line.
-- [x] #3 A regression test covers the startup path that previously froze the visible primary subtitle while sidebar timing continued advancing.
-<!-- AC:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-Root cause: applyStartupState seeded youtubePlaybackFlowPending from initialArgs.youtubePlay, and runYoutubePlaybackFlowMain restored that preexisting true value after startup auto-load. Result: primary subtitle events stayed suppressed for startup-launched YouTube playback while sidebar timing still advanced.
-<!-- SECTION:NOTES:END -->
-
-## Final Summary
-
-<!-- SECTION:FINAL_SUMMARY:BEGIN -->
-Stopped pre-seeding youtubePlaybackFlowPending from startup CLI args so only the actual YouTube playback bootstrap window suppresses subtitle events. Added a regression test covering startup YouTube args and re-ran targeted YouTube/runtime subtitle tests plus typecheck.
-<!-- SECTION:FINAL_SUMMARY:END -->

backlog/tasks/task-226 - Restore-subtitle-sidebar-cues-for-auto-loaded-YouTube-subtitles.md

@@ -1,42 +0,0 @@
----
-id: TASK-226
-title: Restore subtitle sidebar cues for auto-loaded YouTube subtitles
-status: Done
-assignee: []
-created_date: '2026-03-23 20:21'
-updated_date: '2026-03-24 06:41'
-labels:
-  - bug
-  - youtube
-  - subtitle-sidebar
-dependencies:
-  - TASK-224
-  - TASK-225
-priority: high
-ordinal: 148500
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-After fixing startup subtitle event suppression, the primary subtitle overlay updates for auto-loaded YouTube playback but the subtitle sidebar reports no parsed subtitle cues available. Investigate parsed subtitle source registration / refresh for auto-loaded YouTube subtitle files and restore sidebar cue population.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 When YouTube auto-load succeeds, the subtitle sidebar receives parsed cues for the active primary subtitle source.
-- [x] #2 Auto-loaded YouTube subtitle source changes refresh the sidebar snapshot without requiring manual picker interaction.
-- [x] #3 A regression test covers the startup auto-load path where live primary subtitles render but sidebar cues remain empty.
-<!-- AC:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-Root cause: successful YouTube auto-load refreshed visible primary subtitle state, but did not explicitly initialize parsed subtitle cues from the resolved downloaded primary subtitle file. Sidebar cue population depended on later mpv source rediscovery, which could leave snapshots empty.
-<!-- SECTION:NOTES:END -->
-
-## Final Summary
-
-<!-- SECTION:FINAL_SUMMARY:BEGIN -->
-Added a regression test for YouTube auto-load sidebar cue refresh and wired the YouTube subtitle flow to explicitly refresh parsed subtitle cues from the resolved primary subtitle path after a successful load. Verified with targeted YouTube/sidebar/runtime tests plus typecheck.
-<!-- SECTION:FINAL_SUMMARY:END -->

backlog/tasks/task-227 - Assess-and-address-PR-31-latest-CodeRabbit-review-round.md

@@ -1,65 +0,0 @@
----
-id: TASK-227
-title: 'Assess and address PR #31 latest CodeRabbit review round'
-status: Done
-assignee:
-  - codex
-created_date: '2026-03-24 03:53'
-updated_date: '2026-03-24 06:41'
-labels:
-  - pr-review
-  - coderabbit
-dependencies: []
-references:
-  - >-
-    PR #31 feat: add app-owned YouTube subtitle flow with absPlayer-style
-    parsing
-priority: medium
-ordinal: 147500
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-Inspect the latest CodeRabbit review round on PR #31, verify each actionable comment against the current branch, implement only the valid fixes, add regression coverage where appropriate, and prepare thread replies for resolved or declined items.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 Latest CodeRabbit comments on PR #31 are triaged into valid fixes vs non-actioned suggestions with rationale.
-- [x] #2 Confirmed issues are fixed with regression coverage where appropriate.
-- [x] #3 Relevant verification passes for the touched areas.
-- [x] #4 PR reply notes are ready for each addressed or declined latest-review comment.
-<!-- AC:END -->
-
-## Implementation Plan
-
-<!-- SECTION:PLAN:BEGIN -->
-1. Verify the five latest CodeRabbit inline comments against the current branch and separate valid bugs from non-actioned suggestions.
-2. Add failing regression coverage for confirmed issues in launcher playback tests, CLI YouTube flow error handling, and renderer YouTube picker disabled-state behavior.
-3. Implement the minimal production fixes for the confirmed issues, plus remove the duplicate overlay Anki initialization if still redundant.
-4. Inspect the YouTube primary-subtitle failure timer wiring to decide whether a code change is warranted in this round or whether a technical reply declining the comment is more correct.
-5. Run targeted Bun tests for the touched files and prepare concise PR thread replies for each latest-review comment.
-<!-- SECTION:PLAN:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-Triaged the latest PR #31 CodeRabbit round: five inline comments were current action items; implemented all five. Strengthened the launcher playback test fixture so YouTube pause coverage no longer piggybacks on generic overlay auto-pause settings.
-
-Added regression tests for CLI YouTube flow rejection handling, no-track picker disabled-state restoration, and app-owned YouTube notification suppression while subtitle acquisition is still in flight.
-
-Implemented `runAsyncWithOsd(...)` handling for `args.youtubePlay`, kept no-track picker controls disabled after failed continue attempts, added `setAppOwnedFlowInFlight(...)` to the YouTube primary-subtitle notification runtime with main-process wiring around `runYoutubePlaybackFlowMain(...)`, and removed the duplicate `initializeOverlayAnkiIntegrationCore(...)` call from `initializeOverlayRuntime()`.
-
-Verification passed: `bun test launcher/commands/playback-command.test.ts src/core/services/cli-command.test.ts src/renderer/modals/youtube-track-picker.test.ts src/main/runtime/youtube-primary-subtitle-notification.test.ts` and `bun run typecheck`.
-<!-- SECTION:NOTES:END -->
-
-## Final Summary
-
-<!-- SECTION:FINAL_SUMMARY:BEGIN -->
-Assessed the latest CodeRabbit review round on PR #31 and implemented all five current inline action items. Strengthened the launcher playback regression test so app-owned YouTube pause behavior is asserted independently from generic overlay auto-pause settings, wrapped the CLI `youtubePlay` branch in the existing `runAsyncWithOsd(...)` path so probe/download/startup failures surface in logs and OSD, kept the no-track YouTube picker controls disabled after rejected continue attempts, suppressed the generic primary-subtitle failure timer while the app-owned YouTube flow is still probing/downloading and restarted it only after the flow settles, and removed the duplicate overlay Anki initialization from `initializeOverlayRuntime()`.
-
-Verification passed with `bun test launcher/commands/playback-command.test.ts src/core/services/cli-command.test.ts src/renderer/modals/youtube-track-picker.test.ts src/main/runtime/youtube-primary-subtitle-notification.test.ts` and `bun run typecheck`.
-
-Prepared thread-reply notes for the five latest inline comments; did not post them because GitHub replies are an external side effect.
-<!-- SECTION:FINAL_SUMMARY:END -->

backlog/tasks/task-228 - Assess-and-address-PR-31-subsequent-CodeRabbit-review-round.md

@@ -1,64 +0,0 @@
----
-id: TASK-228
-title: 'Assess and address PR #31 subsequent CodeRabbit review round'
-status: Done
-assignee:
-  - codex
-created_date: '2026-03-24 04:10'
-updated_date: '2026-03-24 06:41'
-labels:
-  - pr-review
-  - coderabbit
-dependencies: []
-references:
-  - >-
-    PR #31 feat: add app-owned YouTube subtitle flow with absPlayer-style
-    parsing
-  - 'commit cdb12827 fix: address PR #31 latest review follow-ups'
-priority: medium
-ordinal: 146500
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-Inspect the subsequent CodeRabbit review round on PR #31 after commit cdb12827, verify each newly reported issue against the current branch, implement the valid fixes with regression coverage where appropriate, and prepare/update PR thread replies.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 New CodeRabbit comments after cdb12827 are triaged into valid fixes vs declined suggestions with rationale.
-- [x] #2 Confirmed issues are fixed with regression coverage where appropriate.
-- [x] #3 Relevant verification passes for the touched areas.
-- [x] #4 PR threads are updated for the addressed comments.
-<!-- AC:END -->
-
-## Implementation Plan
-
-<!-- SECTION:PLAN:BEGIN -->
-1. Verify the new CodeRabbit comments after cdb12827 and separate valid bugs from refactor-only suggestions.
-2. Add failing regression coverage for the valid runtime issues: `track.selected` fallback in the YouTube primary-subtitle notifier and consistent no-track handling in the picker.
-3. Inspect existing test seams for the `main.ts` flow-entry guards; if lightweight coverage exists, add it before patching. Otherwise apply the minimal `main.ts` fixes and rely on typecheck plus targeted regression tests around the affected runtime helpers.
-4. Implement the confirmed fixes: picker re-entry guard, broader `inFlight` cleanup, `track.selected` fallback, and a single canonical `hasTracks` check.
-5. Run targeted tests/typecheck and update the new PR threads with landed fix refs.
-<!-- SECTION:PLAN:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-Triaged the post-cdb12827 CodeRabbit round. Implemented the 4 concrete follow-ups: manual picker re-entry guard, broader `setAppOwnedFlowInFlight(...)` cleanup, `track.selected` fallback in the YouTube primary-subtitle notifier, and a single canonical `payloadHasTracks(...)` helper in the picker. Also took the adjacent `replaceChildren()` cleanup while touching the same picker paths.
-
-Verification passed: `bun test src/main/runtime/youtube-primary-subtitle-notification.test.ts src/renderer/modals/youtube-track-picker.test.ts launcher/commands/playback-command.test.ts src/core/services/cli-command.test.ts` and `bun run typecheck`.
-
-Updated the new CodeRabbit inline threads with landed fix refs and left a top-level PR comment noting the large refactor suggestions are intentionally out of scope for this bugfix round.
-<!-- SECTION:NOTES:END -->
-
-## Final Summary
-
-<!-- SECTION:FINAL_SUMMARY:BEGIN -->
-Assessed the subsequent CodeRabbit review round on PR #31 after cdb12827 and applied the valid follow-ups in commit 5f6f93cd. Added a guard in `openYoutubeTrackPickerFromPlayback()` so the manual picker cannot re-enter while another YouTube flow session is active, widened the app-owned in-flight suppression to cover synchronous Windows mpv bootstrap and connect failures, taught the primary-subtitle notifier to honor `track.selected` before `sid` arrives, and unified the picker’s subtitle-availability logic behind `payloadHasTracks(...)` while swapping node clearing to `replaceChildren()`.
-
-Verification passed with `bun test src/main/runtime/youtube-primary-subtitle-notification.test.ts src/renderer/modals/youtube-track-picker.test.ts launcher/commands/playback-command.test.ts src/core/services/cli-command.test.ts` and `bun run typecheck`.
-
-Updated the latest inline CodeRabbit threads plus a top-level PR comment summarizing the round and explicitly deferred the large refactor suggestions as non-blocking maintainability nits.
-<!-- SECTION:FINAL_SUMMARY:END -->

backlog/tasks/task-229 - Address-PR-31-final-CodeRabbit-picker-test-follow-up.md

@@ -1,55 +0,0 @@
----
-id: TASK-229
-title: 'Address PR #31 final CodeRabbit picker test follow-up'
-status: Done
-assignee:
-  - codex
-created_date: '2026-03-24 04:27'
-updated_date: '2026-03-24 06:41'
-labels:
-  - pr-review
-  - coderabbit
-dependencies: []
-references:
-  - >-
-    PR #31 feat: add app-owned YouTube subtitle flow with absPlayer-style
-    parsing
-  - >-
-    CodeRabbit comment on src/renderer/modals/youtube-track-picker.test.ts
-    global restoration / harness duplication
-priority: medium
-ordinal: 145500
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-Fix the remaining CodeRabbit comment on the YouTube picker test file by restoring absent globals correctly and reducing repeated test harness setup so global stubbing is consistent and isolated.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 Picker tests restore `window`, `document`, and `CustomEvent` without leaving undefined-valued globals behind.
-- [x] #2 Repeated picker test setup is consolidated enough to remove the current review complaint.
-- [x] #3 Relevant picker tests pass and PR thread is updated.
-<!-- AC:END -->
-
-## Implementation Plan
-
-<!-- SECTION:PLAN:BEGIN -->
-1. Add a failing regression around global restoration semantics in the YouTube picker test harness.
-2. Extract shared DOM/environment helpers and restore logic using delete when globals were originally absent.
-3. Re-run focused tests and typecheck, then commit/push and reply on the PR thread.
-<!-- SECTION:PLAN:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-Latest CodeRabbit comment targets youtube-track-picker.test.ts harness cleanup and correct restoration of global properties.
-<!-- SECTION:NOTES:END -->
-
-## Final Summary
-
-<!-- SECTION:FINAL_SUMMARY:BEGIN -->
-Addressed the last PR #31 CodeRabbit comment by refactoring the YouTube picker test harness to use shared DOM/env helpers, restoring absent globals via delete semantics, adding a regression for cleanup behavior, and pushing commit 039e2f56 with focused picker tests plus typecheck passing.
-<!-- SECTION:FINAL_SUMMARY:END -->

backlog/tasks/task-231 - Restore-controller-input-while-subtitle-sidebar-is-open.md

@@ -1,57 +0,0 @@
----
-id: TASK-231
-title: Restore controller input while subtitle sidebar is open
-status: Done
-assignee:
-  - '@codex'
-created_date: '2026-03-24 00:15'
-updated_date: '2026-03-24 00:15'
-labels:
-  - bug
-  - controller
-  - subtitle-sidebar
-  - overlay
-dependencies: []
-references:
-  - /home/sudacode/projects/japanese/SubMiner/src/renderer/renderer.ts
-  - /home/sudacode/projects/japanese/SubMiner/src/renderer/controller-interaction-blocking.ts
-  - /home/sudacode/projects/japanese/SubMiner/src/renderer/controller-interaction-blocking.test.ts
-priority: high
-ordinal: 54900
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-
-When keyboard-only mode is active, opening the subtitle sidebar should not disable controller navigation and lookup/mining controls. Restore controller input while the sidebar is open, while keeping true modal dialogs blocking controller actions.
-
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-
-<!-- AC:BEGIN -->
-
-- [x] #1 Opening the subtitle sidebar does not block controller input for keyboard-only mode actions.
-- [x] #2 Controller-select/debug and other true modal dialogs still block controller actions while open.
-- [x] #3 Focused regression coverage exists for the sidebar-open controller gating rule.
-
-<!-- AC:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-
-Root cause: renderer gamepad polling used the broad `isAnyModalOpen()` check as its interaction gate, and that list includes `subtitleSidebarModalOpen`. The subtitle sidebar is non-modal for controller usage, so gamepad input was being suppressed whenever the sidebar was visible.
-
-Fixed by extracting a dedicated controller-interaction blocking helper that excludes the subtitle sidebar but keeps the existing blocking behavior for true modal dialogs.
-
-<!-- SECTION:NOTES:END -->
-
-## Final Summary
-
-<!-- SECTION:FINAL_SUMMARY:BEGIN -->
-
-Restored controller input while the subtitle sidebar is open by switching gamepad polling to a dedicated modal-blocking rule that leaves the sidebar controller-passive. Added a regression test covering the sidebar-open exception and preserving hard blocks for actual modal dialogs.
-
-<!-- SECTION:FINAL_SUMMARY:END -->

backlog/tasks/task-232 - Trim-release-package-size-by-pruning-duplicate-and-source-only-assets.md

@@ -1,66 +0,0 @@
----
-id: TASK-232
-title: Trim release package size by pruning duplicate and source-only assets
-status: Done
-assignee:
-  - '@codex'
-created_date: '2026-03-24 12:05'
-updated_date: '2026-03-24 12:30'
-labels:
-  - release
-  - packaging
-priority: medium
-ordinal: 54700
-dependencies: []
-references:
-  - /home/sudacode/projects/japanese/SubMiner/package.json
-  - /home/sudacode/projects/japanese/SubMiner/src/release-workflow.test.ts
-  - /home/sudacode/projects/japanese/SubMiner/src/core/services/texthooker.ts
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-
-Reduce packaged release artifact size without changing user-visible functionality by pruning files that are duplicated between `app.asar` and `extraResources`, excluding source/test/doc-only trees from Electron packaging, and trimming obviously non-runtime vendored payload.
-
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-
-<!-- AC:BEGIN -->
-
-- [x] #1 Electron packaging excludes repo content that is source-only, test-only, docs-only, or duplicated by `extraResources`.
-- [x] #2 Release packaging tests cover the new exclusion rules.
-- [x] #3 Verification includes at least targeted release-packaging tests and one packaging-oriented validation step.
-<!-- AC:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-
-Completed scope:
-- Exclude `assets`, `plugin`, and `vendor/yomitan-jlpt-vocab` from `files` because they are already staged via `extraResources`.
-- Exclude `dist` sourcemaps/tests, repo docs/tests/packaging metadata, and stats source leftovers from `files`.
-- Exclude non-runtime `vendor/texthooker-ui` payload such as `public/`, `.vscode/`, and package metadata.
-- Exclude Linux musl libsql binary from packaged app payload for AppImage-focused savings.
-
-Verification:
-- `bun test src/release-workflow.test.ts`
-- `bun run build`
-- `node_modules/.bin/electron-builder --linux dir --publish never`
-- `node_modules/.bin/electron-builder --linux AppImage --publish never`
-
-Observed result:
-- `release/linux-unpacked/resources/app.asar` dropped from about `100 MB` to `29 MB`.
-- `release/SubMiner-0.9.0.AppImage` dropped from about `256 MB` to `194 MB`.
-
-<!-- SECTION:NOTES:END -->
-
-## Final Summary
-
-<!-- SECTION:FINAL_SUMMARY:BEGIN -->
-
-Trimmed Electron packaging so release artifacts no longer bundle duplicated `extraResources`, source/test/doc-only repo content, non-runtime `texthooker-ui` files, or the Linux musl libsql binary. Added release-packaging regression coverage and verified the Linux package shrink with fresh local builds.
-
-<!-- SECTION:FINAL_SUMMARY:END -->

backlog/tasks/task-233 - Cut-patch-release-v0.9.1-for-package-size-pruning.md

@@ -1,69 +0,0 @@
----
-id: TASK-233
-title: Cut patch release v0.9.1 for package size pruning
-status: Done
-assignee:
-  - '@codex'
-created_date: '2026-03-24 12:40'
-updated_date: '2026-03-24 12:55'
-labels:
-  - release
-  - patch
-dependencies:
-  - TASK-232
-references:
-  - /home/sudacode/projects/japanese/SubMiner/package.json
-  - /home/sudacode/projects/japanese/SubMiner/CHANGELOG.md
-  - /home/sudacode/projects/japanese/SubMiner/release/release-notes.md
-  - /home/sudacode/projects/japanese/SubMiner/docs-site/changelog.md
-priority: high
-ordinal: 54800
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-
-Publish a patch release for the packaging-size cleanup by bumping the app version to `0.9.1`, generating committed release metadata, and keeping release-facing docs/changelog surfaces aligned.
-
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-
-<!-- AC:BEGIN -->
-
-- [x] #1 Repository version metadata is updated to `0.9.1`.
-- [x] #2 `CHANGELOG.md`, `release/release-notes.md`, and `docs-site/changelog.md` contain the committed `v0.9.1` release line and the consumed fragment is removed.
-- [x] #3 Release-readiness verification passes for changelog, docs, tests, and build lanes.
-<!-- AC:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-
-Completed:
-- Bumped `package.json` to `0.9.1`.
-- Ran `bun run changelog:build --version 0.9.1 --date 2026-03-24`, which generated `CHANGELOG.md` + `release/release-notes.md` and consumed both pending release fragments.
-- Synced `docs-site/changelog.md` with the generated `v0.9.1` release line.
-- Confirmed no additional README/docs wording changes were needed beyond changelog surfaces.
-
-Verification:
-- `bun run changelog:lint`
-- `bun run changelog:check --version 0.9.1`
-- `bun run verify:config-example`
-- `bun run typecheck`
-- `bun run test:fast`
-- `bun run test:env`
-- `bun run build`
-- `bun run docs:test`
-- `bun run docs:build`
-
-<!-- SECTION:NOTES:END -->
-
-## Final Summary
-
-<!-- SECTION:FINAL_SUMMARY:BEGIN -->
-
-Prepared patch release `v0.9.1` locally. Version metadata, committed changelog artifacts, release notes, and docs-site changelog are aligned, and the release gate is green. Pending manual release actions are the release-prep commit, `git tag v0.9.1`, and push/tag publication.
-
-<!-- SECTION:FINAL_SUMMARY:END -->

backlog/tasks/task-234 - Address-PR-35-latest-CodeRabbit-review-round.md

@@ -1,68 +0,0 @@
----
-id: TASK-234
-title: 'Address PR #35 latest CodeRabbit review round'
-status: Done
-assignee:
-  - codex
-created_date: '2026-03-26 03:59'
-updated_date: '2026-03-26 04:01'
-labels:
-  - review-comments
-  - coderabbit
-dependencies: []
-references:
-  - /Users/sudacode/projects/japanese/SubMiner/src/main.ts
-  - /Users/sudacode/projects/japanese/SubMiner/src/cli/args.test.ts
-  - >-
-    /Users/sudacode/projects/japanese/SubMiner/src/main/runtime/cli-command-prechecks.test.ts
-  - >-
-    /Users/sudacode/projects/japanese/SubMiner/src/main/runtime/youtube-playback-launch.ts
-priority: medium
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-Assess and implement the latest actionable CodeRabbit feedback on PR #35 for the Windows YouTube playback flow. Scope includes fixing the overlapping youtubePlay cleanup race in main runtime state and any low-risk follow-up test/clarity comments from the same review round.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 Overlapping youtubePlay requests no longer let an older flow clear active quit-on-disconnect/app-owned-flow state for a newer flow.
-- [x] #2 Latest low-risk CodeRabbit test and clarity follow-ups for this PR round are addressed or intentionally rejected based on code verification.
-- [x] #3 Relevant tests covering the touched areas pass locally.
-<!-- AC:END -->
-
-## Implementation Plan
-
-<!-- SECTION:PLAN:BEGIN -->
-1. Update runYoutubePlaybackFlowMain in src/main.ts to use a per-request generation guard around shared YouTube flow state so overlapping requests cannot clear the active timer, armed flag, or app-owned-flow marker for a newer request.
-2. Address verified low-risk latest-round follow-ups: add direct startup-prereq assertions in src/cli/args.test.ts, extend side-effect assertions in src/main/runtime/cli-command-prechecks.test.ts, and rename the youtube-playback-launch polling variable for clarity.
-3. Run targeted Bun tests for the touched areas and record results in the task notes/final summary.
-<!-- SECTION:PLAN:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-Implemented per-request youtubePlaybackFlowGeneration guard in src/main.ts so superseded youtubePlay flows cannot clear the active arm timer, armed flag, or app-owned-flow state for a newer request.
-
-Added explicit startup-prereq assertions in src/cli/args.test.ts and stronger warmup/log side-effect assertions in src/main/runtime/cli-command-prechecks.test.ts for the latest CodeRabbit follow-ups.
-
-Renamed youtube-playback-launch polling variable from pathChanged to pathDiffersFromInitial for accuracy without behavior change.
-
-Verification: bun test src/cli/args.test.ts; bun test src/main/runtime/cli-command-prechecks.test.ts; bun test src/main/runtime/youtube-playback-launch.test.ts; bun run typecheck.
-<!-- SECTION:NOTES:END -->
-
-## Final Summary
-
-<!-- SECTION:FINAL_SUMMARY:BEGIN -->
-Addressed the latest PR #35 CodeRabbit round by making YouTube playback flow cleanup generation-safe in src/main.ts. Overlapping youtubePlay requests now isolate timer/armed/app-owned-flow cleanup to the currently active request so an older flow cannot clear state for its replacement.
-
-Also folded in the latest low-risk follow-ups: args tests now assert that youtube playback requires overlay startup prerequisites, cli-command precheck tests now assert warmup/log side effects for the youtube transition, and youtube-playback-launch.ts uses a clearer variable name for the initial-path comparison.
-
-Verification:
-- bun test src/cli/args.test.ts
-- bun test src/main/runtime/cli-command-prechecks.test.ts
-- bun test src/main/runtime/youtube-playback-launch.test.ts
-- bun run typecheck
-<!-- SECTION:FINAL_SUMMARY:END -->

backlog/tasks/task-235 - Address-PR-35-autoplay-retry-CodeRabbit-follow-up.md

@@ -1,56 +0,0 @@
----
-id: TASK-235
-title: 'Address PR #35 autoplay retry CodeRabbit follow-up'
-status: Done
-assignee:
-  - codex
-created_date: '2026-03-26 04:30'
-updated_date: '2026-03-26 04:31'
-labels:
-  - review-comments
-  - coderabbit
-dependencies: []
-references:
-  - /Users/sudacode/projects/japanese/SubMiner/src/main.ts
-priority: medium
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-Assess and implement the latest CodeRabbit follow-up on PR #35 concerning stale autoplay-ready fallback retries interfering with a new app-owned YouTube playback flow in main.ts.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 Starting a new app-owned YouTube playback flow invalidates any pending autoplay-ready fallback retries from older playback state before mpv prep begins.
-- [x] #2 Relevant verification for the touched main.ts autoplay retry logic passes locally.
-- [x] #3 Task notes/final summary capture the fix and verification.
-<!-- AC:END -->
-
-## Implementation Plan
-
-<!-- SECTION:PLAN:BEGIN -->
-1. Add a helper in src/main.ts that invalidates pending autoplay-ready fallback retry state by clearing the tracked media path and advancing the autoplay generation counter.
-2. Invoke that helper at the start of runYoutubePlaybackFlowMain before app-owned YouTube playback takes over so stale retries cannot unpause reused playback.
-3. Run relevant verification for the touched main.ts path and record results in the task notes/final summary.
-<!-- SECTION:PLAN:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-Added invalidatePendingAutoplayReadyFallbacks() in src/main.ts to clear the tracked autoplay-ready media path and advance the autoplay generation before a new app-owned YouTube flow claims playback. This invalidates stale fallback retry closures even when the reused playback path is the same.
-
-Verification: bun test src/main/runtime/mpv-main-event-actions.test.ts; bun test src/main/runtime/startup-autoplay-release-policy.test.ts; bun run typecheck.
-<!-- SECTION:NOTES:END -->
-
-## Final Summary
-
-<!-- SECTION:FINAL_SUMMARY:BEGIN -->
-Addressed the latest PR #35 CodeRabbit follow-up by invalidating pending autoplay-ready fallback retries before a new app-owned YouTube playback flow takes over in src/main.ts. The new helper clears the tracked autoplay media path and advances the autoplay generation counter, so retry closures from older playback state cannot later unpause the newly prepared flow when reusing the same media path.
-
-Verification:
-- bun test src/main/runtime/mpv-main-event-actions.test.ts
-- bun test src/main/runtime/startup-autoplay-release-policy.test.ts
-- bun run typecheck
-<!-- SECTION:FINAL_SUMMARY:END -->

changes/2026-03-19-incremental-known-word-cache.md

@@ -0,0 +1,4 @@
+type: fixed
+area: anki
+
+- Known-word cache refreshes now reconcile Anki changes incrementally instead of wiping and rebuilding on startup, mined cards can append their word into the cache immediately through a new default-enabled config flag, and explicit refreshes now run through `subminer doctor --refresh-known-words`.

changes/2026-03-19-known-jlpt-reading-fallback.md

@@ -0,0 +1,4 @@
+type: fixed
+area: subtitle
+
+- Restored known-word coloring and JLPT underlines for subtitle tokens like `大体` when the subtitle token is kanji but the known-word cache only matches the kana reading.

changes/2026-03-19-stats-ended-media-progress.md

@@ -0,0 +1,4 @@
+type: fixed
+area: stats
+
+- Episode progress in the anime page now uses the last ended playback position instead of cumulative active watch time, avoiding distorted percentages after rewatches or repeated sessions.

changes/2026-03-19-stats-session-progress-checkpoint.md

@@ -0,0 +1,4 @@
+type: fixed
+area: stats
+
+- Anime episode progress now keeps the latest known playback position through active-session checkpoints and stale-session recovery, so recently watched episodes no longer lose their progress percentage.

config.example.jsonc

@@ -187,7 +187,7 @@
   // ==========================================
   // Secondary Subtitles
   // Dual subtitle track options.
-  // Used by the YouTube subtitle loading flow as secondary language preferences.
+  // Used by subminer YouTube subtitle generation as secondary language preferences.
   // Hot-reload: defaultMode updates live while SubMiner is running.
   // ==========================================
   "secondarySub": {
@@ -284,30 +284,6 @@
     } // Secondary setting.
   }, // Primary and secondary subtitle styling.
 
-  // ==========================================
-  // Subtitle Sidebar
-  // Parsed-subtitle sidebar cue list styling, behavior, and toggle key.
-  // Hot-reload: subtitle sidebar changes apply live without restarting SubMiner.
-  // ==========================================
-  "subtitleSidebar": {
-    "enabled": false, // Enable the subtitle sidebar feature for parsed subtitle sources. Values: true | false
-    "autoOpen": false, // Automatically open the subtitle sidebar once during overlay startup. Values: true | false
-    "layout": "overlay", // Render the subtitle sidebar as a floating overlay or reserve space inside mpv. Values: overlay | embedded
-    "toggleKey": "Backslash", // KeyboardEvent.code used to toggle the subtitle sidebar open and closed.
-    "pauseVideoOnHover": false, // Pause mpv while hovering the subtitle sidebar, then resume on leave. Values: true | false
-    "autoScroll": true, // Auto-scroll the active subtitle cue into view while playback advances. Values: true | false
-    "maxWidth": 420, // Maximum sidebar width in CSS pixels.
-    "opacity": 0.95, // Base opacity applied to the sidebar shell.
-    "backgroundColor": "rgba(73, 77, 100, 0.9)", // Background color for the subtitle sidebar shell.
-    "textColor": "#cad3f5", // Default cue text color in the subtitle sidebar.
-    "fontFamily": "\"M PLUS 1\", \"Noto Sans CJK JP\", sans-serif", // Font family used for subtitle sidebar cue text.
-    "fontSize": 16, // Base font size for subtitle sidebar cue text in CSS pixels.
-    "timestampColor": "#a5adcb", // Timestamp color in the subtitle sidebar.
-    "activeLineColor": "#f5bde6", // Text color for the active subtitle cue.
-    "activeLineBackgroundColor": "rgba(138, 173, 244, 0.22)", // Background color for the active subtitle cue.
-    "hoverLineBackgroundColor": "rgba(54, 58, 79, 0.84)" // Background color for hovered subtitle cues.
-  }, // Parsed-subtitle sidebar cue list styling, behavior, and toggle key.
-
   // ==========================================
   // Shared AI Provider
   // Canonical OpenAI-compatible provider transport settings shared by Anki and YouTube subtitle fixing.
@@ -414,24 +390,24 @@
   }, // Jimaku API configuration and defaults.
 
   // ==========================================
-  // YouTube Playback Settings
-  // Defaults for SubMiner YouTube subtitle loading and languages.
+  // YouTube Subtitle Generation
+  // Defaults for SubMiner YouTube subtitle generation.
   // ==========================================
   "youtubeSubgen": {
-    "whisperBin": "", // Legacy compatibility path kept for external subtitle fallback tools; not used by default.
-    "whisperModel": "", // Legacy compatibility model path kept for external subtitle fallback tooling; not used by default.
-    "whisperVadModel": "", // Legacy compatibility VAD path kept for external subtitle fallback tooling; not used by default.
-    "whisperThreads": 4, // Legacy thread tuning for subtitle fallback tooling; not used by default.
-    "fixWithAi": false, // Legacy subtitle fallback post-processing switch kept for compatibility; use is currently disabled by default. Values: true | false
+    "whisperBin": "", // Path to whisper.cpp CLI used as fallback transcription engine.
+    "whisperModel": "", // Path to whisper model used for fallback transcription.
+    "whisperVadModel": "", // Path to optional whisper VAD model used for subtitle generation.
+    "whisperThreads": 4, // Thread count passed to whisper.cpp subtitle generation runs.
+    "fixWithAi": false, // Use shared AI provider to post-process whisper-generated YouTube subtitles. Values: true | false
     "ai": {
-      "model": "", // Optional model override for legacy subtitle fallback post-processing; not used by default.
-      "systemPrompt": "" // Optional system prompt override for legacy subtitle fallback post-processing; not used by default.
+      "model": "", // Optional model override for YouTube subtitle AI post-processing.
+      "systemPrompt": "" // Optional system prompt override for YouTube subtitle AI post-processing.
     }, // Ai setting.
     "primarySubLanguages": [
       "ja",
       "jpn"
     ] // Comma-separated primary subtitle language priority used by the launcher.
-  }, // Defaults for SubMiner YouTube subtitle loading and languages.
+  }, // Defaults for SubMiner YouTube subtitle generation.
 
   // ==========================================
   // Anilist

docs-site/.vitepress/config.ts

@@ -95,7 +95,6 @@ export default {
           { text: 'Building & Testing', link: '/development' },
           { text: 'Architecture', link: '/architecture' },
           { text: 'IPC + Runtime Contracts', link: '/ipc-contracts' },
-          { text: 'WebSocket + Texthooker API', link: '/websocket-texthooker-api' },
           { text: 'Changelog', link: '/changelog' },
         ],
       },

docs-site/anki-integration.md

@@ -29,8 +29,7 @@ In both modes, the enrichment workflow is the same:
 4. Fills the translation field from the secondary subtitle or AI.
 5. Writes metadata to the miscInfo field.
 
-Polling mode uses the query `"deck:<ankiConnect.deck>" added:1` to find recently added cards. If no deck is configured, it searches all decks.
-Known-word sync scope is controlled by `ankiConnect.knownWords.decks` (object map), with `ankiConnect.deck` used as legacy fallback.
+Polling mode uses the query `"deck:<your-deck>" added:1` to find recently added cards. If no deck is configured, it searches all decks.
 
 ### Proxy Mode Setup (Yomitan / Texthooker)
 

docs-site/changelog.md

@@ -1,35 +1,5 @@
 # Changelog
 
-## v0.9.1 (2026-03-24)
-- Reduced packaged release size by excluding duplicate `extraResources` payload and pruning docs, tests, sourcemaps, and other source-only files from Electron bundles.
-- Restored controller navigation and lookup/mining controls while the subtitle sidebar is open, while keeping true modal dialogs blocking controller actions.
-- Fixed subtitle annotation clearing so explanatory contrast endings like `んですけど` are excluded consistently across the shared tokenizer filter and annotation stage.
-
-## v0.9.0 (2026-03-23)
-- Added an app-owned YouTube subtitle flow with absPlayer-style timedtext parsing that auto-loads the default primary subtitle plus a best-effort secondary at startup and resumes once the primary is ready.
-- Added a manual YouTube subtitle picker on `Ctrl+Alt+C` so subtitle selection can be retried on demand during active YouTube playback.
-- Added yt-dlp metadata probing so YouTube playback and immersion tracking record canonical video title and channel metadata.
-- Disabled conflicting mpv native subtitle auto-selection for the app-owned flow so injected explicit tracks stay authoritative.
-- Added OSD status updates covering YouTube playback startup, subtitle acquisition, and subtitle loading.
-- Stopped forcing `--ytdl-raw-options=` before user-provided mpv options so existing YouTube cookie integrations are preserved.
-- Improved sidebar startup/resume behavior, scroll handling, and overlay/sidebar subtitle synchronization.
-- Stats Library tab now shows YouTube video title, channel name, and thumbnail for YouTube media entries.
-- Added a new WebSocket / Texthooker API integration guide covering payload formats, custom client patterns, and mpv plugin automation.
-- Fixed Anki media mining for mpv YouTube streams so audio and screenshot capture work correctly during YouTube playback sessions.
-- Fixed YouTube media path handling in immersion tracking so YouTube sessions record correct media references and AniList state transitions do not fire for YouTube media.
-- Reused existing authoritative YouTube subtitle tracks when present, fell back only for missing sides, and kept native mpv secondary subtitle rendering hidden so the overlay remains the visible secondary subtitle surface.
-
-## v0.8.0 (2026-03-22)
-- Added a configurable subtitle sidebar feature (`subtitleSidebar`) with overlay/embedded rendering, click-to-seek cue list, and hot-reloadable visibility and behavior controls.
-- Added a rendered sidebar modal with cue list display, click-to-seek, active-cue highlighting, and embedded layout support.
-- Added sidebar snapshot plumbing between main and renderer for overlay/sidebar synchronization.
-- Added sidebar configuration options for visibility and behavior (enabled, layout, toggle key, autoOpen, pauseOnHover, autoScroll) plus typography and sizing controls.
-- Documented `subtitleSidebar` configuration and behavior in user-facing docs (configuration.md, shortcuts.md, config.example.jsonc).
-- Updated subtitle prefetch/rendering flow to keep overlay and sidebar state in sync through media transitions.
-- Kept sidebar cue tracking stable across playback transitions and timing edge cases.
-- Fixed sidebar startup/resume positioning to jump directly to the first resolved active cue.
-- Prevented stale subtitle refreshes from regressing active-cue state.
-
 ## v0.7.0 (2026-03-19)
 - Added a full local immersion dashboard release line with Overview, Library, Trends, Vocabulary, and Sessions drill-down views backed by SQLite tracking data.
 - Added browser-first stats workflows: `subminer stats`, background stats daemon controls (`-b` / `-s`), stats cleanup, and dashboard-side mining actions with media enrichment.

docs-site/configuration.md

@@ -17,11 +17,6 @@ For most users, start with this minimal configuration:
   "ankiConnect": {
     "enabled": true,
     "deck": "YourDeckName",
-    "knownWords": {
-      "decks": {
-        "YourDeckName": ["Word", "Word Reading", "Expression"]
-      }
-    },
     "fields": {
       "sentence": "Sentence",
       "audio": "Audio",
@@ -31,8 +26,6 @@ For most users, start with this minimal configuration:
 }
 ```
 
-`ankiConnect.deck` is still accepted for backward-compatible polling scope and legacy known-word fallback behavior. For known-word cache scope, prefer `ankiConnect.knownWords.decks` with deck-to-fields mapping.
-
 Then customize as needed using the sections below.
 
 ## Configuration File
@@ -66,7 +59,6 @@ SubMiner watches the active config file (`config.jsonc` or `config.json`) while
 Hot-reloadable fields:
 
 - `subtitleStyle`
-- `subtitleSidebar`
 - `keybindings`
 - `shortcuts`
 - `secondarySub.defaultMode`
@@ -96,7 +88,6 @@ The configuration file includes several main sections:
 **Subtitle Display**
 
 - [**Subtitle Style**](#subtitle-style) - Appearance customization
-- [**Subtitle Sidebar**](#subtitle-sidebar) - Parsed cue list sidebar modal
 - [**Subtitle Position**](#subtitle-position) - Overlay vertical positioning
 - [**Secondary Subtitles**](#secondary-subtitles) - Dual subtitle track support
 
@@ -127,7 +118,7 @@ The configuration file includes several main sections:
 - [**Discord Rich Presence**](#discord-rich-presence) - Optional Discord activity card updates
 - [**Immersion Tracking**](#immersion-tracking) - Track subtitle sessions and mining activity in SQLite
 - [**Stats Dashboard**](#stats-dashboard) - Local dashboard and overlay for immersion progress
-- [**YouTube Playback Settings**](#youtube-playback-settings) - Defaults for YouTube subtitle loading
+- [**YouTube Subtitle Generation**](#youtube-subtitle-generation) - Launcher defaults for yt-dlp + local whisper fallback
 
 ## Core Settings
 
@@ -202,8 +193,6 @@ Defaults warm everything (`true` for all toggles, `lowPowerMode: false`). Settin
 
 The overlay includes a built-in WebSocket server that broadcasts subtitle text to connected clients (such as texthooker-ui) for external processing.
 
-For endpoint details, payload examples, and client patterns, see [WebSocket / Texthooker API & Integration](/websocket-texthooker-api).
-
 By default, the server uses "auto" mode: it starts automatically unless [mpv_websocket](https://github.com/kuroahna/mpv_websocket) is detected at `~/.config/mpv/mpv_websocket`. If you have mpv_websocket installed, the built-in server is skipped to avoid conflicts.
 
 See `config.example.jsonc` for detailed configuration options.
@@ -348,48 +337,6 @@ Secondary subtitle defaults: `fontFamily: "Inter, Noto Sans, Helvetica Neue, san
 
 **See `config.example.jsonc`** for the complete list of subtitle style configuration options.
 
-### Subtitle Sidebar
-
-Configure the parsed-subtitle sidebar modal.
-
-```json
-{
-  "subtitleSidebar": {
-    "enabled": false,
-    "autoOpen": false,
-    "layout": "overlay",
-    "toggleKey": "Backslash",
-    "pauseVideoOnHover": false,
-    "autoScroll": true,
-    "fontFamily": "\"M PLUS 1\", \"Noto Sans CJK JP\", sans-serif",
-    "fontSize": 16
-  }
-}
-```
-
-| Option                      | Values           | Description                                                                      |
-| --------------------------- | ---------------- | -------------------------------------------------------------------------------- |
-| `enabled`                   | boolean          | Enable subtitle sidebar support (`false` by default)                             |
-| `autoOpen`                  | boolean          | Open sidebar automatically on overlay startup (`false` by default)                |
-| `layout`                    | string           | `"overlay"` floats over mpv; `"embedded"` reserves right-side player space to mimic browser-like layout |
-| `toggleKey`                 | string           | `KeyboardEvent.code` used to open/close the sidebar (default: `"Backslash"`)     |
-| `pauseVideoOnHover`         | boolean          | Pause playback while hovering the sidebar cue list                               |
-| `autoScroll`                | boolean          | Keep the active cue in view while playback advances                              |
-| `maxWidth`                  | number           | Maximum sidebar width in CSS pixels (default: `420`)                             |
-| `opacity`                   | number           | Sidebar opacity between `0` and `1` (default: `0.95`)                            |
-| `backgroundColor`           | string           | Sidebar shell background color                                                   |
-| `textColor`                 | hex color        | Default cue text color                                                           |
-| `fontFamily`                | string           | CSS `font-family` value applied to sidebar cue text                             |
-| `fontSize`                  | number           | Base sidebar cue font size in CSS pixels (default: `16`)                        |
-| `timestampColor`            | hex color        | Cue timestamp color                                                              |
-| `activeLineColor`           | hex color        | Active cue text color                                                            |
-| `activeLineBackgroundColor` | string           | Active cue background color                                                      |
-| `hoverLineBackgroundColor`  | string           | Hovered cue background color                                                     |
-
-The sidebar is only available when the active subtitle source has been parsed into a cue list. Default colors use Catppuccin Macchiato with a semi-transparent shell so the panel stays readable without feeling like an opaque settings dialog.
-
-`embedded` layout is intended to act like a split-pane view: it reserves player space with a right-side video margin and keeps interaction in both the player area and sidebar. If you see unexpected offset behavior in your environment, switch back to `overlay` to isolate sidebar placement.
-
 `jlptColors` keys are:
 
 | Key  | Default   | Description             |
@@ -469,7 +416,6 @@ See `config.example.jsonc` for detailed configuration options and more examples.
 | `Space`              | `["cycle", "pause"]`         | Toggle pause                          |
 | `KeyJ`               | `["cycle", "sid"]`           | Cycle primary subtitle track          |
 | `Shift+KeyJ`         | `["cycle", "secondary-sid"]` | Cycle secondary subtitle track        |
-| `Ctrl+Alt+KeyC`    | `["__youtube-picker-open"]`     | Open the manual YouTube subtitle picker |
 | `ArrowRight`         | `["seek", 5]`                | Seek forward 5 seconds                |
 | `ArrowLeft`          | `["seek", -5]`               | Seek backward 5 seconds               |
 | `ArrowUp`            | `["seek", 60]`               | Seek forward 60 seconds               |
@@ -741,7 +687,7 @@ Palette controls:
 ### Shared AI Provider
 
 Shared OpenAI-compatible transport settings live at the top level under `ai`.
-Anki reads this provider directly. Legacy subtitle fallback keeps the same provider shape for compatibility, then applies feature-local overrides where supported.
+Anki and YouTube subtitle cleanup both read this provider, then apply feature-local overrides where supported.
 
 ```json
 {
@@ -761,14 +707,12 @@ Anki reads this provider directly. Legacy subtitle fallback keeps the same provi
 | `apiKey`           | string                | Static API key for the shared provider               |
 | `apiKeyCommand`    | string                | Shell command used to resolve the API key            |
 | `baseUrl`          | string (URL)          | OpenAI-compatible base URL                           |
-| `model`            | string                | Optional model override for shared provider workflows   |
-| `systemPrompt`     | string                | Optional system prompt override for shared provider workflows |
 | `requestTimeoutMs` | integer milliseconds  | Shared request timeout (default: `15000`)            |
 
-SubMiner uses the shared provider for:
+SubMiner uses the shared provider in two places:
 
 - Anki translation/enrichment when `ankiConnect.ai.enabled` is `true`
-- Legacy subtitle fallback compatibility when `youtubeSubgen.fixWithAi` is `true`
+- YouTube whisper subtitle post-processing when `youtubeSubgen.fixWithAi` is `true`
 
 ### AnkiConnect
 
@@ -852,8 +796,8 @@ This example is intentionally compact. The option table below documents availabl
 | `proxy.port`                            | number                                  | Bind port for local AnkiConnect proxy (default: `8766`)                                                                                       |
 | `proxy.upstreamUrl`                     | string (URL)                            | Upstream AnkiConnect URL that proxy forwards to (default: `http://127.0.0.1:8765`)                                                            |
 | `tags`                                  | array of strings                        | Tags automatically added to cards mined/updated by SubMiner (default: `['SubMiner']`; set `[]` to disable automatic tagging).                 |
-| `ankiConnect.deck`                      | string                                  | Legacy Anki polling/compatibility scope. Newer known-word cache scoping should use `ankiConnect.knownWords.decks`.                               |
-| `ankiConnect.knownWords.decks`          | object                                  | Deck→fields mapping for known-word cache queries (for example `{ "Kaishi 1.5k": ["Word", "Word Reading"] }`).                                  |
+| `deck`                                  | string                                  | Anki deck to monitor for new cards                                                                                                            |
+| `ankiConnect.knownWords.decks`          | array of strings                        | Decks used for known-word cache lookups. When omitted/empty, falls back to `ankiConnect.deck`.                                                 |
 | `fields.word`                           | string                                  | Card field for mined word / expression text (default: `Expression`)                                                                            |
 | `fields.audio`                          | string                                  | Card field for audio files (default: `ExpressionAudio`)                                                                                       |
 | `fields.image`                          | string                                  | Card field for images (default: `Picture`)                                                                                                    |
@@ -874,7 +818,6 @@ This example is intentionally compact. The option table below documents availabl
 | `media.animatedMaxWidth`                | number (px)                             | Max width for animated AVIF (default: `640`)                                                                                                  |
 | `media.animatedMaxHeight`               | number (px)                             | Optional max height for animated AVIF. Unset keeps source aspect-constrained height.                                                          |
 | `media.animatedCrf`                     | number (0-63)                           | CRF quality for AVIF; lower = higher quality (default: `35`)                                                                                  |
-| `media.syncAnimatedImageToWordAudio`    | `true`, `false`                         | Whether animated AVIF includes an opening frame synced to sentence word-audio timing (default: `true`). |
 | `media.audioPadding`                    | number (seconds)                        | Padding around audio clip timing (default: `0.5`)                                                                                             |
 | `media.fallbackDuration`                | number (seconds)                        | Default duration if timing unavailable (default: `3.0`)                                                                                       |
 | `media.maxMediaDuration`                | number (seconds)                        | Max duration for generated media from multi-line copy (default: `30`, `0` to disable)                                                         |
@@ -883,11 +826,10 @@ This example is intentionally compact. The option table below documents availabl
 | `behavior.mediaInsertMode`              | `"append"`, `"prepend"`                 | Where to insert new media when overwrite is off (default: `"append"`)                                                                         |
 | `behavior.highlightWord`                | `true`, `false`                         | Highlight the word in sentence context (default: `true`)                                                                                      |
 | `ankiConnect.knownWords.highlightEnabled` | `true`, `false`                       | Enable fast local highlighting for words already known in Anki (default: `false`)                                                             |
-| `ankiConnect.knownWords.addMinedWordsImmediately` | `true`, `false`                         | Add words from successful mines into the local known-word cache immediately (default: `true`)                                                   |
 | `ankiConnect.knownWords.color`          | hex color string                        | Text color for tokens already found in the local known-word cache (default: `"#a6da95"`).                                                     |
 | `ankiConnect.knownWords.matchMode`      | `"headword"`, `"surface"`               | Matching strategy for known-word highlighting (default: `"headword"`). `headword` uses token headwords; `surface` uses visible subtitle text. |
 | `ankiConnect.knownWords.refreshMinutes` | number                                  | Minutes between known-word cache refreshes (default: `1440`)                                                                                  |
-| `ankiConnect.knownWords.decks`          | object                                  | Deck→fields mapping used for known-word cache query scope (e.g. `{ "Kaishi 1.5k": ["Word", "Word Reading"] }`).                              |
+| `ankiConnect.knownWords.decks`          | array of strings                        | Decks used by known-word cache refresh. Leave empty for compatibility with legacy `deck` scope.                                               |
 | `ankiConnect.nPlusOne.nPlusOne`         | hex color string                        | Text color for the single target token to study when exactly one unknown candidate exists in a sentence (default: `"#c6a0f6"`).               |
 | `ankiConnect.nPlusOne.minSentenceWords` | number                                  | Minimum number of words required in a sentence before single unknown-word N+1 highlighting can trigger (default: `3`).                        |
 | `behavior.notificationType`             | `"osd"`, `"system"`, `"both"`, `"none"` | Notification type on card update (default: `"osd"`)                                                                                           |
@@ -933,7 +875,7 @@ Known-word cache policy:
 - `ankiConnect.nPlusOne.nPlusOne` sets the color for the single target token when exactly one eligible unknown word exists.
 - `ankiConnect.nPlusOne.minSentenceWords` sets the minimum token count required in a sentence for N+1 highlighting (default: `3`).
 - `ankiConnect.knownWords.color` sets the known-word highlight color for tokens already in Anki.
-- `ankiConnect.knownWords.decks` accepts an object keyed by deck name. If omitted or empty, it falls back to the legacy `ankiConnect.deck` single-deck scope.
+- `ankiConnect.knownWords.decks` accepts one or more decks. If empty, it uses the legacy single `ankiConnect.deck` value as scope.
 - Cache state is persisted to `known-words-cache.json` under the app `userData` directory.
 - The cache is automatically invalidated when the configured scope changes (for example, when deck changes).
 - Cache lookups are in-memory. By default, token headwords are matched against cached `Expression` / `Word` values; set `ankiConnect.knownWords.matchMode` to `"surface"` for raw subtitle text matching.
@@ -1277,14 +1219,6 @@ Enable or disable local immersion analytics stored in SQLite for mined subtitles
 | `retention.monthlyRollupsDays` | integer (`0`-`36500`)         | Monthly rollup retention window. Default `0` (keep all).                                                    |
 | `retention.vacuumIntervalDays` | integer (`0`-`3650`)          | Minimum spacing between `VACUUM` passes. `0` disables vacuum. Default `0` (disabled).                         |
 
-You can also disable immersion tracking for a single session using:
-
-```bash
-SUBMINER_DISABLE_IMMERSION_TRACKING=1 subminer
-```
-
-When this is set, SubMiner skips immersion-tracker startup and does not initialize or read the immersion SQLite database for that session.
-
 Default behavior keeps raw events, telemetry, sessions, and rollups forever while still maintaining lifetime summary tables and daily/monthly rollups for faster reads. If you later want bounded retention, switch `retentionMode` or set explicit `retention.*` values.
 
 When `dbPath` is blank or omitted, SubMiner writes telemetry and session summaries to the default app-data location:
@@ -1305,7 +1239,7 @@ Configure the local stats UI served from SubMiner and the in-app stats overlay t
 {
   "stats": {
     "toggleKey": "Backquote",
-    "serverPort": 6969,
+    "serverPort": 5175,
     "autoStartServer": true,
     "autoOpenBrowser": true
   }
@@ -1315,7 +1249,7 @@ Configure the local stats UI served from SubMiner and the in-app stats overlay t
 | Option            | Values            | Description                                                                 |
 | ----------------- | ----------------- | --------------------------------------------------------------------------- |
 | `toggleKey`       | Electron key code | Overlay-local key code used to toggle the stats overlay. Default `Backquote`. |
-| `serverPort`      | integer           | Localhost port for the browser stats UI. Default `6969`.                    |
+| `serverPort`      | integer           | Localhost port for the browser stats UI. Default `5175`.                    |
 | `autoStartServer` | `true`, `false`   | Start the local stats HTTP server automatically once immersion tracking is active. Default `true`. |
 | `autoOpenBrowser` | `true`, `false`   | When `subminer stats` starts the server on demand, also open the dashboard in your default browser. Default `true`. |
 
@@ -1326,13 +1260,22 @@ Usage notes:
 - The dashboard reads from the same immersion-tracking database, so keep `immersionTracking.enabled` on if you want data to appear.
 - The UI includes Overview, Library, Trends, Vocabulary, and Sessions tabs.
 
-### YouTube Playback Settings
+### YouTube Subtitle Generation
 
-Set defaults used by the `subminer` launcher for YouTube subtitle loading:
+Set defaults used by the `subminer` launcher for YouTube subtitle generation:
 
 ```json
 {
   "youtubeSubgen": {
+    "whisperBin": "/path/to/whisper-cli",
+    "whisperModel": "/path/to/ggml-model.bin",
+    "whisperVadModel": "/path/to/ggml-vad.bin",
+    "whisperThreads": 4,
+    "fixWithAi": false,
+    "ai": {
+      "model": "openai/gpt-4o-mini",
+      "systemPrompt": "Fix subtitle mistakes only."
+    },
     "primarySubLanguages": ["ja", "jpn"]
   }
 }
@@ -1340,22 +1283,27 @@ Set defaults used by the `subminer` launcher for YouTube subtitle loading:
 
 | Option                | Values               | Description                                                                                    |
 | --------------------- | -------------------- | ---------------------------------------------------------------------------------------------- |
-| `primarySubLanguages` | string[]             | Primary subtitle language priority for YouTube auto-loading (default `["ja", "jpn"]`) |
+| `whisperBin`          | string path          | Path to `whisper.cpp` CLI binary used as fallback transcription engine                        |
+| `whisperModel`        | string path          | Path to whisper model used by fallback transcription                                          |
+| `whisperVadModel`     | string path          | Optional whisper VAD model path                                                               |
+| `whisperThreads`      | integer              | Thread count passed to whisper runs                                                           |
+| `fixWithAi`           | `true`, `false`      | Run shared AI post-processing on whisper-generated subtitles                                  |
+| `ai.model`            | string               | Optional model override for YouTube AI subtitle cleanup                                       |
+| `ai.systemPrompt`     | string               | Optional system prompt override for YouTube AI subtitle cleanup                               |
+| `primarySubLanguages` | string[]             | Primary subtitle language priority for YouTube subtitle generation (default `["ja", "jpn"]`) |
 
-Current launcher behavior:
+Launcher behavior:
 
-- For YouTube URLs, SubMiner probes subtitle tracks with yt-dlp after mpv bootstrap and binds auto-selected tracks before normal playback resumes.
-- If YouTube/mpv already exposes an authoritative matching subtitle track, SubMiner reuses it; otherwise it downloads and injects only the missing side.
-- SubMiner loads the primary subtitle plus a best-effort secondary subtitle.
-- Playback waits only for primary subtitle readiness; secondary failures do not block playback.
-- English secondary subtitles are selected from `secondarySub.secondarySubLanguages` when primary language matches are unavailable.
-- Native mpv secondary subtitle rendering stays hidden during this flow so the SubMiner overlay remains the visible secondary subtitle surface.
-- If primary subtitle loading fails, use `Ctrl+Alt+C` to open the subtitle modal and pick a track.
+- For YouTube URLs, subtitle generation now runs before mpv launch.
+- SubMiner probes manual/native YouTube subtitle tracks first.
+- Missing tracks fall back to local `whisper.cpp`.
+- English secondary subtitles can use whisper translate fallback when no manual track exists.
+- If `fixWithAi` is enabled, only whisper-generated `.srt` output is post-processed with the shared top-level `ai` provider.
 
 Language targets are derived from subtitle config:
 
 - primary track: `youtubeSubgen.primarySubLanguages` (falls back to `["ja","jpn"]`)
 - secondary track: `secondarySub.secondarySubLanguages` (falls back to English when empty)
-- Tracks are resolved and loaded before mpv starts; the older launcher mode switch has been removed.
+- Subtitle files are generated or downloaded before mpv starts; the older launcher mode switch has been removed.
 
 Precedence for launcher defaults is: CLI flag > environment variable > `config.jsonc` > built-in default.

docs-site/development.md

@@ -231,13 +231,13 @@ Run `make help` for a full list of targets. Key ones:
 | `SUBMINER_ROFI_THEME`              | Override rofi theme path for launcher picker                                   |
 | `SUBMINER_LOG_LEVEL`               | Override app logger level (`debug`, `info`, `warn`, `error`)                   |
 | `SUBMINER_MPV_LOG`                 | Override mpv/app shared log file path                                          |
-| `SUBMINER_WHISPER_BIN`             | Override legacy `youtubeSubgen.whisperBin` fallback compatibility path          |
-| `SUBMINER_WHISPER_MODEL`           | Override legacy `youtubeSubgen.whisperModel` fallback compatibility path        |
-| `SUBMINER_WHISPER_VAD_MODEL`       | Override legacy `youtubeSubgen.whisperVadModel` fallback compatibility path      |
-| `SUBMINER_WHISPER_THREADS`         | Override legacy `youtubeSubgen.whisperThreads` fallback compatibility value      |
-| `SUBMINER_YT_SUBGEN_OUT_DIR`       | Override legacy fallback subtitle output directory                               |
-| `SUBMINER_YT_SUBGEN_AUDIO_FORMAT`  | Override extraction format used by legacy fallback subtitle path                  |
-| `SUBMINER_YT_SUBGEN_KEEP_TEMP`     | Set to `1` to keep legacy fallback subtitle workspace                          |
+| `SUBMINER_WHISPER_BIN`             | Override `youtubeSubgen.whisperBin` for launcher                               |
+| `SUBMINER_WHISPER_MODEL`           | Override `youtubeSubgen.whisperModel` for launcher                             |
+| `SUBMINER_WHISPER_VAD_MODEL`       | Override `youtubeSubgen.whisperVadModel` for launcher                          |
+| `SUBMINER_WHISPER_THREADS`         | Override `youtubeSubgen.whisperThreads` for launcher                           |
+| `SUBMINER_YT_SUBGEN_OUT_DIR`       | Override generated subtitle output directory                                   |
+| `SUBMINER_YT_SUBGEN_AUDIO_FORMAT`  | Override extraction format used for whisper fallback                           |
+| `SUBMINER_YT_SUBGEN_KEEP_TEMP`     | Set to `1` to keep temporary subtitle-generation workspace                     |
 | `SUBMINER_JIMAKU_API_KEY`          | Override Jimaku API key for launcher subtitle downloads                        |
 | `SUBMINER_JIMAKU_API_KEY_COMMAND`  | Command used to resolve Jimaku API key at runtime                              |
 | `SUBMINER_JIMAKU_API_BASE_URL`     | Override Jimaku API base URL                                                   |

docs-site/docs-sync.test.ts

@@ -20,7 +20,7 @@ function extractReleaseHeadings(content: string, count: number): string[] {
 test('docs reflect current launcher and release surfaces', () => {
   expect(usageContents).not.toContain('--mode preprocess');
   expect(usageContents).not.toContain('"automatic" (default)');
-  expect(usageContents).toContain('during startup while mpv is paused');
+  expect(usageContents).toContain('before mpv starts');
 
   expect(installationContents).toContain('bun run build:appimage');
   expect(installationContents).toContain('bun run build:win');

docs-site/immersion-tracking.md

@@ -28,7 +28,7 @@ The same immersion data powers the stats dashboard.
 - Launcher command: run `subminer stats` to start the local stats server on demand and open the dashboard in your browser.
 - Background server: run `subminer stats -b` to start or reuse a dedicated background stats daemon without keeping the launcher attached, and `subminer stats -s` to stop that daemon.
 - Maintenance command: run `subminer stats cleanup` or `subminer stats cleanup -v` to backfill/repair vocabulary metadata (`headword`, `reading`, POS) and purge stale or excluded rows from `imm_words` on demand.
-- Browser page: open `http://127.0.0.1:6969` directly if the local stats server is already running.
+- Browser page: open `http://127.0.0.1:5175` directly if the local stats server is already running.
 
 ### Dashboard Tabs
 
@@ -42,8 +42,6 @@ Recent sessions, streak calendar, watch-time history, and a tracking snapshot wi
 
 Cover-art library with search and sorting, per-series progress, episode drill-down, and direct links into mined cards.
 
-When YouTube channel metadata is available, the Library tab groups videos by creator/channel and treats each tracked video as an episode-like entry inside that channel section.
-
 ![Stats Library](/screenshots/stats-library.png)
 
 #### Trends
@@ -70,7 +68,7 @@ Stats server config lives under `stats`:
 {
   "stats": {
     "toggleKey": "Backquote",
-    "serverPort": 6969,
+    "serverPort": 5175,
     "autoStartServer": true,
     "autoOpenBrowser": true
   }

docs-site/index.md

@@ -51,8 +51,8 @@ features:
   - icon:
       src: /assets/video.svg
       alt: Video playback icon
-    title: YouTube Playback
-    details: Play YouTube URLs or ytsearch targets directly — SubMiner automatically selects and loads subtitles for the video.
+    title: YouTube & Whisper
+    details: Play YouTube URLs or searches with native subtitles, or generate them with whisper.cpp and optional AI cleanup.
     link: /usage#youtube-playback
     linkText: YouTube playback
   - icon:
@@ -72,10 +72,10 @@ features:
   - icon:
       src: /assets/tokenization.svg
       alt: Tracking chart icon
-    title: Stats Dashboard
-    details: Browse session history, streak calendars, vocabulary frequency, and per-series progress in a local dashboard — then mine cards straight from your viewing history.
+    title: Immersion Tracking
+    details: Logs watch time, words encountered, and cards mined to SQLite, then surfaces the same data in a local stats dashboard with rollups and session drill-down.
     link: /immersion-tracking
-    linkText: Dashboard & tracking
+    linkText: Stats details
   - icon:
       src: /assets/cross-platform.svg
       alt: Cross-platform icon
@@ -120,7 +120,7 @@ const demoAssetVersion = '20260223-2';
       <div class="workflow-step" style="animation-delay: 240ms">
         <div class="step-number">05</div>
         <div class="step-title">Track</div>
-        <div class="step-desc">Open the stats dashboard to review sessions, vocabulary trends, and mine cards from past viewing history.</div>
+        <div class="step-desc">Review immersion history and repeat high-value patterns over time.</div>
       </div>
     </div>
   </section>

docs-site/launcher-script.md

@@ -58,21 +58,15 @@ subminer --start video.mkv      # optional explicit overlay start when plugin au
 subminer -S video.mkv           # same as above via --start-overlay
 subminer https://youtu.be/...   # YouTube playback (requires yt-dlp)
 subminer ytsearch:"jp news"     # YouTube search
-subminer stats                  # open immersion dashboard
-subminer stats -b               # start background stats daemon
-subminer stats -s               # stop background stats daemon
 subminer --setup                # Open first-run setup popup
 ```
 
 ## Subcommands
 
 | Subcommand                 | Purpose                                                    |
-| ---------------------------- | ---------------------------------------------------------- |
+| -------------------------- | ---------------------------------------------------------- |
 | `subminer jellyfin` / `jf` | Jellyfin workflows (`-d` discovery, `-p` play, `-l` login) |
-| `subminer stats`             | Start stats server and open immersion dashboard in browser |
-| `subminer stats -b`          | Start or reuse background stats daemon (non-blocking)      |
-| `subminer stats -s`          | Stop the background stats daemon                           |
-| `subminer stats cleanup`     | Backfill vocabulary metadata and prune stale rows          |
+| `subminer yt` / `youtube`  | YouTube shorthand (`-o`, `-m`)                             |
 | `subminer doctor`          | Dependency + config + socket diagnostics                   |
 | `subminer config path`     | Print active config file path                              |
 | `subminer config show`     | Print active config contents                               |

docs-site/mining-workflow.md

@@ -6,28 +6,11 @@ This guide walks through the sentence mining loop — from watching a video to c
 
 SubMiner runs as a transparent overlay on top of mpv. As subtitles play, the overlay displays them as interactive text. You hover a word, trigger Yomitan lookup with your configured lookup key/modifier, then create an Anki card with a single action. SubMiner automatically attaches the sentence, audio clip, and screenshot.
 
-```mermaid
-flowchart LR
-  classDef step fill:#c6a0f6,stroke:#494d64,color:#24273a,stroke-width:1.5px
-  classDef action fill:#8aadf4,stroke:#494d64,color:#24273a,stroke-width:1.5px
-  classDef result fill:#a6da95,stroke:#494d64,color:#24273a,stroke-width:1.5px
-  classDef enrich fill:#8bd5ca,stroke:#494d64,color:#24273a,stroke-width:1.5px
-
-  Watch["Watch Video"]:::step
-  Sub["Subtitle Appears"]:::step
-  Hover["Hover Word"]:::action
-  Lookup["Trigger Lookup"]:::action
-  Yomi["Yomitan Popup"]:::result
-  Add["Add to Anki"]:::result
-
-  Watch --> Sub --> Hover --> Lookup --> Yomi --> Add
-
-  Add --> Enrich["SubMiner Enriches"]:::enrich
-
-  Enrich --> S["Sentence"]:::enrich
-  Enrich --> A["Audio Clip"]:::enrich
-  Enrich --> I["Screenshot"]:::enrich
-  Enrich --> T["Translation"]:::enrich
+```text
+Watch video → See subtitle → Hover word + trigger lookup → Yomitan popup → Add to Anki
+                                                              ↓
+                                              SubMiner auto-fills:
+                                              sentence, audio, image, translation
 ```
 
 ## Subtitle Delivery Path (Startup + Runtime)
@@ -193,8 +176,6 @@ SubMiner runs a local HTTP server at `http://127.0.0.1:5174` (configurable port)
 
 The texthooker page displays the current subtitle and updates as new lines arrive. This is useful if you prefer to do lookups in a browser rather than through the overlay's built-in Yomitan.
 
-If you want to build your own browser client, websocket consumer, or automation relay, see [WebSocket / Texthooker API & Integration](/websocket-texthooker-api).
-
 ## Subtitle Sync (Subsync)
 
 If your subtitle file is out of sync with the audio, SubMiner can resynchronize it using [alass](https://github.com/kaegi/alass) or [ffsubsync](https://github.com/smacke/ffsubsync).
@@ -225,7 +206,7 @@ Enable it in your config:
 }
 ```
 
-Open the dashboard in the overlay with `stats.toggleKey` (default: `` ` ``), launch it in a browser with `subminer stats`, keep a dedicated background server alive with `subminer stats -b`, stop that background server with `subminer stats -s`, or visit `http://127.0.0.1:6969` directly if the local stats server is already running. The dashboard covers overview totals, anime progress, session detail, and vocabulary drill-down from the same local immersion database.
+Open the dashboard in the overlay with `stats.toggleKey` (default: `` ` ``), launch it in a browser with `subminer stats`, keep a dedicated background server alive with `subminer stats -b`, stop that background server with `subminer stats -s`, or visit `http://127.0.0.1:5175` directly if the local stats server is already running. The dashboard covers overview totals, anime progress, session detail, and vocabulary drill-down from the same local immersion database.
 
 See [Immersion Tracking](/immersion-tracking) for dashboard details, schema, and retention settings.
 

docs-site/public/config.example.jsonc

@@ -187,7 +187,7 @@
   // ==========================================
   // Secondary Subtitles
   // Dual subtitle track options.
-  // Used by the YouTube subtitle loading flow as secondary language preferences.
+  // Used by subminer YouTube subtitle generation as secondary language preferences.
   // Hot-reload: defaultMode updates live while SubMiner is running.
   // ==========================================
   "secondarySub": {
@@ -284,30 +284,6 @@
     } // Secondary setting.
   }, // Primary and secondary subtitle styling.
 
-  // ==========================================
-  // Subtitle Sidebar
-  // Parsed-subtitle sidebar cue list styling, behavior, and toggle key.
-  // Hot-reload: subtitle sidebar changes apply live without restarting SubMiner.
-  // ==========================================
-  "subtitleSidebar": {
-    "enabled": false, // Enable the subtitle sidebar feature for parsed subtitle sources. Values: true | false
-    "autoOpen": false, // Automatically open the subtitle sidebar once during overlay startup. Values: true | false
-    "layout": "overlay", // Render the subtitle sidebar as a floating overlay or reserve space inside mpv. Values: overlay | embedded
-    "toggleKey": "Backslash", // KeyboardEvent.code used to toggle the subtitle sidebar open and closed.
-    "pauseVideoOnHover": false, // Pause mpv while hovering the subtitle sidebar, then resume on leave. Values: true | false
-    "autoScroll": true, // Auto-scroll the active subtitle cue into view while playback advances. Values: true | false
-    "maxWidth": 420, // Maximum sidebar width in CSS pixels.
-    "opacity": 0.95, // Base opacity applied to the sidebar shell.
-    "backgroundColor": "rgba(73, 77, 100, 0.9)", // Background color for the subtitle sidebar shell.
-    "textColor": "#cad3f5", // Default cue text color in the subtitle sidebar.
-    "fontFamily": "\"M PLUS 1\", \"Noto Sans CJK JP\", sans-serif", // Font family used for subtitle sidebar cue text.
-    "fontSize": 16, // Base font size for subtitle sidebar cue text in CSS pixels.
-    "timestampColor": "#a5adcb", // Timestamp color in the subtitle sidebar.
-    "activeLineColor": "#f5bde6", // Text color for the active subtitle cue.
-    "activeLineBackgroundColor": "rgba(138, 173, 244, 0.22)", // Background color for the active subtitle cue.
-    "hoverLineBackgroundColor": "rgba(54, 58, 79, 0.84)" // Background color for hovered subtitle cues.
-  }, // Parsed-subtitle sidebar cue list styling, behavior, and toggle key.
-
   // ==========================================
   // Shared AI Provider
   // Canonical OpenAI-compatible provider transport settings shared by Anki and YouTube subtitle fixing.
@@ -414,24 +390,24 @@
   }, // Jimaku API configuration and defaults.
 
   // ==========================================
-  // YouTube Playback Settings
-  // Defaults for SubMiner YouTube subtitle loading and languages.
+  // YouTube Subtitle Generation
+  // Defaults for SubMiner YouTube subtitle generation.
   // ==========================================
   "youtubeSubgen": {
-    "whisperBin": "", // Legacy compatibility path kept for external subtitle fallback tools; not used by default.
-    "whisperModel": "", // Legacy compatibility model path kept for external subtitle fallback tooling; not used by default.
-    "whisperVadModel": "", // Legacy compatibility VAD path kept for external subtitle fallback tooling; not used by default.
-    "whisperThreads": 4, // Legacy thread tuning for subtitle fallback tooling; not used by default.
-    "fixWithAi": false, // Legacy subtitle fallback post-processing switch kept for compatibility; use is currently disabled by default. Values: true | false
+    "whisperBin": "", // Path to whisper.cpp CLI used as fallback transcription engine.
+    "whisperModel": "", // Path to whisper model used for fallback transcription.
+    "whisperVadModel": "", // Path to optional whisper VAD model used for subtitle generation.
+    "whisperThreads": 4, // Thread count passed to whisper.cpp subtitle generation runs.
+    "fixWithAi": false, // Use shared AI provider to post-process whisper-generated YouTube subtitles. Values: true | false
     "ai": {
-      "model": "", // Optional model override for legacy subtitle fallback post-processing; not used by default.
-      "systemPrompt": "" // Optional system prompt override for legacy subtitle fallback post-processing; not used by default.
+      "model": "", // Optional model override for YouTube subtitle AI post-processing.
+      "systemPrompt": "" // Optional system prompt override for YouTube subtitle AI post-processing.
     }, // Ai setting.
     "primarySubLanguages": [
       "ja",
       "jpn"
     ] // Comma-separated primary subtitle language priority used by the launcher.
-  }, // Defaults for SubMiner YouTube subtitle loading and languages.
+  }, // Defaults for SubMiner YouTube subtitle generation.
 
   // ==========================================
   // Anilist

docs-site/shortcuts.md

@@ -67,15 +67,11 @@ Mouse-hover playback behavior is configured separately from shortcuts: `subtitle
 | `Ctrl/Cmd+Shift+V` | Cycle secondary subtitle mode (hidden → visible → hover) | `shortcuts.toggleSecondarySub` |
 | `Ctrl/Cmd+Shift+O` | Open runtime options palette                             | `shortcuts.openRuntimeOptions` |
 | `Ctrl+Shift+J`     | Open Jimaku subtitle search modal                        | `shortcuts.openJimaku`         |
-| `Ctrl+Alt+C`       | Open the manual YouTube subtitle picker                  | `keybindings`                  |
 | `Ctrl+Alt+S`       | Open subtitle sync (subsync) modal                       | `shortcuts.triggerSubsync`     |
-| `\`                | Toggle subtitle sidebar                                  | `subtitleSidebar.toggleKey`    |
 | `` ` ``            | Toggle stats overlay                                     | `stats.toggleKey`              |
 
 The stats toggle is handled inside the focused visible overlay window. It is configurable through the top-level `stats.toggleKey` setting and defaults to `Backquote`.
 
-The subtitle sidebar toggle is overlay-local and only opens when SubMiner has a parsed cue list for the active subtitle source.
-
 ## Controller Shortcuts
 
 These overlay-local shortcuts are fixed and open controller utilities for the Chrome Gamepad API integration.
@@ -137,4 +133,4 @@ The `keybindings` array overrides or extends the overlay's built-in key handling
 }
 ```
 
-Both `shortcuts`, `keybindings`, and `subtitleSidebar` are [hot-reloadable](/configuration#hot-reload-behavior) — changes take effect without restarting SubMiner.
+Both `shortcuts` and `keybindings` are [hot-reloadable](/configuration#hot-reload-behavior) — changes take effect without restarting SubMiner.

docs-site/subtitle-annotations.md

@@ -24,7 +24,7 @@ N+1 highlighting identifies sentences where you know every word except one, maki
 | --- | --- | --- |
 | `ankiConnect.knownWords.highlightEnabled` | `false` | Enable known-word cache lookups used by N+1 highlighting |
 | `ankiConnect.knownWords.refreshMinutes` | `1440` | Minutes between Anki cache refreshes |
-| `ankiConnect.knownWords.decks` | `{}` | Deck→fields map for known-word cache queries (legacy fallback: `ankiConnect.deck`) |
+| `ankiConnect.knownWords.decks` | `[]` | Decks to query (falls back to `ankiConnect.deck`) |
 | `ankiConnect.knownWords.matchMode` | `"headword"` | `"headword"` (dictionary form) or `"surface"` (raw text) |
 | `ankiConnect.nPlusOne.minSentenceWords` | `3` | Minimum tokens in a sentence for N+1 to trigger |
 | `ankiConnect.nPlusOne.nPlusOne` | `#c6a0f6` | Color for the single unknown target word |

docs-site/troubleshooting.md

@@ -29,7 +29,7 @@ SubMiner retries the connection automatically with increasing delays (200 ms, 50
 - Common spikes come from:
   - first subtitle parse/tokenization bursts
   - media generation (`ffmpeg` audio/image and AVIF paths)
-  - media sync and subtitle tooling (`alass`, `ffsubsync`)
+  - media sync and subtitle tooling (`alass`, `ffsubsync`, `whisper` fallback path)
   - `ankiConnect` enrichment (plus polling overhead when proxy mode is disabled)
 
 ### If playback feels sluggish
@@ -57,7 +57,7 @@ SubMiner retries the connection automatically with increasing delays (200 ms, 50
 
 - disable AI translation when not needed (`ankiConnect.ai.enabled: false`)
 - if needed, run immersion telemetry with lower duration expectations (`immersionTracking.enabled: false` for constrained sessions)
-- favor the default lightweight YouTube subtitle startup settings on low-resource systems
+- prefer YouTube `--mode automatic` over `preprocess` on low-resource systems
 
 ### Practical low-impact profile
 

docs-site/usage.md

@@ -78,6 +78,8 @@ subminer mpv idle                 # Launch detached idle mpv with SubMiner defau
 subminer dictionary /path/to/file-or-directory  # Generate character dictionary ZIP from target (manual Yomitan import)
 subminer texthooker               # Launch texthooker-only mode
 subminer app --anilist            # Pass args directly to SubMiner binary (example: AniList login flow)
+subminer yt -o ~/subs https://youtu.be/...  # YouTube subcommand: output directory shortcut
+subminer yt --keep-temp --whisper-bin /path/to/whisper-cli --whisper-model /path/to/model.bin --whisper-vad-model /path/to/ggml-vad.bin https://youtu.be/...  # Keep generated subtitle workspace for debugging
 
 # Direct packaged app control
 SubMiner.AppImage --background             # Start in background (tray + IPC wait, minimal logs)
@@ -135,13 +137,14 @@ This flow requires `mpv.exe` to be on `PATH`. If it is installed elsewhere, set
 ### Launcher Subcommands
 
 - `subminer jellyfin` / `subminer jf`: Jellyfin-focused workflow aliases.
+- `subminer yt` / `subminer youtube`: YouTube-focused shorthand flags (`-o`, `--keep-temp`, `--whisper-*`).
 - `subminer doctor`: health checks for core dependencies and runtime paths.
 - `subminer config`: config 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.
 - `subminer texthooker`: texthooker-only shortcut (same behavior as `--texthooker`).
 - `subminer app` / `subminer bin`: direct passthrough to the SubMiner binary/AppImage.
-- Subcommand help pages are available (for example `subminer jellyfin -h`).
+- Subcommand help pages are available (for example `subminer jellyfin -h`, `subminer yt -h`).
 
 ### First-Run Setup
 
@@ -171,8 +174,8 @@ AniList character dictionary auto-sync (optional):
 - SubMiner syncs the currently watched AniList media into a per-media snapshot, then rebuilds one merged `SubMiner Character Dictionary` from the most recently used snapshots.
 - Rotation limit defaults to 3 recent media snapshots in that merged dictionary (`maxLoaded`).
 
-Use subcommands for Jellyfin workflows (`subminer jellyfin ...`).
-Top-level launcher flags like `--jellyfin-*` are intentionally rejected.
+Use subcommands for Jellyfin/YouTube command families (`subminer jellyfin ...`, `subminer yt ...`).
+Top-level launcher flags like `--jellyfin-*` and `--yt-subgen-*` are intentionally rejected.
 
 ### MPV Profile Example (mpv.conf)
 
@@ -225,18 +228,26 @@ If you also use Yomitan in a browser, configure that browser profile separately;
 ### YouTube Playback
 
 `subminer` accepts direct URLs (for example, YouTube links) and `ytsearch:` targets.
-For YouTube playback, SubMiner resolves subtitle selection during startup while mpv is paused: it auto-selects the default primary subtitle track plus a best-effort secondary track, then resumes when primary subtitles are ready.
+For YouTube playback, SubMiner now generates or downloads subtitle tracks before mpv starts, then launches mpv with the resolved subtitle files attached.
 
 Notes:
 
 - Install `yt-dlp` so mpv can resolve YouTube streams and subtitle tracks reliably.
-- For YouTube URLs, startup no longer requires opening the picker first; SubMiner loads subtitles and keeps the overlay available for retries.
-- Press `Ctrl+Alt+C` during active YouTube playback to open the manual YouTube subtitle picker and retry track selection.
-- For YouTube URLs, `subminer` probes available YouTube subtitle tracks, reuses existing authoritative tracks when available, and downloads only missing sides.
-- Native mpv secondary subtitle rendering stays hidden so the overlay remains the visible secondary subtitle surface.
+- For YouTube URLs, `subminer` now generates any missing subtitles before mpv launch.
+- It probes manual/native YouTube subtitle tracks first, then falls back to local `whisper.cpp` only for missing tracks.
 - Primary subtitle target languages come from `youtubeSubgen.primarySubLanguages` (defaults to `["ja","jpn"]`).
 - Secondary target languages come from `secondarySub.secondarySubLanguages` (defaults to English if unset).
-- Configure defaults in `$XDG_CONFIG_HOME/SubMiner/config.jsonc` (or `~/.config/SubMiner/config.jsonc`) under `youtubeSubgen` and `secondarySub`.
+- Whisper translation fallback currently only supports English secondary targets; non-English secondary targets rely on native/manual subtitle availability.
+- Optional AI cleanup for whisper-generated subtitles is controlled by `youtubeSubgen.fixWithAi` plus the shared top-level `ai` config (with optional `youtubeSubgen.ai` overrides).
+- Configure defaults in `$XDG_CONFIG_HOME/SubMiner/config.jsonc` (or `~/.config/SubMiner/config.jsonc`) under `youtubeSubgen`, `secondarySub`, and `ai`.
+- CLI overrides are available through `subminer yt` / `subminer youtube`:
+  - `-o, --out-dir`
+  - `--keep-temp`
+  - `--whisper-bin`
+  - `--whisper-model`
+  - `--whisper-vad-model`
+  - `--whisper-threads`
+  - `--yt-subgen-audio-format`
 
 ## Controller Support
 

docs-site/websocket-texthooker-api.md

@@ -1,357 +0,0 @@
-# WebSocket / Texthooker API & Integration
-
-SubMiner exposes a small set of local integration surfaces for browser tools, automation helpers, and mpv-driven workflows:
-
-- **Subtitle WebSocket** at `ws://127.0.0.1:6677` by default for plain subtitle pushes.
-- **Annotation WebSocket** at `ws://127.0.0.1:6678` by default for token-aware clients.
-- **Texthooker HTTP UI** at `http://127.0.0.1:5174` by default for browser-based subtitle consumption.
-- **mpv plugin script messages** for in-player automation and extension.
-
-This page documents those integration points and shows how to build custom consumers around them.
-
-## Quick Reference
-
-| Surface | Default | Purpose |
-| --- | --- | --- |
-| `websocket` | `ws://127.0.0.1:6677` | Basic subtitle broadcast stream |
-| `annotationWebsocket` | `ws://127.0.0.1:6678` | Structured stream with token metadata |
-| `texthooker` | `http://127.0.0.1:5174` | Local texthooker UI with injected websocket config |
-| mpv plugin | `script-message subminer-*` | Start/stop/toggle/status automation inside mpv |
-
-## Enable and Configure the Services
-
-SubMiner's integration ports are configured in `config.jsonc`.
-
-```jsonc
-{
-  "websocket": {
-    "enabled": "auto",
-    "port": 6677
-  },
-  "annotationWebsocket": {
-    "enabled": true,
-    "port": 6678
-  },
-  "texthooker": {
-    "launchAtStartup": true,
-    "openBrowser": true
-  }
-}
-```
-
-### How startup behaves
-
-- `websocket.enabled: "auto"` starts the basic subtitle websocket unless SubMiner detects the external `mpv_websocket` plugin.
-- `annotationWebsocket` is independent from `websocket` and stays enabled unless you explicitly disable it.
-- `texthooker.launchAtStartup` starts the local HTTP UI automatically.
-- `texthooker.openBrowser` controls whether SubMiner opens the texthooker page in your browser when it starts.
-
-If you use the [mpv plugin](/mpv-plugin), it can also start a texthooker-only helper process and override the texthooker port in `subminer.conf`.
-
-## Developer API Documentation
-
-### 1. Subtitle WebSocket
-
-Use the basic subtitle websocket when you only need the current subtitle line and a ready-to-render HTML sentence string.
-
-- **Default URL:** `ws://127.0.0.1:6677`
-- **Transport:** local WebSocket server bound to `127.0.0.1`
-- **Direction:** server push only
-- **Client auth:** none
-- **Reconnects:** client-managed
-
-When a client connects, SubMiner immediately sends the latest subtitle payload if one is available. After that, it pushes a new message each time the current subtitle changes.
-
-#### Message shape
-
-```json
-{
-  "version": 1,
-  "text": "無事",
-  "sentence": "<span class=\"word word-known word-jlpt-n2\" data-reading=\"ぶじ\" data-headword=\"無事\" data-frequency-rank=\"745\" data-jlpt-level=\"N2\">無事</span>",
-  "tokens": [
-    {
-      "surface": "無事",
-      "reading": "ぶじ",
-      "headword": "無事",
-      "startPos": 0,
-      "endPos": 2,
-      "partOfSpeech": "other",
-      "isMerged": false,
-      "isKnown": true,
-      "isNPlusOneTarget": false,
-      "isNameMatch": false,
-      "jlptLevel": "N2",
-      "frequencyRank": 745,
-      "className": "word word-known word-jlpt-n2",
-      "frequencyRankLabel": "745",
-      "jlptLevelLabel": "N2"
-    }
-  ]
-}
-```
-
-#### Field reference
-
-| Field | Type | Notes |
-| --- | --- | --- |
-| `version` | number | Current websocket payload version. Today this is `1`. |
-| `text` | string | Raw subtitle text. |
-| `sentence` | string | HTML string with `<span>` wrappers and `data-*` attributes for client rendering. |
-| `tokens` | array | Token metadata; empty when the subtitle is not tokenized yet. |
-
-Each token may include:
-
-| Token field | Type | Notes |
-| --- | --- | --- |
-| `surface` | string | Display text for the token |
-| `reading` | string | Kana reading when available |
-| `headword` | string | Dictionary headword when available |
-| `startPos` / `endPos` | number | Character offsets in the subtitle text |
-| `partOfSpeech` | string | SubMiner token POS label |
-| `isMerged` | boolean | Whether this token represents merged content |
-| `isKnown` | boolean | Marked known by SubMiner's known-word logic |
-| `isNPlusOneTarget` | boolean | True when the token is the sentence's N+1 target |
-| `isNameMatch` | boolean | True for prioritized character-name matches |
-| `frequencyRank` | number | Frequency rank when available |
-| `jlptLevel` | string | JLPT level when available |
-| `className` | string | CSS-ready class list derived from token state |
-| `frequencyRankLabel` | string or `null` | Preformatted rank label for UIs |
-| `jlptLevelLabel` | string or `null` | Preformatted JLPT label for UIs |
-
-### 2. Annotation WebSocket
-
-Use the annotation websocket for custom clients that want the same structured token payload the bundled texthooker UI consumes.
-
-- **Default URL:** `ws://127.0.0.1:6678`
-- **Payload shape:** same JSON contract as the basic subtitle websocket
-- **Primary difference:** this stream is intended to stay on even when the basic websocket auto-disables because `mpv_websocket` is installed
-
-In practice, if you are building a new client, prefer `annotationWebsocket` unless you specifically need compatibility with an existing `websocket` consumer.
-
-### 3. HTML markup conventions
-
-The `sentence` field is pre-rendered HTML generated by SubMiner. Depending on token state, it can include classes such as:
-
-- `word`
-- `word-known`
-- `word-n-plus-one`
-- `word-name-match`
-- `word-jlpt-n1` through `word-jlpt-n5`
-- `word-frequency-single`
-- `word-frequency-band-1` through `word-frequency-band-5`
-
-SubMiner also adds tooltip-friendly data attributes when available:
-
-- `data-reading`
-- `data-headword`
-- `data-frequency-rank`
-- `data-jlpt-level`
-
-If you need a fully custom UI, ignore `sentence` and render from `tokens` instead.
-
-## Texthooker Integration Guide
-
-### When to use the bundled texthooker page
-
-Use texthooker when you want a browser tab that:
-
-- updates live from current subtitles
-- works well with browser-based Yomitan setups
-- inherits SubMiner's coloring preferences and websocket URL automatically
-
-Start it with either:
-
-```bash
-subminer texthooker
-```
-
-or by leaving `texthooker.launchAtStartup` enabled.
-
-### What SubMiner injects into the page
-
-When SubMiner serves the local texthooker UI, it injects bootstrap values into `window.localStorage`, including:
-
-- `bannou-texthooker-websocketUrl`
-- coloring toggles for known/N+1/name/frequency/JLPT styling
-- CSS custom properties for SubMiner's token colors
-
-That means the bundled page already knows which websocket to connect to and which color palette to use.
-
-### Build a custom websocket client
-
-Here is a minimal browser client for the annotation stream:
-
-```html
-<!doctype html>
-<meta charset="utf-8" />
-<title>SubMiner client</title>
-<div id="subtitle">Waiting for subtitles...</div>
-<script>
-  const subtitle = document.getElementById('subtitle');
-  const ws = new WebSocket('ws://127.0.0.1:6678');
-
-  ws.addEventListener('message', (event) => {
-    const payload = JSON.parse(event.data);
-    subtitle.innerHTML = payload.sentence || payload.text;
-  });
-
-  ws.addEventListener('close', () => {
-    subtitle.textContent = 'Connection closed; reload or reconnect.';
-  });
-</script>
-```
-
-### Build a custom Node client
-
-```js
-import WebSocket from 'ws';
-
-const ws = new WebSocket('ws://127.0.0.1:6678');
-
-ws.on('message', (raw) => {
-  const payload = JSON.parse(String(raw));
-  console.log({
-    text: payload.text,
-    tokens: payload.tokens.length,
-    firstToken: payload.tokens[0]?.surface ?? null,
-  });
-});
-```
-
-### Integration tips
-
-- Bind only to `127.0.0.1`; these services are local-only by design.
-- Handle empty `tokens` arrays gracefully because subtitle text can arrive before tokenization completes.
-- Reconnect on disconnect; SubMiner does not manage client reconnects for you.
-- Prefer `payload.text` for logging/automation and `payload.sentence` or `payload.tokens` for UI rendering.
-
-## Plugin Development
-
-SubMiner does **not** currently expose a general-purpose third-party plugin SDK inside the app itself. Today, the supported extension surfaces are:
-
-1. the local websocket streams
-2. the local texthooker UI
-3. the mpv Lua plugin's script-message API
-4. the launcher CLI
-
-### mpv script messages
-
-The mpv plugin accepts these script messages:
-
-```text
-script-message subminer-start
-script-message subminer-stop
-script-message subminer-toggle
-script-message subminer-menu
-script-message subminer-options
-script-message subminer-restart
-script-message subminer-status
-script-message subminer-autoplay-ready
-script-message subminer-aniskip-refresh
-script-message subminer-skip-intro
-```
-
-The start command also accepts inline overrides:
-
-```text
-script-message subminer-start backend=hyprland socket=/custom/path texthooker=no log-level=debug
-```
-
-### Practical extension patterns
-
-#### Add another mpv script that coordinates with SubMiner
-
-Examples:
-
-- send `subminer-start` after your own media-selection script chooses a file
-- send `subminer-status` before running follow-up automation
-- send `subminer-aniskip-refresh` after you update title/episode metadata
-
-#### Build a launcher wrapper
-
-Examples:
-
-- open a media picker, then call `subminer /path/to/file.mkv`
-- launch browser-only subtitle tooling with `subminer texthooker`
-- disable the helper UI for a session with `subminer --no-texthooker video.mkv`
-
-#### Build an overlay-adjacent client
-
-Examples:
-
-- browser widget showing current subtitle + token breakdown
-- local vocabulary capture helper that writes interesting lines to a file
-- bridge service that forwards websocket events into your own workflow engine
-
-## Webhook Examples
-
-SubMiner does **not** currently send outbound webhooks by itself. The supported pattern is to consume the websocket locally and relay events into another system.
-
-That still makes webhook-style automation straightforward.
-
-### Example: forward subtitle lines to a local webhook receiver
-
-```js
-import WebSocket from 'ws';
-
-const ws = new WebSocket('ws://127.0.0.1:6678');
-
-ws.on('message', async (raw) => {
-  const payload = JSON.parse(String(raw));
-
-  await fetch('http://127.0.0.1:5678/subminer/subtitle', {
-    method: 'POST',
-    headers: { 'content-type': 'application/json' },
-    body: JSON.stringify({
-      text: payload.text,
-      tokens: payload.tokens,
-      receivedAt: new Date().toISOString(),
-    }),
-  });
-});
-```
-
-### Automation ideas
-
-- **n8n / Make / Zapier relay:** send each subtitle line into an automation workflow for logging, translation, or summarization.
-- **Discord / Slack notifier:** post only lines that contain unknown words or N+1 targets.
-- **Obsidian / Markdown capture:** append subtitle lines plus token metadata to a daily immersion note.
-- **Local LLM pipeline:** trigger a glossary, translation, or sentence-mining workflow whenever a new line arrives.
-
-### Filtering example: only forward N+1 lines
-
-```js
-import WebSocket from 'ws';
-
-const ws = new WebSocket('ws://127.0.0.1:6678');
-
-ws.on('message', async (raw) => {
-  const payload = JSON.parse(String(raw));
-  const hasNPlusOne = payload.tokens.some((token) => token.isNPlusOneTarget);
-
-  if (!hasNPlusOne) return;
-
-  await fetch('http://127.0.0.1:5678/subminer/n-plus-one', {
-    method: 'POST',
-    headers: { 'content-type': 'application/json' },
-    body: JSON.stringify({ text: payload.text, tokens: payload.tokens }),
-  });
-});
-```
-
-## Recommended Integration Combinations
-
-- **Browser Yomitan client:** `texthooker` + `annotationWebsocket`
-- **Custom dashboard:** `annotationWebsocket` only
-- **Lightweight subtitle mirror:** `websocket` only
-- **mpv-side automation:** mpv plugin script messages + optional websocket relay
-- **Webhook-style workflows:** `annotationWebsocket` + your own local relay service
-
-## Related Pages
-
-- [Configuration](/configuration#websocket-server)
-- [Mining Workflow — Texthooker](/mining-workflow#texthooker)
-- [MPV Plugin](/mpv-plugin)
-- [Launcher Script](/launcher-script)
-- [Anki Integration](/anki-integration#proxy-mode-setup-yomitan--texthooker)

docs/RELEASING.md

@@ -8,7 +8,6 @@
 4. Bump `package.json` to the release version.
 5. Build release metadata before tagging:
    `bun run changelog:build --version <version> --date <yyyy-mm-dd>`
-   - Release CI now also auto-runs this step when releasing directly from a tag and `changes/*.md` fragments remain.
 6. Review `CHANGELOG.md` and `release/release-notes.md`.
 7. Run release gate locally:
    `bun run changelog:check --version <version>`
@@ -30,8 +29,6 @@ Notes:
 - Pass `--date` explicitly when you want the release stamped with the local cut date; otherwise the generator uses the current ISO date, which can roll over to the next UTC day late at night.
 - `changelog:check` now rejects tag/package version mismatches.
 - `changelog:build` generates `CHANGELOG.md` + `release/release-notes.md` and removes the released `changes/*.md` fragments.
-- In the same way, the release workflow now auto-runs `changelog:build` when it detects unreleased `changes/*.md` on a tag-based run, then verifies and publishes.
 - Do not tag while `changes/*.md` fragments still exist.
-- If you need to repair a published release body (for example, a prior version’s section was omitted), regenerate notes from `CHANGELOG.md` and re-edit the release with `gh release edit --notes-file`.
 - Tagged release workflow now also attempts to update `subminer-bin` on the AUR after GitHub Release publication.
 - Required GitHub Actions secret: `AUR_SSH_PRIVATE_KEY`. Add the matching public key to your AUR account before relying on the automation.

docs/architecture/domains.md

@@ -22,7 +22,6 @@ Read when: you need to find the owner module for a behavior or test surface
 - Subtitle/token pipeline: `src/core/services/tokenizer*`, `src/subtitle/`, `src/tokenizers/`
 - Anki workflow: `src/anki-integration/`, `src/core/services/anki-jimaku*.ts`
 - Immersion tracking: `src/core/services/immersion-tracker/`
-  Includes stats storage/query schema such as `imm_videos`, `imm_media_art`, and `imm_youtube_videos` for per-video and YouTube-specific library metadata.
 - AniList tracking: `src/core/services/anilist/`, `src/main/runtime/composers/anilist-*`
 - Jellyfin integration: `src/core/services/jellyfin*.ts`, `src/main/runtime/composers/jellyfin-*`
 - Window trackers: `src/window-trackers/`

docs/plans/2026-03-19-known-word-cache-incremental-sync-design.md

@@ -0,0 +1,46 @@
+<!-- read_when: changing known-word cache lifecycle, stats cache semantics, or Anki sync behavior -->
+
+# Incremental Known-Word Cache Sync
+
+## Goal
+
+Stop rebuilding the entire known-word cache on startup or routine refreshes. Keep the cache correct through incremental reconciliation on the configured sync cadence, with an immediate append path for freshly mined cards.
+
+## Scope
+
+- Persist per-note extracted known-word snapshots beside the existing global `words` list.
+- Replace startup refresh with load-only behavior.
+- Make timed refresh diff current Anki note IDs against cached note IDs, then apply add/remove/edit deltas.
+- Add `ankiConnect.knownWords.addMinedWordsImmediately`, default `true`.
+- Keep full rebuild out of normal lifecycle; reserve it for explicit doctor tooling.
+
+## Data Model
+
+Persist versioned cache state with:
+
+- `words`: deduplicated global known-word set for stats/UI consumers
+- `notes`: record of `noteId -> extractedWords[]`
+- `refreshedAtMs`
+- `scope`
+
+The in-memory manager derives the global set from the per-note snapshots during sync updates so deletes and edits can remove stale words safely.
+
+## Sync Behavior
+
+- Startup: load persisted state only
+- Interval tick or explicit refresh command: run incremental sync
+- Incremental sync:
+  - query tracked note IDs for configured deck scope
+  - remove note snapshots for note IDs that disappeared
+  - fetch `notesInfo` for note IDs that are new or need field reconciliation
+  - compare extracted words per note and update the global set
+
+## Immediate Mining Path
+
+When SubMiner already has fresh `noteInfo` after mining or updating a note, append/update that note snapshot immediately if `addMinedWordsImmediately` is enabled.
+
+## Verification
+
+- focused cache manager tests for add/delete/edit reconciliation
+- focused integration/config tests for startup behavior and new config flag
+- config verification lane because defaults/schema/example change

launcher/aniskip-metadata.ts

@@ -553,12 +553,10 @@ export function buildSubminerScriptOpts(
   socketPath: string,
   aniSkipMetadata: AniSkipMetadata | null,
   logLevel: LogLevel = 'info',
-  extraParts: string[] = [],
 ): string {
   const parts = [
     `subminer-binary_path=${sanitizeScriptOptValue(appPath)}`,
     `subminer-socket_path=${sanitizeScriptOptValue(socketPath)}`,
-    ...extraParts.map(sanitizeScriptOptValue),
   ];
   if (logLevel !== 'info') {
     parts.push(`subminer-log_level=${sanitizeScriptOptValue(logLevel)}`);

launcher/commands/command-modules.test.ts

@@ -48,64 +48,6 @@ function createContext(overrides: Partial<LauncherCommandContext> = {}): Launche
   };
 }
 
-type StatsTestArgOverrides = {
-  stats?: boolean;
-  statsBackground?: boolean;
-  statsCleanup?: boolean;
-  statsCleanupVocab?: boolean;
-  statsCleanupLifetime?: boolean;
-  statsStop?: boolean;
-  logLevel?: LauncherCommandContext['args']['logLevel'];
-};
-
-function createStatsTestHarness(overrides: StatsTestArgOverrides = {}) {
-  const context = createContext();
-  const forwarded: string[][] = [];
-  const removedPaths: string[] = [];
-  const createTempDir = (_prefix: string) => {
-    const created = `/tmp/subminer-stats-test`;
-    return created;
-  };
-  const joinPath = (...parts: string[]) => parts.join('/');
-  const removeDir = (targetPath: string) => {
-    removedPaths.push(targetPath);
-  };
-  const runAppCommandAttachedStub = async (
-    _appPath: string,
-    appArgs: string[],
-    _logLevel: LauncherCommandContext['args']['logLevel'],
-    _label: string,
-  ) => {
-    forwarded.push(appArgs);
-    return 0;
-  };
-  const waitForStatsResponseStub = async () => ({ ok: true, url: 'http://127.0.0.1:5175' });
-
-  context.args = {
-    ...context.args,
-    stats: true,
-    ...overrides,
-  };
-
-  return {
-    context,
-    forwarded,
-    removedPaths,
-    createTempDir,
-    joinPath,
-    removeDir,
-    runAppCommandAttachedStub,
-    waitForStatsResponseStub,
-    commandDeps: {
-      createTempDir,
-      joinPath,
-      runAppCommandAttached: runAppCommandAttachedStub,
-      waitForStatsResponse: waitForStatsResponseStub,
-      removeDir,
-    },
-  };
-}
-
 test('config command writes newline-terminated path via process adapter', () => {
   const writes: string[] = [];
   const context = createContext();
@@ -149,16 +91,20 @@ test('doctor command forwards refresh-known-words to app binary', () => {
   context.args.doctorRefreshKnownWords = true;
   const forwarded: string[][] = [];
 
-  const handled = runDoctorCommand(context, {
+  assert.throws(
+    () =>
+      runDoctorCommand(context, {
         commandExists: () => false,
         configExists: () => true,
         resolveMainConfigPath: () => '/tmp/SubMiner/config.jsonc',
         runAppCommandWithInherit: (_appPath, appArgs) => {
           forwarded.push(appArgs);
+          throw new ExitSignal(0);
         },
-  });
+      }),
+    (error: unknown) => error instanceof ExitSignal && error.code === 0,
+  );
 
-  assert.equal(handled, true);
   assert.deepEqual(forwarded, [['--refresh-known-words']]);
 });
 
@@ -183,33 +129,52 @@ test('dictionary command forwards --dictionary and target path to app binary', (
   context.args.dictionaryTarget = '/tmp/anime';
   const forwarded: string[][] = [];
 
-  const handled = runDictionaryCommand(context, {
+  assert.throws(
+    () =>
+      runDictionaryCommand(context, {
         runAppCommandWithInherit: (_appPath, appArgs) => {
           forwarded.push(appArgs);
+          throw new ExitSignal(0);
         },
-  });
+      }),
+    (error: unknown) => error instanceof ExitSignal && error.code === 0,
+  );
 
-  assert.equal(handled, true);
   assert.deepEqual(forwarded, [['--dictionary', '--dictionary-target', '/tmp/anime']]);
 });
 
-test('dictionary command returns after app handoff starts', () => {
+test('dictionary command throws if app handoff unexpectedly returns', () => {
   const context = createContext();
   context.args.dictionary = true;
 
-  const handled = runDictionaryCommand(context, {
-    runAppCommandWithInherit: () => undefined,
-  });
-
-  assert.equal(handled, true);
+  assert.throws(
+    () =>
+      runDictionaryCommand(context, {
+        runAppCommandWithInherit: () => undefined as never,
+      }),
+    /unexpectedly returned/,
+  );
 });
 
 test('stats command launches attached app command with response path', async () => {
-  const harness = createStatsTestHarness({ stats: true, logLevel: 'debug' });
-  const handled = await runStatsCommand(harness.context, harness.commandDeps);
+  const context = createContext();
+  context.args.stats = true;
+  context.args.logLevel = 'debug';
+  const forwarded: string[][] = [];
+
+  const handled = await runStatsCommand(context, {
+    createTempDir: () => '/tmp/subminer-stats-test',
+    joinPath: (...parts) => parts.join('/'),
+    runAppCommandAttached: async (_appPath, appArgs) => {
+      forwarded.push(appArgs);
+      return 0;
+    },
+    waitForStatsResponse: async () => ({ ok: true, url: 'http://127.0.0.1:5175' }),
+    removeDir: () => {},
+  });
 
   assert.equal(handled, true);
-  assert.deepEqual(harness.forwarded, [
+  assert.deepEqual(forwarded, [
     [
       '--stats',
       '--stats-response-path',
@@ -218,34 +183,50 @@ test('stats command launches attached app command with response path', async ()
       'debug',
     ],
   ]);
-  assert.equal(harness.removedPaths.length, 1);
 });
 
 test('stats background command launches attached daemon control command with response path', async () => {
-  const harness = createStatsTestHarness({ stats: true, statsBackground: true });
-  const handled = await runStatsCommand(harness.context, harness.commandDeps);
+  const context = createContext();
+  context.args.stats = true;
+  (context.args as typeof context.args & { statsBackground?: boolean }).statsBackground = true;
+  const forwarded: string[][] = [];
+
+  const handled = await runStatsCommand(context, {
+    createTempDir: () => '/tmp/subminer-stats-test',
+    joinPath: (...parts) => parts.join('/'),
+    runAppCommandAttached: async (_appPath, appArgs) => {
+      forwarded.push(appArgs);
+      return 0;
+    },
+    waitForStatsResponse: async () => ({ ok: true, url: 'http://127.0.0.1:5175' }),
+    removeDir: () => {},
+  } as Parameters<typeof runStatsCommand>[1]);
 
   assert.equal(handled, true);
-  assert.deepEqual(harness.forwarded, [
+  assert.deepEqual(forwarded, [
     [
       '--stats-daemon-start',
       '--stats-response-path',
       '/tmp/subminer-stats-test/response.json',
     ],
   ]);
-  assert.equal(harness.removedPaths.length, 1);
 });
 
 test('stats command waits for attached app exit after startup response', async () => {
-  const harness = createStatsTestHarness({ stats: true });
+  const context = createContext();
+  context.args.stats = true;
+  const forwarded: string[][] = [];
   const started = new Promise<number>((resolve) => setTimeout(() => resolve(0), 20));
 
-  const statsCommand = runStatsCommand(harness.context, {
-    ...harness.commandDeps,
-    runAppCommandAttached: async (...args) => {
-      await harness.runAppCommandAttachedStub(...args);
+  const statsCommand = runStatsCommand(context, {
+    createTempDir: () => '/tmp/subminer-stats-test',
+    joinPath: (...parts) => parts.join('/'),
+    runAppCommandAttached: async (_appPath, appArgs) => {
+      forwarded.push(appArgs);
       return started;
     },
+    waitForStatsResponse: async () => ({ ok: true, url: 'http://127.0.0.1:5175' }),
+    removeDir: () => {},
   });
   const result = await Promise.race([
     statsCommand.then(() => 'resolved'),
@@ -256,46 +237,53 @@ test('stats command waits for attached app exit after startup response', async (
 
   const final = await statsCommand;
   assert.equal(final, true);
-  assert.deepEqual(harness.forwarded, [
+  assert.deepEqual(forwarded, [
     [
       '--stats',
       '--stats-response-path',
       '/tmp/subminer-stats-test/response.json',
     ],
   ]);
-  assert.equal(harness.removedPaths.length, 1);
 });
 
 test('stats command throws when attached app exits non-zero after startup response', async () => {
-  const harness = createStatsTestHarness({ stats: true });
+  const context = createContext();
+  context.args.stats = true;
 
   await assert.rejects(async () => {
-    await runStatsCommand(harness.context, {
-      ...harness.commandDeps,
-      runAppCommandAttached: async (...args) => {
-        await harness.runAppCommandAttachedStub(...args);
+    await runStatsCommand(context, {
+      createTempDir: () => '/tmp/subminer-stats-test',
+      joinPath: (...parts) => parts.join('/'),
+      runAppCommandAttached: async () => {
         await new Promise((resolve) => setTimeout(resolve, 10));
         return 3;
       },
+      waitForStatsResponse: async () => ({ ok: true, url: 'http://127.0.0.1:5175' }),
+      removeDir: () => {},
     });
   }, /Stats app exited with status 3\./);
-
-  assert.equal(harness.removedPaths.length, 1);
 });
 
 test('stats cleanup command forwards cleanup vocab flags to the app', async () => {
-  const harness = createStatsTestHarness({
-    stats: true,
-    statsCleanup: true,
-    statsCleanupVocab: true,
-  });
-  const handled = await runStatsCommand(harness.context, {
-    ...harness.commandDeps,
+  const context = createContext();
+  context.args.stats = true;
+  context.args.statsCleanup = true;
+  context.args.statsCleanupVocab = true;
+  const forwarded: string[][] = [];
+
+  const handled = await runStatsCommand(context, {
+    createTempDir: () => '/tmp/subminer-stats-test',
+    joinPath: (...parts) => parts.join('/'),
+    runAppCommandAttached: async (_appPath, appArgs) => {
+      forwarded.push(appArgs);
+      return 0;
+    },
     waitForStatsResponse: async () => ({ ok: true }),
+    removeDir: () => {},
   });
 
   assert.equal(handled, true);
-  assert.deepEqual(harness.forwarded, [
+  assert.deepEqual(forwarded, [
     [
       '--stats',
       '--stats-response-path',
@@ -304,62 +292,76 @@ test('stats cleanup command forwards cleanup vocab flags to the app', async () =
       '--stats-cleanup-vocab',
     ],
   ]);
-  assert.equal(harness.removedPaths.length, 1);
 });
 
 test('stats stop command forwards stop flag to the app', async () => {
-  const harness = createStatsTestHarness({ stats: true, statsStop: true });
+  const context = createContext();
+  context.args.stats = true;
+  (context.args as typeof context.args & { statsStop?: boolean }).statsStop = true;
+  const forwarded: string[][] = [];
 
-  const handled = await runStatsCommand(harness.context, {
-    ...harness.commandDeps,
+  const handled = await runStatsCommand(context, {
+    createTempDir: () => '/tmp/subminer-stats-test',
+    joinPath: (...parts) => parts.join('/'),
+    runAppCommandAttached: async (_appPath, appArgs) => {
+      forwarded.push(appArgs);
+      return 0;
+    },
     waitForStatsResponse: async () => ({ ok: true }),
+    removeDir: () => {},
   });
 
   assert.equal(handled, true);
-  assert.deepEqual(harness.forwarded, [
+  assert.deepEqual(forwarded, [
     [
       '--stats-daemon-stop',
       '--stats-response-path',
       '/tmp/subminer-stats-test/response.json',
     ],
   ]);
-  assert.equal(harness.removedPaths.length, 1);
 });
 
 test('stats stop command exits on process exit without waiting for startup response', async () => {
-  const harness = createStatsTestHarness({ stats: true, statsStop: true });
+  const context = createContext();
+  context.args.stats = true;
+  (context.args as typeof context.args & { statsStop?: boolean }).statsStop = true;
   let waitedForResponse = false;
 
-  const handled = await runStatsCommand(harness.context, {
-    ...harness.commandDeps,
-    runAppCommandAttached: async (...args) => {
-      await harness.runAppCommandAttachedStub(...args);
-      return 0;
-    },
+  const handled = await runStatsCommand(context, {
+    createTempDir: () => '/tmp/subminer-stats-test',
+    joinPath: (...parts) => parts.join('/'),
+    runAppCommandAttached: async () => 0,
     waitForStatsResponse: async () => {
       waitedForResponse = true;
       return { ok: true };
     },
+    removeDir: () => {},
   });
 
   assert.equal(handled, true);
   assert.equal(waitedForResponse, false);
-  assert.equal(harness.removedPaths.length, 1);
 });
 
 test('stats cleanup command forwards lifetime rebuild flag to the app', async () => {
-  const harness = createStatsTestHarness({
-    stats: true,
-    statsCleanup: true,
-    statsCleanupLifetime: true,
-  });
-  const handled = await runStatsCommand(harness.context, {
-    ...harness.commandDeps,
+  const context = createContext();
+  context.args.stats = true;
+  context.args.statsCleanup = true;
+  context.args.statsCleanupLifetime = true;
+  const forwarded: string[][] = [];
+
+  const handled = await runStatsCommand(context, {
+    createTempDir: () => '/tmp/subminer-stats-test',
+    joinPath: (...parts) => parts.join('/'),
+    runAppCommandAttached: async (_appPath, appArgs) => {
+      forwarded.push(appArgs);
+      return 0;
+    },
     waitForStatsResponse: async () => ({ ok: true }),
+    removeDir: () => {},
   });
 
   assert.equal(handled, true);
-  assert.deepEqual(harness.forwarded, [
+  assert.deepEqual(forwarded, [
     [
       '--stats',
       '--stats-response-path',
@@ -368,64 +370,56 @@ test('stats cleanup command forwards lifetime rebuild flag to the app', async ()
       '--stats-cleanup-lifetime',
     ],
   ]);
-  assert.equal(harness.removedPaths.length, 1);
 });
 
 test('stats command throws when stats response reports an error', async () => {
-  const harness = createStatsTestHarness({ stats: true });
+  const context = createContext();
+  context.args.stats = true;
 
   await assert.rejects(async () => {
-    await runStatsCommand(harness.context, {
-      ...harness.commandDeps,
-      runAppCommandAttached: async (...args) => {
-        await harness.runAppCommandAttachedStub(...args);
-        return 0;
-      },
+    await runStatsCommand(context, {
+      createTempDir: () => '/tmp/subminer-stats-test',
+      joinPath: (...parts) => parts.join('/'),
+      runAppCommandAttached: async () => 0,
       waitForStatsResponse: async () => ({
         ok: false,
         error: 'Immersion tracking is disabled in config.',
       }),
+      removeDir: () => {},
     });
   }, /Immersion tracking is disabled in config\./);
-
-  assert.equal(harness.removedPaths.length, 1);
 });
 
 test('stats cleanup command fails if attached app exits before startup response', async () => {
-  const harness = createStatsTestHarness({
-    stats: true,
-    statsCleanup: true,
-    statsCleanupVocab: true,
-  });
+  const context = createContext();
+  context.args.stats = true;
+  context.args.statsCleanup = true;
+  context.args.statsCleanupVocab = true;
 
   await assert.rejects(async () => {
-    await runStatsCommand(harness.context, {
-      ...harness.commandDeps,
-      runAppCommandAttached: async (...args) => {
-        await harness.runAppCommandAttachedStub(...args);
-        return 2;
-      },
+    await runStatsCommand(context, {
+      createTempDir: () => '/tmp/subminer-stats-test',
+      joinPath: (...parts) => parts.join('/'),
+      runAppCommandAttached: async () => 2,
       waitForStatsResponse: async () => {
         await new Promise((resolve) => setTimeout(resolve, 25));
         return { ok: true, url: 'http://127.0.0.1:5175' };
       },
+      removeDir: () => {},
     });
   }, /Stats app exited before startup response \(status 2\)\./);
-
-  assert.equal(harness.removedPaths.length, 1);
 });
 
 test('stats command aborts pending response wait when app exits before startup response', async () => {
-  const harness = createStatsTestHarness({ stats: true });
+  const context = createContext();
+  context.args.stats = true;
   let aborted = false;
 
   await assert.rejects(async () => {
-    await runStatsCommand(harness.context, {
-      ...harness.commandDeps,
-      runAppCommandAttached: async (...args) => {
-        await harness.runAppCommandAttachedStub(...args);
-        return 2;
-      },
+    await runStatsCommand(context, {
+      createTempDir: () => '/tmp/subminer-stats-test',
+      joinPath: (...parts) => parts.join('/'),
+      runAppCommandAttached: async () => 2,
       waitForStatsResponse: async (_responsePath, signal) =>
         await new Promise((resolve) => {
           signal?.addEventListener(
@@ -437,24 +431,25 @@ test('stats command aborts pending response wait when app exits before startup r
             { once: true },
           );
         }),
+      removeDir: () => {},
     });
   }, /Stats app exited before startup response \(status 2\)\./);
 
   assert.equal(aborted, true);
-  assert.equal(harness.removedPaths.length, 1);
 });
 
 test('stats command aborts pending response wait when attached app fails to spawn', async () => {
-  const harness = createStatsTestHarness({ stats: true });
+  const context = createContext();
+  context.args.stats = true;
   const spawnError = new Error('spawn failed');
   let aborted = false;
 
   await assert.rejects(
     async () => {
-      await runStatsCommand(harness.context, {
-        ...harness.commandDeps,
-        runAppCommandAttached: async (...args) => {
-          await harness.runAppCommandAttachedStub(...args);
+      await runStatsCommand(context, {
+        createTempDir: () => '/tmp/subminer-stats-test',
+        joinPath: (...parts) => parts.join('/'),
+        runAppCommandAttached: async () => {
           throw spawnError;
         },
         waitForStatsResponse: async (_responsePath, signal) =>
@@ -468,30 +463,27 @@ test('stats command aborts pending response wait when attached app fails to spaw
               { once: true },
             );
           }),
+        removeDir: () => {},
       });
     },
     (error: unknown) => error === spawnError,
   );
 
   assert.equal(aborted, true);
-  assert.equal(harness.removedPaths.length, 1);
 });
 
 test('stats cleanup command aborts pending response wait when app exits before startup response', async () => {
-  const harness = createStatsTestHarness({
-    stats: true,
-    statsCleanup: true,
-    statsCleanupVocab: true,
-  });
+  const context = createContext();
+  context.args.stats = true;
+  context.args.statsCleanup = true;
+  context.args.statsCleanupVocab = true;
   let aborted = false;
 
   await assert.rejects(async () => {
-    await runStatsCommand(harness.context, {
-      ...harness.commandDeps,
-      runAppCommandAttached: async (...args) => {
-        await harness.runAppCommandAttachedStub(...args);
-        return 2;
-      },
+    await runStatsCommand(context, {
+      createTempDir: () => '/tmp/subminer-stats-test',
+      joinPath: (...parts) => parts.join('/'),
+      runAppCommandAttached: async () => 2,
       waitForStatsResponse: async (_responsePath, signal) =>
         await new Promise((resolve) => {
           signal?.addEventListener(
@@ -503,9 +495,9 @@ test('stats cleanup command aborts pending response wait when app exits before s
             { once: true },
           );
         }),
+      removeDir: () => {},
     });
   }, /Stats app exited before startup response \(status 2\)\./);
 
   assert.equal(aborted, true);
-  assert.equal(harness.removedPaths.length, 1);
 });

launcher/commands/dictionary-command.ts

@@ -2,7 +2,7 @@ import { runAppCommandWithInherit } from '../mpv.js';
 import type { LauncherCommandContext } from './context.js';
 
 interface DictionaryCommandDeps {
-  runAppCommandWithInherit: (appPath: string, appArgs: string[]) => void;
+  runAppCommandWithInherit: (appPath: string, appArgs: string[]) => never;
 }
 
 const defaultDeps: DictionaryCommandDeps = {
@@ -27,5 +27,5 @@ export function runDictionaryCommand(
   }
 
   deps.runAppCommandWithInherit(appPath, forwarded);
-  return true;
+  throw new Error('Dictionary command app handoff unexpectedly returned.');
 }

launcher/commands/doctor-command.ts

@@ -9,7 +9,7 @@ interface DoctorCommandDeps {
   commandExists(command: string): boolean;
   configExists(path: string): boolean;
   resolveMainConfigPath(): string;
-  runAppCommandWithInherit(appPath: string, appArgs: string[]): void;
+  runAppCommandWithInherit(appPath: string, appArgs: string[]): never;
 }
 
 const defaultDeps: DoctorCommandDeps = {
@@ -51,7 +51,7 @@ export function runDoctorCommand(
       ok: deps.commandExists('ffmpeg'),
       detail: deps.commandExists('ffmpeg')
         ? 'found'
-        : 'missing (optional unless legacy subtitle fallback is enabled)',
+        : 'missing (optional unless subtitle generation)',
     },
     {
       label: 'fzf',
@@ -85,7 +85,6 @@ export function runDoctorCommand(
       return true;
     }
     deps.runAppCommandWithInherit(appPath, ['--refresh-known-words']);
-    return true;
   }
 
   const hasHardFailure = checks.some((entry) =>

launcher/commands/jellyfin-command.ts

@@ -21,7 +21,6 @@ export async function runJellyfinCommand(context: LauncherCommandContext): Promi
     if (args.logLevel !== 'info') forwarded.push('--log-level', args.logLevel);
     appendPasswordStore(forwarded);
     runAppCommandWithInherit(appPath, forwarded);
-    return true;
   }
 
   if (args.jellyfinLogin) {
@@ -45,7 +44,6 @@ export async function runJellyfinCommand(context: LauncherCommandContext): Promi
     if (args.logLevel !== 'info') forwarded.push('--log-level', args.logLevel);
     appendPasswordStore(forwarded);
     runAppCommandWithInherit(appPath, forwarded);
-    return true;
   }
 
   if (args.jellyfinLogout) {
@@ -53,7 +51,6 @@ export async function runJellyfinCommand(context: LauncherCommandContext): Promi
     if (args.logLevel !== 'info') forwarded.push('--log-level', args.logLevel);
     appendPasswordStore(forwarded);
     runAppCommandWithInherit(appPath, forwarded);
-    return true;
   }
 
   if (args.jellyfinPlay) {
@@ -72,8 +69,13 @@ export async function runJellyfinCommand(context: LauncherCommandContext): Promi
     if (args.logLevel !== 'info') forwarded.push('--log-level', args.logLevel);
     appendPasswordStore(forwarded);
     runAppCommandWithInherit(appPath, forwarded);
-    return true;
   }
 
-  return false;
+  return Boolean(
+    args.jellyfin ||
+    args.jellyfinLogin ||
+    args.jellyfinLogout ||
+    args.jellyfinPlay ||
+    args.jellyfinDiscovery,
+  );
 }

launcher/commands/playback-command.test.ts

@@ -1,133 +0,0 @@
-import test from 'node:test';
-import assert from 'node:assert/strict';
-import type { LauncherCommandContext } from './context.js';
-import { runPlaybackCommandWithDeps } from './playback-command.js';
-
-function createContext(): LauncherCommandContext {
-  return {
-    args: {
-      backend: 'auto',
-      directory: '.',
-      recursive: false,
-      profile: '',
-      startOverlay: false,
-      youtubeMode: 'download',
-      whisperBin: '',
-      whisperModel: '',
-      whisperVadModel: '',
-      whisperThreads: 0,
-      youtubeSubgenOutDir: '',
-      youtubeSubgenAudioFormat: '',
-      youtubeSubgenKeepTemp: false,
-      youtubeFixWithAi: false,
-      youtubePrimarySubLangs: [],
-      youtubeSecondarySubLangs: [],
-      youtubeAudioLangs: [],
-      youtubeWhisperSourceLanguage: '',
-      aiConfig: {},
-      useTexthooker: false,
-      autoStartOverlay: false,
-      texthookerOnly: false,
-      useRofi: false,
-      logLevel: 'info',
-      passwordStore: '',
-      target: 'https://www.youtube.com/watch?v=65Ovd7t8sNw',
-      targetKind: 'url',
-      jimakuApiKey: '',
-      jimakuApiKeyCommand: '',
-      jimakuApiBaseUrl: '',
-      jimakuLanguagePreference: 'ja',
-      jimakuMaxEntryResults: 20,
-      jellyfin: false,
-      jellyfinLogin: false,
-      jellyfinLogout: false,
-      jellyfinPlay: false,
-      jellyfinDiscovery: false,
-      dictionary: false,
-      stats: false,
-      doctor: false,
-      doctorRefreshKnownWords: false,
-      configPath: false,
-      configShow: false,
-      mpvIdle: false,
-      mpvSocket: false,
-      mpvStatus: false,
-      mpvArgs: '',
-      appPassthrough: false,
-      appArgs: [],
-      jellyfinServer: '',
-      jellyfinUsername: '',
-      jellyfinPassword: '',
-    },
-    scriptPath: '/tmp/subminer',
-    scriptName: 'subminer',
-    mpvSocketPath: '/tmp/subminer.sock',
-    pluginRuntimeConfig: {
-      socketPath: '/tmp/subminer.sock',
-      autoStart: true,
-      autoStartVisibleOverlay: true,
-      autoStartPauseUntilReady: true,
-    },
-    appPath: '/tmp/SubMiner.AppImage',
-    launcherJellyfinConfig: {},
-    processAdapter: {
-      platform: () => 'linux',
-      onSignal: () => {},
-      writeStdout: () => {},
-      exit: (_code: number): never => {
-        throw new Error('unexpected exit');
-      },
-      setExitCode: () => {},
-    },
-  };
-}
-
-test('youtube playback launches overlay with app-owned youtube flow args', async () => {
-  const calls: string[] = [];
-  const context = createContext();
-  context.pluginRuntimeConfig = {
-    ...context.pluginRuntimeConfig,
-    autoStart: false,
-    autoStartVisibleOverlay: false,
-    autoStartPauseUntilReady: false,
-  };
-  let receivedStartMpvOptions: Record<string, unknown> | null = null;
-
-  await runPlaybackCommandWithDeps(context, {
-    ensurePlaybackSetupReady: async () => {},
-    chooseTarget: async (_args, _scriptPath) => ({ target: context.args.target, kind: 'url' }),
-    checkDependencies: () => {},
-    registerCleanup: () => {},
-    startMpv: async (
-      _target,
-      _targetKind,
-      _args,
-      _socketPath,
-      _appPath,
-      _preloadedSubtitles,
-      options,
-    ) => {
-      receivedStartMpvOptions = options ?? null;
-      calls.push('startMpv');
-    },
-    waitForUnixSocketReady: async () => true,
-    startOverlay: async (_appPath, _args, _socketPath, extraAppArgs = []) => {
-      calls.push(`startOverlay:${extraAppArgs.join(' ')}`);
-    },
-    launchAppCommandDetached: (_appPath: string, appArgs: string[]) => {
-      calls.push(`launch:${appArgs.join(' ')}`);
-    },
-    log: () => {},
-    cleanupPlaybackSession: async () => {},
-    getMpvProc: () => null,
-  });
-
-  assert.deepEqual(calls, [
-    'startMpv',
-    'startOverlay:--youtube-play https://www.youtube.com/watch?v=65Ovd7t8sNw --youtube-mode download',
-  ]);
-  assert.deepEqual(receivedStartMpvOptions, {
-    startPaused: true,
-    disableYoutubeSubtitleAutoLoad: true,
-  });
-});

launcher/commands/playback-command.ts

@@ -6,13 +6,13 @@ import { commandExists, isYoutubeTarget, realpathMaybe, resolvePathMaybe } from
 import { collectVideos, showFzfMenu, showRofiMenu } from '../picker.js';
 import {
   cleanupPlaybackSession,
-  launchAppCommandDetached,
   startMpv,
   startOverlay,
   state,
   stopOverlay,
   waitForUnixSocketReady,
 } from '../mpv.js';
+import { generateYoutubeSubtitles } from '../youtube.js';
 import type { Args } from '../types.js';
 import type { LauncherCommandContext } from './context.js';
 import { ensureLauncherSetupReady } from '../setup-gate.js';
@@ -31,12 +31,11 @@ function checkDependencies(args: Args): void {
 
   if (!commandExists('mpv')) missing.push('mpv');
 
-  const isYoutubeUrl = args.targetKind === 'url' && isYoutubeTarget(args.target);
-  if (args.targetKind === 'url' && !isYoutubeUrl && !commandExists('yt-dlp')) {
+  if (args.targetKind === 'url' && isYoutubeTarget(args.target) && !commandExists('yt-dlp')) {
     missing.push('yt-dlp');
   }
 
-  if (args.targetKind === 'url' && !isYoutubeUrl && !commandExists('ffmpeg')) {
+  if (args.targetKind === 'url' && isYoutubeTarget(args.target) && !commandExists('ffmpeg')) {
     missing.push('ffmpeg');
   }
 
@@ -127,66 +126,30 @@ async function ensurePlaybackSetupReady(context: LauncherCommandContext): Promis
 }
 
 export async function runPlaybackCommand(context: LauncherCommandContext): Promise<void> {
-  return runPlaybackCommandWithDeps(context, {
-    ensurePlaybackSetupReady,
-    chooseTarget,
-    checkDependencies,
-    registerCleanup,
-    startMpv,
-    waitForUnixSocketReady,
-    startOverlay,
-    launchAppCommandDetached,
-    log,
-    cleanupPlaybackSession,
-    getMpvProc: () => state.mpvProc,
-  });
-}
-
-type PlaybackCommandDeps = {
-  ensurePlaybackSetupReady: (context: LauncherCommandContext) => Promise<void>;
-  chooseTarget: (
-    args: Args,
-    scriptPath: string,
-  ) => Promise<{ target: string; kind: 'file' | 'url' } | null>;
-  checkDependencies: (args: Args) => void;
-  registerCleanup: (context: LauncherCommandContext) => void;
-  startMpv: typeof startMpv;
-  waitForUnixSocketReady: typeof waitForUnixSocketReady;
-  startOverlay: typeof startOverlay;
-  launchAppCommandDetached: typeof launchAppCommandDetached;
-  log: typeof log;
-  cleanupPlaybackSession: typeof cleanupPlaybackSession;
-  getMpvProc: () => typeof state.mpvProc;
-};
-
-export async function runPlaybackCommandWithDeps(
-  context: LauncherCommandContext,
-  deps: PlaybackCommandDeps,
-): Promise<void> {
   const { args, appPath, scriptPath, mpvSocketPath, pluginRuntimeConfig, processAdapter } = context;
   if (!appPath) {
     fail('SubMiner AppImage not found. Install to ~/.local/bin/ or set SUBMINER_APPIMAGE_PATH.');
   }
 
-  await deps.ensurePlaybackSetupReady(context);
+  await ensurePlaybackSetupReady(context);
 
   if (!args.target) {
     checkPickerDependencies(args);
   }
 
-  const targetChoice = await deps.chooseTarget(args, scriptPath);
+  const targetChoice = await chooseTarget(args, scriptPath);
   if (!targetChoice) {
-    deps.log('info', args.logLevel, 'No video selected, exiting');
+    log('info', args.logLevel, 'No video selected, exiting');
     processAdapter.exit(0);
   }
 
-  deps.checkDependencies({
+  checkDependencies({
     ...args,
     target: targetChoice ? targetChoice.target : args.target,
     targetKind: targetChoice ? targetChoice.kind : 'url',
   });
 
-  deps.registerCleanup(context);
+  registerCleanup(context);
 
   const selectedTarget = targetChoice
     ? {
@@ -196,11 +159,30 @@ export async function runPlaybackCommandWithDeps(
     : { target: args.target, kind: 'url' as const };
 
   const isYoutubeUrl = selectedTarget.kind === 'url' && isYoutubeTarget(selectedTarget.target);
-  const isAppOwnedYoutubeFlow = isYoutubeUrl;
-  const youtubeMode = args.youtubeMode ?? 'download';
+  let preloadedSubtitles: { primaryPath?: string; secondaryPath?: string } | undefined;
 
   if (isYoutubeUrl) {
-    deps.log('info', args.logLevel, 'YouTube subtitle flow: app-owned picker after mpv bootstrap');
+    log('info', args.logLevel, 'YouTube subtitle generation: preload before mpv');
+    const generated = await generateYoutubeSubtitles(selectedTarget.target, args);
+    preloadedSubtitles = {
+      primaryPath: generated.primaryPath,
+      secondaryPath: generated.secondaryPath,
+    };
+    const primaryStatus = generated.primaryPath
+      ? 'ready'
+      : generated.primaryNative
+        ? 'native'
+        : 'missing';
+    const secondaryStatus = generated.secondaryPath
+      ? 'ready'
+      : generated.secondaryNative
+        ? 'native'
+        : 'missing';
+    log(
+      'info',
+      args.logLevel,
+      `YouTube subtitle result: primary=${primaryStatus}, secondary=${secondaryStatus}`,
+    );
   }
 
   const shouldPauseUntilOverlayReady =
@@ -209,57 +191,47 @@ export async function runPlaybackCommandWithDeps(
     pluginRuntimeConfig.autoStartPauseUntilReady;
 
   if (shouldPauseUntilOverlayReady) {
-    deps.log('info', args.logLevel, 'Configured to pause mpv until overlay and tokenization are ready');
+    log('info', args.logLevel, 'Configured to pause mpv until overlay and tokenization are ready');
   }
 
-  await deps.startMpv(
+  await startMpv(
     selectedTarget.target,
     selectedTarget.kind,
     args,
     mpvSocketPath,
     appPath,
-    undefined,
-    {
-      startPaused: shouldPauseUntilOverlayReady || isAppOwnedYoutubeFlow,
-      disableYoutubeSubtitleAutoLoad: isAppOwnedYoutubeFlow,
-    },
+    preloadedSubtitles,
+    { startPaused: shouldPauseUntilOverlayReady },
   );
 
-  const ready = await deps.waitForUnixSocketReady(mpvSocketPath, 10000);
+  const ready = await waitForUnixSocketReady(mpvSocketPath, 10000);
   const pluginAutoStartEnabled = pluginRuntimeConfig.autoStart;
-  const shouldStartOverlay = args.startOverlay || args.autoStartOverlay || isAppOwnedYoutubeFlow;
+  const shouldStartOverlay = args.startOverlay || args.autoStartOverlay;
   if (shouldStartOverlay) {
     if (ready) {
-      deps.log('info', args.logLevel, 'MPV IPC socket ready, starting SubMiner overlay');
+      log('info', args.logLevel, 'MPV IPC socket ready, starting SubMiner overlay');
     } else {
-      deps.log(
+      log(
         'info',
         args.logLevel,
         'MPV IPC socket not ready after timeout, starting SubMiner overlay anyway',
       );
     }
-    await deps.startOverlay(
-      appPath,
-      args,
-      mpvSocketPath,
-      isAppOwnedYoutubeFlow
-        ? ['--youtube-play', selectedTarget.target, '--youtube-mode', youtubeMode]
-        : [],
-    );
+    await startOverlay(appPath, args, mpvSocketPath);
   } else if (pluginAutoStartEnabled) {
     if (ready) {
-      deps.log('info', args.logLevel, 'MPV IPC socket ready, relying on mpv plugin auto-start');
+      log('info', args.logLevel, 'MPV IPC socket ready, relying on mpv plugin auto-start');
     } else {
-      deps.log('info', args.logLevel, 'MPV IPC socket not ready yet, relying on mpv plugin auto-start');
+      log('info', args.logLevel, 'MPV IPC socket not ready yet, relying on mpv plugin auto-start');
     }
   } else if (ready) {
-    deps.log(
+    log(
       'info',
       args.logLevel,
       'MPV IPC socket ready, overlay auto-start disabled (use y-s to start)',
     );
   } else {
-    deps.log(
+    log(
       'info',
       args.logLevel,
       'MPV IPC socket not ready yet, overlay auto-start disabled (use y-s to start)',
@@ -267,7 +239,7 @@ export async function runPlaybackCommandWithDeps(
   }
 
   await new Promise<void>((resolve) => {
-    const mpvProc = deps.getMpvProc();
+    const mpvProc = state.mpvProc;
     if (!mpvProc) {
       stopOverlay(args);
       resolve();
@@ -275,7 +247,7 @@ export async function runPlaybackCommandWithDeps(
     }
 
     const finalize = (code: number | null | undefined) => {
-      void deps.cleanupPlaybackSession(args).finally(() => {
+      void cleanupPlaybackSession(args).finally(() => {
         processAdapter.setExitCode(code ?? 0);
         resolve();
       });

launcher/config.test.ts

@@ -10,6 +10,7 @@ test('launcher root help lists subcommands', () => {
 
   assert.match(output, /Commands:/);
   assert.match(output, /jellyfin\|jf/);
+  assert.match(output, /yt\|youtube/);
   assert.match(output, /doctor/);
   assert.match(output, /config/);
   assert.match(output, /mpv/);

launcher/config/args-normalizer.ts

@@ -249,6 +249,26 @@ export function applyInvocationsToArgs(parsed: Args, invocations: CliInvocations
     parsed.jellyfinLogout = Boolean(modeFlags.logout);
   }
 
+  if (invocations.ytInvocation) {
+    if (invocations.ytInvocation.logLevel)
+      parsed.logLevel = parseLogLevel(invocations.ytInvocation.logLevel);
+    if (invocations.ytInvocation.outDir)
+      parsed.youtubeSubgenOutDir = invocations.ytInvocation.outDir;
+    if (invocations.ytInvocation.keepTemp) parsed.youtubeSubgenKeepTemp = true;
+    if (invocations.ytInvocation.whisperBin)
+      parsed.whisperBin = invocations.ytInvocation.whisperBin;
+    if (invocations.ytInvocation.whisperModel)
+      parsed.whisperModel = invocations.ytInvocation.whisperModel;
+    if (invocations.ytInvocation.whisperVadModel)
+      parsed.whisperVadModel = invocations.ytInvocation.whisperVadModel;
+    if (invocations.ytInvocation.whisperThreads)
+      parsed.whisperThreads = invocations.ytInvocation.whisperThreads;
+    if (invocations.ytInvocation.ytSubgenAudioFormat) {
+      parsed.youtubeSubgenAudioFormat = invocations.ytInvocation.ytSubgenAudioFormat;
+    }
+    if (invocations.ytInvocation.target) ensureTarget(invocations.ytInvocation.target, parsed);
+  }
+
   if (invocations.dictionaryLogLevel) {
     parsed.logLevel = parseLogLevel(invocations.dictionaryLogLevel);
   }

launcher/config/cli-parser-builder.ts

@@ -14,6 +14,18 @@ export interface JellyfinInvocation {
   logLevel?: string;
 }
 
+export interface YtInvocation {
+  target?: string;
+  outDir?: string;
+  keepTemp?: boolean;
+  whisperBin?: string;
+  whisperModel?: string;
+  whisperVadModel?: string;
+  whisperThreads?: number;
+  ytSubgenAudioFormat?: string;
+  logLevel?: string;
+}
+
 export interface CommandActionInvocation {
   action: string;
   logLevel?: string;
@@ -21,6 +33,7 @@ export interface CommandActionInvocation {
 
 export interface CliInvocations {
   jellyfinInvocation: JellyfinInvocation | null;
+  ytInvocation: YtInvocation | null;
   configInvocation: CommandActionInvocation | null;
   mpvInvocation: CommandActionInvocation | null;
   appInvocation: { appArgs: string[] } | null;
@@ -76,6 +89,8 @@ function getTopLevelCommand(argv: string[]): { name: string; index: number } | n
   const commandNames = new Set([
     'jellyfin',
     'jf',
+    'yt',
+    'youtube',
     'doctor',
     'config',
     'mpv',
@@ -127,6 +142,7 @@ export function parseCliPrograms(
   invocations: CliInvocations;
 } {
   let jellyfinInvocation: JellyfinInvocation | null = null;
+  let ytInvocation: YtInvocation | null = null;
   let configInvocation: CommandActionInvocation | null = null;
   let mpvInvocation: CommandActionInvocation | null = null;
   let appInvocation: { appArgs: string[] } | null = null;
@@ -201,6 +217,38 @@ export function parseCliPrograms(
       };
     });
 
+  commandProgram
+    .command('yt')
+    .alias('youtube')
+    .description('YouTube workflows')
+    .argument('[target]', 'YouTube URL or ytsearch: query')
+    .option('-o, --out-dir <dir>', 'Subtitle output dir')
+    .option('--keep-temp', 'Keep temp files')
+    .option('--whisper-bin <path>', 'whisper.cpp CLI path')
+    .option('--whisper-model <path>', 'whisper model path')
+    .option('--whisper-vad-model <path>', 'whisper.cpp VAD model path')
+    .option('--whisper-threads <n>', 'whisper.cpp thread count')
+    .option('--yt-subgen-audio-format <format>', 'Audio extraction format')
+    .option('--log-level <level>', 'Log level')
+    .action((target: string | undefined, options: Record<string, unknown>) => {
+      ytInvocation = {
+        target,
+        outDir: typeof options.outDir === 'string' ? options.outDir : undefined,
+        keepTemp: options.keepTemp === true,
+        whisperBin: typeof options.whisperBin === 'string' ? options.whisperBin : undefined,
+        whisperModel: typeof options.whisperModel === 'string' ? options.whisperModel : undefined,
+        whisperVadModel:
+          typeof options.whisperVadModel === 'string' ? options.whisperVadModel : undefined,
+        whisperThreads:
+          typeof options.whisperThreads === 'number' && Number.isFinite(options.whisperThreads)
+            ? Math.floor(options.whisperThreads)
+            : undefined,
+        ytSubgenAudioFormat:
+          typeof options.ytSubgenAudioFormat === 'string' ? options.ytSubgenAudioFormat : undefined,
+        logLevel: typeof options.logLevel === 'string' ? options.logLevel : undefined,
+      };
+    });
+
   commandProgram
     .command('dictionary')
     .alias('dict')
@@ -334,6 +382,7 @@ export function parseCliPrograms(
     rootTarget: rootProgram.processedArgs[0],
     invocations: {
       jellyfinInvocation,
+      ytInvocation,
       configInvocation,
       mpvInvocation,
       appInvocation,

launcher/log.test.ts

@@ -1,10 +1,9 @@
 import test from 'node:test';
 import assert from 'node:assert/strict';
 import path from 'node:path';
-import { getDefaultLauncherLogFile, getDefaultMpvLogFile } from './types.js';
+import { getDefaultMpvLogFile } from './types.js';
 
 test('getDefaultMpvLogFile uses APPDATA on windows', () => {
-  const today = new Date().toISOString().slice(0, 10);
   const resolved = getDefaultMpvLogFile({
     platform: 'win32',
     homeDir: 'C:\\Users\\tester',
@@ -18,27 +17,8 @@ test('getDefaultMpvLogFile uses APPDATA on windows', () => {
         'C:\\Users\\tester\\AppData\\Roaming',
         'SubMiner',
         'logs',
-        `mpv-${today}.log`,
+        `SubMiner-${new Date().toISOString().slice(0, 10)}.log`,
       ),
     ),
   );
 });
-
-test('getDefaultLauncherLogFile uses launcher prefix', () => {
-  const today = new Date().toISOString().slice(0, 10);
-  const resolved = getDefaultLauncherLogFile({
-    platform: 'linux',
-    homeDir: '/home/tester',
-  });
-
-  assert.equal(
-    resolved,
-    path.join(
-      '/home/tester',
-      '.config',
-      'SubMiner',
-      'logs',
-      `launcher-${today}.log`,
-    ),
-  );
-});

launcher/log.ts

@@ -1,6 +1,7 @@
+import fs from 'node:fs';
+import path from 'node:path';
 import type { LogLevel } from './types.js';
-import { DEFAULT_MPV_LOG_FILE, getDefaultLauncherLogFile } from './types.js';
-import { appendLogLine, resolveDefaultLogFilePath } from '../src/shared/log-files.js';
+import { DEFAULT_MPV_LOG_FILE } from './types.js';
 
 export const COLORS = {
   red: '\x1b[0;31m',
@@ -27,32 +28,14 @@ export function getMpvLogPath(): string {
   return DEFAULT_MPV_LOG_FILE;
 }
 
-export function getLauncherLogPath(): string {
-  const envPath = process.env.SUBMINER_LAUNCHER_LOG?.trim();
-  if (envPath) return envPath;
-  return getDefaultLauncherLogFile();
-}
-
-export function getAppLogPath(): string {
-  const envPath = process.env.SUBMINER_APP_LOG?.trim();
-  if (envPath) return envPath;
-  return resolveDefaultLogFilePath('app');
-}
-
-function appendTimestampedLog(logPath: string, message: string): void {
-  appendLogLine(logPath, `[${new Date().toISOString()}] ${message}`);
-}
-
 export function appendToMpvLog(message: string): void {
-  appendTimestampedLog(getMpvLogPath(), message);
+  const logPath = getMpvLogPath();
+  try {
+    fs.mkdirSync(path.dirname(logPath), { recursive: true });
+    fs.appendFileSync(logPath, `[${new Date().toISOString()}] ${message}\n`, { encoding: 'utf8' });
+  } catch {
+    // ignore logging failures
   }
-
-export function appendToLauncherLog(message: string): void {
-  appendTimestampedLog(getLauncherLogPath(), message);
-}
-
-export function appendToAppLog(message: string): void {
-  appendTimestampedLog(getAppLogPath(), message);
 }
 
 export function log(level: LogLevel, configured: LogLevel, message: string): void {
@@ -66,11 +49,11 @@ export function log(level: LogLevel, configured: LogLevel, message: string): voi
           ? COLORS.red
           : COLORS.cyan;
   process.stdout.write(`${color}[${level.toUpperCase()}]${COLORS.reset} ${message}\n`);
-  appendToLauncherLog(`[${level.toUpperCase()}] ${message}`);
+  appendToMpvLog(`[${level.toUpperCase()}] ${message}`);
 }
 
 export function fail(message: string): never {
   process.stderr.write(`${COLORS.red}[ERROR]${COLORS.reset} ${message}\n`);
-  appendToLauncherLog(`[ERROR] ${message}`);
+  appendToMpvLog(`[ERROR] ${message}`);
   process.exit(1);
 }

launcher/main.test.ts

@@ -205,6 +205,136 @@ test('doctor refresh-known-words forwards app refresh command without requiring
   });
 });
 
+test('youtube command rejects removed --mode option', () => {
+  withTempDir((root) => {
+    const homeDir = path.join(root, 'home');
+    const xdgConfigHome = path.join(root, 'xdg');
+    const appPath = path.join(root, 'fake-subminer.sh');
+    fs.writeFileSync(appPath, '#!/bin/sh\nexit 0\n');
+    fs.chmodSync(appPath, 0o755);
+
+    const env = {
+      ...makeTestEnv(homeDir, xdgConfigHome),
+      SUBMINER_APPIMAGE_PATH: appPath,
+    };
+    const result = runLauncher(
+      ['youtube', 'https://www.youtube.com/watch?v=test123', '--mode', 'automatic'],
+      env,
+    );
+
+    assert.equal(result.status, 1);
+    assert.match(result.stderr, /unknown option '--mode'/i);
+  });
+});
+
+test('youtube playback generates subtitles before mpv launch', { timeout: 15000 }, () => {
+  withTempDir((root) => {
+    const homeDir = path.join(root, 'home');
+    const xdgConfigHome = path.join(root, 'xdg');
+    const binDir = path.join(root, 'bin');
+    const appPath = path.join(root, 'fake-subminer.sh');
+    const ytdlpLogPath = path.join(root, 'yt-dlp.log');
+    const mpvCapturePath = path.join(root, 'mpv-order.txt');
+    const mpvArgsPath = path.join(root, 'mpv-args.txt');
+    const socketPath = path.join(root, 'mpv.sock');
+    const bunBinary = JSON.stringify(process.execPath.replace(/\\/g, '/'));
+
+    fs.mkdirSync(binDir, { recursive: true });
+    fs.mkdirSync(path.join(xdgConfigHome, 'SubMiner'), { recursive: true });
+    fs.mkdirSync(path.join(xdgConfigHome, 'mpv', 'script-opts'), { recursive: true });
+    fs.writeFileSync(
+      path.join(xdgConfigHome, 'SubMiner', 'setup-state.json'),
+      JSON.stringify({
+        version: 1,
+        status: 'completed',
+        completedAt: '2026-03-08T00:00:00.000Z',
+        completionSource: 'user',
+        lastSeenYomitanDictionaryCount: 0,
+        pluginInstallStatus: 'installed',
+        pluginInstallPathSummary: null,
+      }),
+    );
+    fs.writeFileSync(
+      path.join(xdgConfigHome, 'mpv', 'script-opts', 'subminer.conf'),
+      `socket_path=${socketPath}\nauto_start=no\nauto_start_visible_overlay=no\nauto_start_pause_until_ready=no\n`,
+    );
+    fs.writeFileSync(appPath, '#!/bin/sh\nexit 0\n');
+    fs.chmodSync(appPath, 0o755);
+
+    fs.writeFileSync(
+      path.join(binDir, 'yt-dlp'),
+      `#!/bin/sh
+set -eu
+printf '%s\\n' "$*" >> "$SUBMINER_TEST_YTDLP_LOG"
+if printf '%s\\n' "$*" | grep -q -- '--dump-single-json'; then
+  printf '{"id":"video123"}\\n'
+  exit 0
+fi
+out_dir=""
+prev=""
+for arg in "$@"; do
+  if [ "$prev" = "-o" ]; then
+    out_dir=$(dirname "$arg")
+    break
+  fi
+  prev="$arg"
+done
+mkdir -p "$out_dir"
+printf '1\\n00:00:00,000 --> 00:00:01,000\\nこんにちは\\n' > "$out_dir/video123.ja.srt"
+printf '1\\n00:00:00,000 --> 00:00:01,000\\nhello\\n' > "$out_dir/video123.en.srt"
+`,
+      'utf8',
+    );
+    fs.chmodSync(path.join(binDir, 'yt-dlp'), 0o755);
+
+    fs.writeFileSync(path.join(binDir, 'ffmpeg'), '#!/bin/sh\nexit 0\n', 'utf8');
+    fs.chmodSync(path.join(binDir, 'ffmpeg'), 0o755);
+
+    fs.writeFileSync(
+      path.join(binDir, 'mpv'),
+      `#!/bin/sh
+set -eu
+if [ -s "$SUBMINER_TEST_YTDLP_LOG" ]; then
+  printf 'generated-before-mpv\\n' > "$SUBMINER_TEST_MPV_ORDER"
+else
+  printf 'mpv-before-generation\\n' > "$SUBMINER_TEST_MPV_ORDER"
+fi
+printf '%s\\n' "$@" > "$SUBMINER_TEST_MPV_ARGS"
+socket_path=""
+for arg in "$@"; do
+  case "$arg" in
+    --input-ipc-server=*)
+      socket_path="\${arg#--input-ipc-server=}"
+      ;;
+  esac
+done
+		${bunBinary} -e "const net=require('node:net'); const fs=require('node:fs'); const path=require('node:path'); const socket=process.argv[1]||''; try{ if(socket) fs.mkdirSync(path.dirname(socket),{recursive:true}); }catch{} try{ if(socket) fs.rmSync(socket,{force:true}); }catch{} const server=net.createServer((c)=>c.end()); server.on('error',()=>process.exit(0)); if(!socket) process.exit(0); try{ server.listen(socket,()=>setTimeout(()=>server.close(()=>process.exit(0)),250)); } catch { process.exit(0); }" "$socket_path"
+		`,
+      'utf8',
+    );
+    fs.chmodSync(path.join(binDir, 'mpv'), 0o755);
+
+    const env = {
+      ...makeTestEnv(homeDir, xdgConfigHome),
+      PATH: `${binDir}${path.delimiter}${process.env.Path || process.env.PATH || ''}`,
+      Path: `${binDir}${path.delimiter}${process.env.Path || process.env.PATH || ''}`,
+      SUBMINER_APPIMAGE_PATH: appPath,
+      SUBMINER_TEST_YTDLP_LOG: ytdlpLogPath,
+      SUBMINER_TEST_MPV_ORDER: mpvCapturePath,
+      SUBMINER_TEST_MPV_ARGS: mpvArgsPath,
+    };
+    const result = runLauncher(['youtube', 'https://www.youtube.com/watch?v=test123'], env);
+
+    assert.equal(result.status, 0, `stdout:\n${result.stdout}\nstderr:\n${result.stderr}`);
+    assert.equal(fs.readFileSync(mpvCapturePath, 'utf8').trim(), 'generated-before-mpv');
+    assert.match(
+      fs.readFileSync(mpvArgsPath, 'utf8'),
+      /https:\/\/www\.youtube\.com\/watch\?v=test123/,
+    );
+    assert.match(fs.readFileSync(ytdlpLogPath, 'utf8'), /--dump-single-json/);
+  });
+});
+
 test('launcher forwards --args to mpv as parsed tokens', { timeout: 15000 }, () => {
   withTempDir((root) => {
     const homeDir = path.join(root, 'home');
@@ -257,10 +387,6 @@ ${bunBinary} -e "const net=require('node:net'); const fs=require('node:fs'); con
       'utf8',
     );
     fs.chmodSync(path.join(binDir, 'mpv'), 0o755);
-    fs.writeFileSync(path.join(binDir, 'yt-dlp'), '#!/bin/sh\nexit 0\n', 'utf8');
-    fs.writeFileSync(path.join(binDir, 'ffmpeg'), '#!/bin/sh\nexit 0\n', 'utf8');
-    fs.chmodSync(path.join(binDir, 'yt-dlp'), 0o755);
-    fs.chmodSync(path.join(binDir, 'ffmpeg'), 0o755);
 
     const env = {
       ...makeTestEnv(homeDir, xdgConfigHome),
@@ -340,10 +466,6 @@ ${bunBinary} -e "const net=require('node:net'); const fs=require('node:fs'); con
       'utf8',
     );
     fs.chmodSync(path.join(binDir, 'mpv'), 0o755);
-    fs.writeFileSync(path.join(binDir, 'yt-dlp'), '#!/bin/sh\nexit 0\n', 'utf8');
-    fs.writeFileSync(path.join(binDir, 'ffmpeg'), '#!/bin/sh\nexit 0\n', 'utf8');
-    fs.chmodSync(path.join(binDir, 'yt-dlp'), 0o755);
-    fs.chmodSync(path.join(binDir, 'ffmpeg'), 0o755);
 
     const env = {
       ...makeTestEnv(homeDir, xdgConfigHome),
@@ -362,87 +484,6 @@ ${bunBinary} -e "const net=require('node:net'); const fs=require('node:fs'); con
   });
 });
 
-test('launcher routes youtube urls through regular playback startup', { timeout: 15000 }, () => {
-  withTempDir((root) => {
-    const homeDir = path.join(root, 'home');
-    const xdgConfigHome = path.join(root, 'xdg');
-    const binDir = path.join(root, 'bin');
-    const appPath = path.join(root, 'fake-subminer.sh');
-    const mpvArgsPath = path.join(root, 'mpv-args.txt');
-    const socketPath = path.join(root, 'mpv.sock');
-    const bunBinary = JSON.stringify(process.execPath.replace(/\\/g, '/'));
-
-    fs.mkdirSync(binDir, { recursive: true });
-    fs.mkdirSync(path.join(xdgConfigHome, 'SubMiner'), { recursive: true });
-    fs.mkdirSync(path.join(xdgConfigHome, 'mpv', 'script-opts'), { recursive: true });
-    fs.writeFileSync(
-      path.join(xdgConfigHome, 'SubMiner', 'setup-state.json'),
-      JSON.stringify({
-        version: 1,
-        status: 'completed',
-        completedAt: '2026-03-08T00:00:00.000Z',
-        completionSource: 'user',
-        lastSeenYomitanDictionaryCount: 0,
-        pluginInstallStatus: 'installed',
-        pluginInstallPathSummary: null,
-      }),
-    );
-    fs.writeFileSync(
-      path.join(xdgConfigHome, 'mpv', 'script-opts', 'subminer.conf'),
-      `socket_path=${socketPath}\nauto_start=yes\nauto_start_visible_overlay=yes\nauto_start_pause_until_ready=yes\n`,
-    );
-    fs.writeFileSync(
-      appPath,
-      '#!/bin/sh\nif [ -n "$SUBMINER_TEST_CAPTURE" ]; then printf "%s\\n" "$@" > "$SUBMINER_TEST_CAPTURE"; fi\nexit 0\n',
-    );
-    fs.chmodSync(appPath, 0o755);
-
-    fs.writeFileSync(
-      path.join(binDir, 'mpv'),
-      `#!/bin/sh
-set -eu
-printf '%s\\n' "$@" > "$SUBMINER_TEST_MPV_ARGS"
-socket_path=""
-for arg in "$@"; do
-  case "$arg" in
-    --input-ipc-server=*)
-      socket_path="\${arg#--input-ipc-server=}"
-      ;;
-  esac
-done
-${bunBinary} -e "const net=require('node:net'); const fs=require('node:fs'); const path=require('node:path'); const socket=process.argv[1]||''; try{ if (socket) fs.mkdirSync(path.dirname(socket),{recursive:true}); }catch{} try{ if (socket) fs.rmSync(socket,{force:true}); }catch{} if(!socket) process.exit(0); const server=net.createServer((c)=>c.end()); server.on('error',()=>process.exit(0)); try{ server.listen(socket,()=>setTimeout(()=>server.close(()=>process.exit(0)),250)); } catch { process.exit(0); }" "$socket_path"
-`,
-      'utf8',
-    );
-    fs.chmodSync(path.join(binDir, 'mpv'), 0o755);
-    fs.writeFileSync(path.join(binDir, 'yt-dlp'), '#!/bin/sh\nexit 0\n', 'utf8');
-    fs.writeFileSync(path.join(binDir, 'ffmpeg'), '#!/bin/sh\nexit 0\n', 'utf8');
-    fs.chmodSync(path.join(binDir, 'yt-dlp'), 0o755);
-    fs.chmodSync(path.join(binDir, 'ffmpeg'), 0o755);
-
-    const env = {
-      ...makeTestEnv(homeDir, xdgConfigHome),
-      PATH: `${binDir}${path.delimiter}${process.env.Path || process.env.PATH || ''}`,
-      Path: `${binDir}${path.delimiter}${process.env.Path || process.env.PATH || ''}`,
-      DISPLAY: ':99',
-      XDG_SESSION_TYPE: 'x11',
-      SUBMINER_APPIMAGE_PATH: appPath,
-      SUBMINER_TEST_MPV_ARGS: mpvArgsPath,
-      SUBMINER_TEST_CAPTURE: path.join(root, 'captured-args.txt'),
-    };
-    const result = runLauncher(['https://www.youtube.com/watch?v=abc123'], env);
-
-    assert.equal(result.status, 0, `stdout:\n${result.stdout}\nstderr:\n${result.stderr}`);
-    const forwardedArgs = fs
-      .readFileSync(mpvArgsPath, 'utf8')
-      .trim()
-      .split('\n')
-      .map((item) => item.trim())
-      .filter(Boolean);
-    assert.equal(forwardedArgs.includes('https://www.youtube.com/watch?v=abc123'), true);
-  });
-});
-
 test('dictionary command forwards --dictionary and --dictionary-target to app command path', () => {
   withTempDir((root) => {
     const homeDir = path.join(root, 'home');

launcher/mpv.test.ts

@@ -302,47 +302,7 @@ test('startOverlay resolves without fixed 2s sleep when readiness signals arrive
   }
 });
 
-test('startOverlay captures app stdout and stderr into app log', async () => {
-  const { dir, socketPath } = createTempSocketPath();
-  const appPath = path.join(dir, 'fake-subminer.sh');
-  const appLogPath = path.join(dir, 'app.log');
-  const originalAppLog = process.env.SUBMINER_APP_LOG;
-  fs.writeFileSync(
-    appPath,
-    '#!/bin/sh\nprintf "hello from stdout\\n"\nprintf "hello from stderr\\n" >&2\nexit 0\n',
-  );
-  fs.chmodSync(appPath, 0o755);
-  fs.writeFileSync(socketPath, '');
-  const originalCreateConnection = net.createConnection;
-  try {
-    process.env.SUBMINER_APP_LOG = appLogPath;
-    net.createConnection = (() => {
-      const socket = new EventEmitter() as net.Socket;
-      socket.destroy = (() => socket) as net.Socket['destroy'];
-      socket.setTimeout = (() => socket) as net.Socket['setTimeout'];
-      setTimeout(() => socket.emit('connect'), 10);
-      return socket;
-    }) as typeof net.createConnection;
-
-    await startOverlay(appPath, makeArgs(), socketPath);
-
-    const logText = fs.readFileSync(appLogPath, 'utf8');
-    assert.match(logText, /\[STDOUT\] hello from stdout/);
-    assert.match(logText, /\[STDERR\] hello from stderr/);
-  } finally {
-    net.createConnection = originalCreateConnection;
-    state.overlayProc = null;
-    state.overlayManagedByLauncher = false;
-    if (originalAppLog === undefined) {
-      delete process.env.SUBMINER_APP_LOG;
-    } else {
-      process.env.SUBMINER_APP_LOG = originalAppLog;
-    }
-    fs.rmSync(dir, { recursive: true, force: true });
-  }
-});
-
-test('cleanupPlaybackSession stops launcher-managed overlay app and mpv-owned children', async () => {
+test('cleanupPlaybackSession preserves background app while stopping mpv-owned children', async () => {
   const { dir } = createTempSocketPath();
   const appPath = path.join(dir, 'fake-subminer.sh');
   const appInvocationsPath = path.join(dir, 'app-invocations.log');
@@ -385,8 +345,8 @@ test('cleanupPlaybackSession stops launcher-managed overlay app and mpv-owned ch
   try {
     await cleanupPlaybackSession(makeArgs());
 
-    assert.deepEqual(calls, ['overlay-kill', 'mpv-kill', 'helper-kill']);
-    assert.match(fs.readFileSync(appInvocationsPath, 'utf8'), /--stop/);
+    assert.deepEqual(calls, ['mpv-kill', 'helper-kill']);
+    assert.equal(fs.existsSync(appInvocationsPath), false);
   } finally {
     state.overlayProc = null;
     state.mpvProc = null;

launcher/mpv.ts

@@ -5,7 +5,7 @@ import net from 'node:net';
 import { spawn, spawnSync } from 'node:child_process';
 import type { LogLevel, Backend, Args, MpvTrack } from './types.js';
 import { DEFAULT_MPV_SUBMINER_ARGS, DEFAULT_YOUTUBE_YTDL_FORMAT } from './types.js';
-import { appendToAppLog, getAppLogPath, log, fail, getMpvLogPath } from './log.js';
+import { log, fail, getMpvLogPath } from './log.js';
 import { buildSubminerScriptOpts, resolveAniSkipMetadataForFile } from './aniskip-metadata.js';
 import {
   commandExists,
@@ -542,7 +542,7 @@ export async function startMpv(
   socketPath: string,
   appPath: string,
   preloadedSubtitles?: { primaryPath?: string; secondaryPath?: string },
-  options?: { startPaused?: boolean; disableYoutubeSubtitleAutoLoad?: boolean },
+  options?: { startPaused?: boolean },
 ): Promise<void> {
   if (targetKind === 'file' && (!fs.existsSync(target) || !fs.statSync(target).isFile())) {
     fail(`Video file not found: ${target}`);
@@ -557,9 +557,15 @@ export async function startMpv(
   const mpvArgs: string[] = [];
   if (args.profile) mpvArgs.push(`--profile=${args.profile}`);
   mpvArgs.push(...DEFAULT_MPV_SUBMINER_ARGS);
+  if (args.mpvArgs) {
+    mpvArgs.push(...parseMpvArgString(args.mpvArgs));
+  }
+
   if (targetKind === 'url' && isYoutubeTarget(target)) {
     log('info', args.logLevel, 'Applying URL playback options');
-    mpvArgs.push('--ytdl=yes');
+    mpvArgs.push('--ytdl=yes', '--ytdl-raw-options=');
+
+    if (isYoutubeTarget(target)) {
       const subtitleLangs = uniqueNormalizedLangCodes([
         ...args.youtubePrimarySubLangs,
         ...args.youtubeSecondarySubLangs,
@@ -569,7 +575,6 @@ export async function startMpv(
       log('debug', args.logLevel, `YouTube subtitle langs: ${subtitleLangs}`);
       log('debug', args.logLevel, `YouTube audio langs: ${audioLangs}`);
       mpvArgs.push(`--ytdl-format=${DEFAULT_YOUTUBE_YTDL_FORMAT}`, `--alang=${audioLangs}`);
-    if (options?.disableYoutubeSubtitleAutoLoad !== true) {
       mpvArgs.push(
         '--sub-auto=fuzzy',
         `--slang=${subtitleLangs}`,
@@ -577,13 +582,8 @@ export async function startMpv(
         '--ytdl-raw-options-append=sub-format=vtt/best',
         `--ytdl-raw-options-append=sub-langs=${subtitleLangs}`,
       );
-    } else {
-      mpvArgs.push('--sub-auto=no');
     }
   }
-  if (args.mpvArgs) {
-    mpvArgs.push(...parseMpvArgString(args.mpvArgs));
-  }
 
   if (preloadedSubtitles?.primaryPath) {
     mpvArgs.push(`--sub-file=${preloadedSubtitles.primaryPath}`);
@@ -597,17 +597,7 @@ export async function startMpv(
   const aniSkipMetadata = shouldResolveAniSkipMetadata(target, targetKind, preloadedSubtitles)
     ? await resolveAniSkipMetadataForFile(target)
     : null;
-  const extraScriptOpts =
-    targetKind === 'url' && isYoutubeTarget(target) && options?.disableYoutubeSubtitleAutoLoad === true
-      ? ['subminer-auto_start_pause_until_ready=no']
-      : [];
-  const scriptOpts = buildSubminerScriptOpts(
-    appPath,
-    socketPath,
-    aniSkipMetadata,
-    args.logLevel,
-    extraScriptOpts,
-  );
+  const scriptOpts = buildSubminerScriptOpts(appPath, socketPath, aniSkipMetadata, args.logLevel);
   if (aniSkipMetadata) {
     log(
       'debug',
@@ -671,25 +661,19 @@ async function waitForOverlayStartCommandSettled(
   });
 }
 
-export async function startOverlay(
-  appPath: string,
-  args: Args,
-  socketPath: string,
-  extraAppArgs: string[] = [],
-): Promise<void> {
+export async function startOverlay(appPath: string, args: Args, socketPath: string): Promise<void> {
   const backend = detectBackend(args.backend);
   log('info', args.logLevel, `Starting SubMiner overlay (backend: ${backend})...`);
 
-  const overlayArgs = ['--start', '--backend', backend, '--socket', socketPath, ...extraAppArgs];
+  const overlayArgs = ['--start', '--backend', backend, '--socket', socketPath];
   if (args.logLevel !== 'info') overlayArgs.push('--log-level', args.logLevel);
   if (args.useTexthooker) overlayArgs.push('--texthooker');
 
   const target = resolveAppSpawnTarget(appPath, overlayArgs);
   state.overlayProc = spawn(target.command, target.args, {
-    stdio: ['ignore', 'pipe', 'pipe'],
+    stdio: 'inherit',
     env: buildAppEnv(),
   });
-  attachAppProcessLogging(state.overlayProc);
   state.overlayManagedByLauncher = true;
 
   const [socketReady] = await Promise.all([
@@ -715,7 +699,10 @@ export function launchTexthookerOnly(appPath: string, args: Args): never {
   if (args.logLevel !== 'info') overlayArgs.push('--log-level', args.logLevel);
 
   log('info', args.logLevel, 'Launching texthooker mode...');
-  const result = runSyncAppCommand(appPath, overlayArgs, true);
+  const result = spawnSync(appPath, overlayArgs, {
+    stdio: 'inherit',
+    env: buildAppEnv(),
+  });
   if (result.error) {
     fail(`Failed to launch texthooker mode: ${result.error.message}`);
   }
@@ -726,7 +713,30 @@ export function stopOverlay(args: Args): void {
   if (state.stopRequested) return;
   state.stopRequested = true;
 
-  stopManagedOverlayApp(args);
+  if (state.overlayManagedByLauncher && state.appPath) {
+    log('info', args.logLevel, 'Stopping SubMiner overlay...');
+
+    const stopArgs = ['--stop'];
+    if (args.logLevel !== 'info') stopArgs.push('--log-level', args.logLevel);
+
+    const result = spawnSync(state.appPath, stopArgs, {
+      stdio: 'ignore',
+      env: buildAppEnv(),
+    });
+    if (result.error) {
+      log('warn', args.logLevel, `Failed to stop SubMiner overlay: ${result.error.message}`);
+    } else if (typeof result.status === 'number' && result.status !== 0) {
+      log('warn', args.logLevel, `SubMiner overlay stop command exited with status ${result.status}`);
+    }
+
+    if (state.overlayProc && !state.overlayProc.killed) {
+      try {
+        state.overlayProc.kill('SIGTERM');
+      } catch {
+        // ignore
+      }
+    }
+  }
 
   if (state.mpvProc && !state.mpvProc.killed) {
     try {
@@ -751,8 +761,6 @@ export function stopOverlay(args: Args): void {
 }
 
 export async function cleanupPlaybackSession(args: Args): Promise<void> {
-  stopManagedOverlayApp(args);
-
   if (state.mpvProc && !state.mpvProc.killed) {
     try {
       state.mpvProc.kill('SIGTERM');
@@ -775,40 +783,9 @@ export async function cleanupPlaybackSession(args: Args): Promise<void> {
   await terminateTrackedDetachedMpv(args.logLevel);
 }
 
-function stopManagedOverlayApp(args: Args): void {
-  if (!(state.overlayManagedByLauncher && state.appPath)) {
-    return;
-  }
-
-  log('info', args.logLevel, 'Stopping SubMiner overlay...');
-
-  const stopArgs = ['--stop'];
-  if (args.logLevel !== 'info') stopArgs.push('--log-level', args.logLevel);
-
-  const target = resolveAppSpawnTarget(state.appPath, stopArgs);
-  const result = spawnSync(target.command, target.args, {
-    stdio: 'ignore',
-    env: buildAppEnv(),
-  });
-  if (result.error) {
-    log('warn', args.logLevel, `Failed to stop SubMiner overlay: ${result.error.message}`);
-  } else if (typeof result.status === 'number' && result.status !== 0) {
-    log('warn', args.logLevel, `SubMiner overlay stop command exited with status ${result.status}`);
-  }
-
-  if (state.overlayProc && !state.overlayProc.killed) {
-    try {
-      state.overlayProc.kill('SIGTERM');
-    } catch {
-      // ignore
-    }
-  }
-}
-
 function buildAppEnv(): NodeJS.ProcessEnv {
   const env: Record<string, string | undefined> = {
     ...process.env,
-    SUBMINER_APP_LOG: getAppLogPath(),
     SUBMINER_MPV_LOG: getMpvLogPath(),
   };
   delete env.ELECTRON_RUN_AS_NODE;
@@ -827,64 +804,6 @@ function buildAppEnv(): NodeJS.ProcessEnv {
   return env;
 }
 
-function appendCapturedAppOutput(kind: 'STDOUT' | 'STDERR', chunk: string): void {
-  const normalized = chunk.replace(/\r\n/g, '\n');
-  for (const line of normalized.split('\n')) {
-    if (!line) continue;
-    appendToAppLog(`[${kind}] ${line}`);
-  }
-}
-
-function attachAppProcessLogging(
-  proc: ReturnType<typeof spawn>,
-  options?: {
-    mirrorStdout?: boolean;
-    mirrorStderr?: boolean;
-  },
-): void {
-  proc.stdout?.setEncoding('utf8');
-  proc.stderr?.setEncoding('utf8');
-  proc.stdout?.on('data', (chunk: string) => {
-    appendCapturedAppOutput('STDOUT', chunk);
-    if (options?.mirrorStdout) process.stdout.write(chunk);
-  });
-  proc.stderr?.on('data', (chunk: string) => {
-    appendCapturedAppOutput('STDERR', chunk);
-    if (options?.mirrorStderr) process.stderr.write(chunk);
-  });
-}
-
-function runSyncAppCommand(
-  appPath: string,
-  appArgs: string[],
-  mirrorOutput: boolean,
-): {
-  status: number;
-  stdout: string;
-  stderr: string;
-  error?: Error;
-} {
-  const target = resolveAppSpawnTarget(appPath, appArgs);
-  const result = spawnSync(target.command, target.args, {
-    env: buildAppEnv(),
-    encoding: 'utf8',
-  });
-  if (result.stdout) {
-    appendCapturedAppOutput('STDOUT', result.stdout);
-    if (mirrorOutput) process.stdout.write(result.stdout);
-  }
-  if (result.stderr) {
-    appendCapturedAppOutput('STDERR', result.stderr);
-    if (mirrorOutput) process.stderr.write(result.stderr);
-  }
-  return {
-    status: result.status ?? 1,
-    stdout: result.stdout ?? '',
-    stderr: result.stderr ?? '',
-    error: result.error ?? undefined,
-  };
-}
-
 function maybeCaptureAppArgs(appArgs: string[]): boolean {
   const capturePath = process.env.SUBMINER_TEST_CAPTURE?.trim();
   if (!capturePath) {
@@ -902,23 +821,20 @@ function resolveAppSpawnTarget(appPath: string, appArgs: string[]): SpawnTarget
   return resolveCommandInvocation(appPath, appArgs);
 }
 
-export function runAppCommandWithInherit(appPath: string, appArgs: string[]): void {
+export function runAppCommandWithInherit(appPath: string, appArgs: string[]): never {
   if (maybeCaptureAppArgs(appArgs)) {
     process.exit(0);
   }
 
   const target = resolveAppSpawnTarget(appPath, appArgs);
-  const proc = spawn(target.command, target.args, {
-    stdio: ['ignore', 'pipe', 'pipe'],
+  const result = spawnSync(target.command, target.args, {
+    stdio: 'inherit',
     env: buildAppEnv(),
   });
-  attachAppProcessLogging(proc, { mirrorStdout: true, mirrorStderr: true });
-  proc.once('error', (error) => {
-    fail(`Failed to run app command: ${error.message}`);
-  });
-  proc.once('close', (code) => {
-    process.exit(code ?? 0);
-  });
+  if (result.error) {
+    fail(`Failed to run app command: ${result.error.message}`);
+  }
+  process.exit(result.status ?? 0);
 }
 
 export function runAppCommandCaptureOutput(
@@ -938,7 +854,18 @@ export function runAppCommandCaptureOutput(
     };
   }
 
-  return runSyncAppCommand(appPath, appArgs, false);
+  const target = resolveAppSpawnTarget(appPath, appArgs);
+  const result = spawnSync(target.command, target.args, {
+    env: buildAppEnv(),
+    encoding: 'utf8',
+  });
+
+  return {
+    status: result.status ?? 1,
+    stdout: result.stdout ?? '',
+    stderr: result.stderr ?? '',
+    error: result.error ?? undefined,
+  };
 }
 
 export function runAppCommandAttached(
@@ -960,14 +887,13 @@ export function runAppCommandAttached(
 
   return new Promise((resolve, reject) => {
     const proc = spawn(target.command, target.args, {
-      stdio: ['ignore', 'pipe', 'pipe'],
+      stdio: 'inherit',
       env: buildAppEnv(),
     });
-    attachAppProcessLogging(proc, { mirrorStdout: true, mirrorStderr: true });
     proc.once('error', (error) => {
       reject(error);
     });
-    proc.once('close', (code, signal) => {
+    proc.once('exit', (code, signal) => {
       if (code !== null) {
         resolve(code);
       } else if (signal) {
@@ -995,7 +921,10 @@ export function runAppCommandWithInheritLogged(
     logLevel,
     `${label}: launching app with args: ${[target.command, ...target.args].join(' ')}`,
   );
-  const result = runSyncAppCommand(appPath, appArgs, true);
+  const result = spawnSync(target.command, target.args, {
+    stdio: 'inherit',
+    env: buildAppEnv(),
+  });
   if (result.error) {
     fail(`Failed to run app command: ${result.error.message}`);
   }
@@ -1024,13 +953,8 @@ export function launchAppCommandDetached(
     logLevel,
     `${label}: launching detached app with args: ${[target.command, ...target.args].join(' ')}`,
   );
-  const appLogPath = getAppLogPath();
-  fs.mkdirSync(path.dirname(appLogPath), { recursive: true });
-  const stdoutFd = fs.openSync(appLogPath, 'a');
-  const stderrFd = fs.openSync(appLogPath, 'a');
-  try {
   const proc = spawn(target.command, target.args, {
-      stdio: ['ignore', stdoutFd, stderrFd],
+    stdio: 'ignore',
     detached: true,
     env: buildAppEnv(),
   });
@@ -1038,10 +962,6 @@ export function launchAppCommandDetached(
     log('warn', logLevel, `${label}: failed to launch detached app: ${error.message}`);
   });
   proc.unref();
-  } finally {
-    fs.closeSync(stdoutFd);
-    fs.closeSync(stderrFd);
-  }
 }
 
 export function launchMpvIdleDetached(

launcher/smoke.e2e.test.ts

@@ -310,7 +310,6 @@ test(
       const appStartPath = path.join(smokeCase.artifactsDir, 'fake-app-start.log');
       const appStopPath = path.join(smokeCase.artifactsDir, 'fake-app-stop.log');
       await waitForJsonLines(appStartPath, 1);
-      await waitForJsonLines(appStopPath, 1);
 
       const appStartEntries = readJsonLines(appStartPath);
       const appStopEntries = readJsonLines(appStopPath);
@@ -325,7 +324,7 @@ test(
       assert.match(result.stdout, /Starting SubMiner overlay/i);
 
       assert.equal(appStartEntries.length, 1);
-      assert.equal(appStopEntries.length, 1);
+      assert.equal(appStopEntries.length, 0);
       assert.equal(mpvEntries.length >= 1, true);
 
       const appStartArgs = appStartEntries[0]?.argv;

launcher/types.ts

@@ -1,6 +1,5 @@
 import path from 'node:path';
 import os from 'node:os';
-import { resolveDefaultLogFilePath } from '../src/shared/log-files.js';
 export { VIDEO_EXTENSIONS } from '../src/shared/video-extensions.js';
 
 export const ROFI_THEME_FILE = 'subminer.rasi';
@@ -30,28 +29,21 @@ export const DEFAULT_YOUTUBE_SUBGEN_OUT_DIR = path.join(
   'subminer',
   'youtube-subs',
 );
-export function getDefaultLauncherLogFile(options?: {
-  platform?: NodeJS.Platform;
-  homeDir?: string;
-  appDataDir?: string;
-}): string {
-  return resolveDefaultLogFilePath('launcher', {
-    platform: options?.platform ?? process.platform,
-    homeDir: options?.homeDir ?? os.homedir(),
-    appDataDir: options?.appDataDir,
-  });
-}
-
 export function getDefaultMpvLogFile(options?: {
   platform?: NodeJS.Platform;
   homeDir?: string;
   appDataDir?: string;
 }): string {
-  return resolveDefaultLogFilePath('mpv', {
-    platform: options?.platform ?? process.platform,
-    homeDir: options?.homeDir ?? os.homedir(),
-    appDataDir: options?.appDataDir,
-  });
+  const platform = options?.platform ?? process.platform;
+  const homeDir = options?.homeDir ?? os.homedir();
+  const baseDir =
+    platform === 'win32'
+      ? path.join(
+          options?.appDataDir?.trim() || path.join(homeDir, 'AppData', 'Roaming'),
+          'SubMiner',
+        )
+      : path.join(homeDir, '.config', 'SubMiner');
+  return path.join(baseDir, 'logs', `SubMiner-${new Date().toISOString().slice(0, 10)}.log`);
 }
 
 export const DEFAULT_MPV_LOG_FILE = getDefaultMpvLogFile();
@@ -87,7 +79,6 @@ export interface Args {
   recursive: boolean;
   profile: string;
   startOverlay: boolean;
-  youtubeMode?: 'download' | 'generate';
   whisperBin: string;
   whisperModel: string;
   whisperVadModel: string;

package.json

@@ -1,6 +1,6 @@
 {
   "name": "subminer",
-  "version": "0.9.2",
+  "version": "0.7.0",
   "description": "All-in-one sentence mining overlay with AnkiConnect and dictionary integration",
   "packageManager": "bun@1.3.5",
   "main": "dist/main-entry.js",
@@ -77,12 +77,6 @@
     "build:win": "bun run build && electron-builder --win nsis zip --publish never",
     "build:win:unsigned": "bun run build && node scripts/build-win-unsigned.mjs"
   },
-  "overrides": {
-    "app-builder-lib": "26.8.2",
-    "electron-builder-squirrel-windows": "26.8.2",
-    "minimatch": "10.2.3",
-    "tar": "7.5.11"
-  },
   "keywords": [
     "anki",
     "ankiconnect",
@@ -111,7 +105,7 @@
     "@types/node": "^25.3.0",
     "@types/ws": "^8.18.1",
     "electron": "^37.10.3",
-    "electron-builder": "26.8.2",
+    "electron-builder": "^26.8.1",
     "esbuild": "^0.25.12",
     "prettier": "^3.8.1",
     "typescript": "^5.9.3"
@@ -165,45 +159,12 @@
       "include": "build/installer.nsh"
     },
     "files": [
-      "**/*",
-      "!assets{,/**/*}",
-      "!src{,/**/*}",
-      "!launcher{,/**/*}",
-      "!docs{,/**/*}",
-      "!tests{,/**/*}",
-      "!packaging{,/**/*}",
-      "!README.md",
-      "!CHANGELOG.md",
-      "!AGENTS.md",
-      "!CLAUDE.md",
-      "!stats/src{,/**/*}",
-      "!stats/index.html",
-      "!stats/public{,/**/*}",
-      "!stats/package.json",
-      "!stats/tsconfig.json",
-      "!stats/vite.config.ts",
-      "!docs-site{,/**/*}",
-      "!changes{,/**/*}",
-      "!backlog{,/**/*}",
-      "!.tmp{,/**/*}",
-      "!release-*{,/**/*}",
-      "!dist/**/*.map",
-      "!dist/**/*.test.*",
-      "!dist/**/__tests__{,/**/*}",
-      "!scripts/**/*.test.*",
-      "!plugin{,/**/*}",
-      "!vendor/subminer-yomitan{,/**/*}",
-      "!vendor/yomitan-jlpt-vocab{,/**/*}",
-      "!vendor/texthooker-ui/src{,/**/*}",
-      "!vendor/texthooker-ui/node_modules{,/**/*}",
-      "!vendor/texthooker-ui/.svelte-kit{,/**/*}",
-      "!vendor/texthooker-ui/.vscode{,/**/*}",
-      "!vendor/texthooker-ui/public{,/**/*}",
-      "!vendor/texthooker-ui/README.md",
-      "!vendor/texthooker-ui/package.json",
-      "!vendor/texthooker-ui/package-lock.json",
-      "!vendor/texthooker-ui/tsconfig*.json",
-      "!node_modules/@libsql/linux-x64-musl{,/**/*}"
+      "dist/**/*",
+      "stats/dist/**/*",
+      "vendor/texthooker-ui/docs/**/*",
+      "vendor/texthooker-ui/package.json",
+      "package.json",
+      "scripts/get-mpv-window-macos.swift"
     ],
     "extraResources": [
       {

plugin/subminer/options.lua

@@ -33,7 +33,6 @@ function M.load(options_lib, default_socket_path)
 		auto_start = true,
 		auto_start_visible_overlay = true,
 		auto_start_pause_until_ready = true,
-		auto_start_pause_until_ready_timeout_seconds = 15,
 		osd_messages = true,
 		log_level = "info",
 		aniskip_enabled = true,

plugin/subminer/process.lua

@@ -2,9 +2,9 @@ local M = {}
 
 local OVERLAY_START_RETRY_DELAY_SECONDS = 0.2
 local OVERLAY_START_MAX_ATTEMPTS = 6
+local AUTO_PLAY_READY_TIMEOUT_SECONDS = 15
 local AUTO_PLAY_READY_LOADING_OSD = "Loading subtitle tokenization..."
 local AUTO_PLAY_READY_READY_OSD = "Subtitle tokenization ready"
-local DEFAULT_AUTO_PLAY_READY_TIMEOUT_SECONDS = 15
 
 function M.create(ctx)
 	local mp = ctx.mp
@@ -34,23 +34,6 @@ function M.create(ctx)
 		return options_helper.coerce_bool(raw_pause_until_ready, false)
 	end
 
-	local function resolve_pause_until_ready_timeout_seconds()
-		local raw_timeout_seconds = opts.auto_start_pause_until_ready_timeout_seconds
-		if raw_timeout_seconds == nil then
-			raw_timeout_seconds = opts["auto-start-pause-until-ready-timeout-seconds"]
-		end
-		if type(raw_timeout_seconds) == "number" then
-			return raw_timeout_seconds
-		end
-		if type(raw_timeout_seconds) == "string" then
-			local parsed = tonumber(raw_timeout_seconds)
-			if parsed ~= nil then
-				return parsed
-			end
-		end
-		return DEFAULT_AUTO_PLAY_READY_TIMEOUT_SECONDS
-	end
-
 	local function normalize_socket_path(path)
 		if type(path) ~= "string" then
 			return nil
@@ -135,9 +118,7 @@ function M.create(ctx)
 			end)
 		end
 		subminer_log("info", "process", "Pausing playback until SubMiner overlay/tokenization readiness signal")
-		local timeout_seconds = resolve_pause_until_ready_timeout_seconds()
-		if timeout_seconds and timeout_seconds > 0 then
-			state.auto_play_ready_timeout = mp.add_timeout(timeout_seconds, function()
+		state.auto_play_ready_timeout = mp.add_timeout(AUTO_PLAY_READY_TIMEOUT_SECONDS, function()
 			if not state.auto_play_ready_gate_armed then
 				return
 			end
@@ -149,7 +130,6 @@ function M.create(ctx)
 			release_auto_play_ready_gate("timeout")
 		end)
 	end
-	end
 
 	local function notify_auto_play_ready()
 		release_auto_play_ready_gate("tokenization-ready")
@@ -392,9 +372,12 @@ function M.create(ctx)
 			end)
 		end
 
-		launch_overlay_with_retry(1)
 		if texthooker_enabled then
-			ensure_texthooker_running(function() end)
+			ensure_texthooker_running(function()
+				launch_overlay_with_retry(1)
+			end)
+		else
+			launch_overlay_with_retry(1)
 		end
 	end
 
@@ -498,6 +481,7 @@ function M.create(ctx)
 			state.texthooker_running = false
 			disarm_auto_play_ready_gate()
 
+			ensure_texthooker_running(function()
 				local start_args = build_command_args("start")
 				subminer_log("info", "process", "Starting overlay: " .. table.concat(start_args, " "))
 
@@ -521,10 +505,7 @@ function M.create(ctx)
 						show_osd("Restarted successfully")
 					end
 				end)
-
-			if opts.texthooker_enabled then
-				ensure_texthooker_running(function() end)
-			end
+			end)
 		end)
 	end
 

scripts/build-changelog.test.ts

@@ -95,43 +95,6 @@ test('writeChangelogArtifacts ignores README, groups fragments by type, writes r
   }
 });
 
-test('writeChangelogArtifacts skips changelog prepend when release section already exists', async () => {
-  const { writeChangelogArtifacts } = await loadModule();
-  const workspace = createWorkspace('write-artifacts-existing-version');
-  const projectRoot = path.join(workspace, 'SubMiner');
-  const existingChangelog = [
-    '# Changelog',
-    '',
-    '## v0.4.1 (2026-03-07)',
-    '### Added',
-    '- Existing release bullet.',
-    '',
-  ].join('\n');
-
-  fs.mkdirSync(projectRoot, { recursive: true });
-  fs.mkdirSync(path.join(projectRoot, 'changes'), { recursive: true });
-  fs.writeFileSync(path.join(projectRoot, 'CHANGELOG.md'), existingChangelog, 'utf8');
-  fs.writeFileSync(path.join(projectRoot, 'changes', '001.md'), ['type: added', 'area: overlay', '', '- Stale release fragment.'].join('\n'), 'utf8');
-
-  try {
-    const result = writeChangelogArtifacts({
-      cwd: projectRoot,
-      version: '0.4.1',
-      date: '2026-03-08',
-    });
-
-    assert.deepEqual(result.deletedFragmentPaths, [path.join(projectRoot, 'changes', '001.md')]);
-    assert.equal(fs.existsSync(path.join(projectRoot, 'changes', '001.md')), false);
-
-    const changelog = fs.readFileSync(path.join(projectRoot, 'CHANGELOG.md'), 'utf8');
-    assert.equal(changelog, existingChangelog);
-    const releaseNotes = fs.readFileSync(path.join(projectRoot, 'release', 'release-notes.md'), 'utf8');
-    assert.match(releaseNotes, /## Highlights\n### Added\n- Existing release bullet\./);
-  } finally {
-    fs.rmSync(workspace, { recursive: true, force: true });
-  }
-});
-
 test('verifyChangelogReadyForRelease ignores README but rejects pending fragments and missing version sections', async () => {
   const { verifyChangelogReadyForRelease } = await loadModule();
   const workspace = createWorkspace('verify-release');

scripts/build-changelog.ts

@@ -341,34 +341,12 @@ export function writeChangelogArtifacts(options?: ChangelogOptions): {
   const version = resolveVersion(options ?? {});
   const date = resolveDate(options?.date);
   const fragments = readChangeFragments(cwd, options?.deps);
+  const releaseSection = buildReleaseSection(version, date, fragments);
   const existingChangelogPath = path.join(cwd, 'CHANGELOG.md');
   const existingChangelog = existsSync(existingChangelogPath)
     ? readFileSync(existingChangelogPath, 'utf8')
     : '';
   const outputPaths = resolveChangelogOutputPaths({ cwd });
-  const existingReleaseSection = extractReleaseSectionBody(existingChangelog, version);
-  if (existingReleaseSection !== null) {
-    log(`Existing section found for v${version}; skipping changelog prepend.`);
-    for (const fragment of fragments) {
-      rmSync(fragment.path);
-      log(`Removed ${fragment.path}`);
-    }
-
-    const releaseNotesPath = writeReleaseNotesFile(
-      cwd,
-      existingReleaseSection,
-      options?.deps,
-    );
-    log(`Generated ${releaseNotesPath}`);
-
-    return {
-      deletedFragmentPaths: fragments.map((fragment) => fragment.path),
-      outputPaths,
-      releaseNotesPath,
-    };
-  }
-
-  const releaseSection = buildReleaseSection(version, date, fragments);
   const nextChangelog = prependReleaseSection(existingChangelog, releaseSection, version);
 
   for (const outputPath of outputPaths) {