fix(ci): install stats deps in release builds

- Move completed tasks (85, 117, 118, 155) to backlog/completed/
- Delete superseded task files (166 verification, 172 drilldown)
- Add stats dashboard milestone m-1
- Add new tasks (190, 194)
- Update task metadata across remaining backlog items
- Add changelog fragments for stats, mpv args, and subtitle filtering

.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

@@ -1,142 +1,5 @@
 # Changelog
 
-## v0.9.3 (2026-03-25)
-
-### Changed
-- Launcher: Moved YouTube primary subtitle language defaults to `youtube.primarySubLanguages`.
-- Launcher: Removed the placeholder YouTube subtitle retime step and now uses downloaded primary subtitle tracks directly, so there is no fake path rewrite before playback/sidebar loading.
-- YouTube: Removed the `src/core/services/youtube/retime` helper and its tests after retiring the internal retime strategy.
-- Docs: Clarified optional `alass` / `ffsubsync` subtitle-sync requirements and setup steps, including fallback behavior when sync tools are absent.
-- Launcher: Removed the old `youtubeSubgen.primarySubLanguages` config path from the generated config and docs.
-
-## 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
-- Immersion: Added Mine Word, Mine Sentence, and Mine Audio buttons to word detail example lines in the stats dashboard.
-- Immersion: Mine Word creates a full Yomitan card (definition, reading, pitch accent) via the hidden search page bridge, then enriches with sentence audio, screenshot, and metadata extracted from the source video.
-- Immersion: Mine Sentence and Mine Audio create cards directly with appropriate Lapis/Kiku flags, sentence highlighting, and media from the source file.
-- Immersion: Media generation (audio + image/AVIF) runs in parallel and respects all AnkiConnect config options.
-- Immersion: Added word exclusion list to the Vocabulary tab with localStorage persistence and a management modal.
-- Immersion: Fixed truncated readings in the frequency rank table (e.g. お前 now shows おまえ instead of まえ).
-- Immersion: Clicking a bar in the Top Repeated Words chart now opens the word detail panel.
-- Immersion: Secondary subtitle text is now stored alongside primary subtitle lines for use as translation when mining cards from the stats page.
-- Stats: Added `subminer stats -b` to start or reuse a dedicated background stats server without blocking normal SubMiner instances.
-- Stats: Added `subminer stats -s` to stop the dedicated background stats server without closing browser tabs.
-- Stats: Stats server startup now reuses a running background stats daemon instead of trying to bind a second local server in another SubMiner instance.
-- Launcher: Added launcher passthrough for `-a/--args` so mpv receives raw extra launch flags (`--fs`, `--ytdl-format`, custom audio/video settings, etc.) from the `subminer` command.
-- Launcher: Added `subminer stats` to launch the local stats dashboard, force-start the stats server on demand, and open the dashboard in your browser.
-- Launcher: Added `subminer stats cleanup` to backfill vocabulary metadata and prune stale or excluded immersion rows on demand.
-- Launcher: Added `stats.autoOpenBrowser` so browser launch after `subminer stats` can be enabled or disabled explicitly.
-- Immersion: Added a local stats dashboard for immersion tracking with Overview, Anime, Trends, Vocabulary, and Sessions views.
-- Immersion: Added anime progress, episode completion, Anki card links, and occurrence drill-down across the stats dashboard.
-- Immersion: Added richer session timelines with new-word activity, cumulative totals, and pause/seek/card event markers.
-- Immersion: Added completed-episodes and completed-anime totals to the Overview tracking snapshot.
-
-### Changed
-- Anki: Changed known-word cache settings to live under `ankiConnect.knownWords` instead of mixing them into `ankiConnect.nPlusOne`.
-- Anki: Kept legacy `ankiConnect.nPlusOne` known-word keys and older `ankiConnect.behavior.nPlusOne*` keys as deprecated compatibility fallbacks.
-- Stats: Added session deletion to the Sessions tab with the same confirmation prompt used by anime episode/session deletes, and removed all associated session rows from the stats database.
-- Immersion: Kept immersion tracking history by default while preserving daily/monthly rollup maintenance.
-- Immersion: Added exact lifetime summary reads for overview/anime/media stats so dashboard totals no longer depend on rescanning raw telemetry.
-- Immersion: Reduced tracker storage overhead by removing duplicated subtitle text from subtitle-line event payloads.
-- Immersion: Deduplicated episode cover-art blobs through a shared blob store and updated cover-art reads/writes to resolve shared images correctly.
-- Immersion: Added indexes for large-history session, telemetry, vocabulary, kanji, and cover-art queries to keep dashboard reads fast as the SQLite database grows.
-- Immersion: Renamed the stats dashboard's Anime tab to Library so the media browser label matches non-anime sources like YouTube and other yt-dlp-backed content.
-- Anilist: Standardized episode completion threshold by introducing `DEFAULT_MIN_WATCH_RATIO` and using it for both local watched state transitions and AniList post-watch progress updates.
-- Anilist: Episode auto-marking now uses the same threshold as AniList (`85%`), removing divergent completion behavior.
-- Overlay: Excluded interjections and sound-effect tokens from subtitle annotation styling so they no longer inherit misleading lexical highlight treatment while still remaining visible and hoverable as plain subtitle tokens.
-- Overlay: Expanded subtitle annotation noise filtering to also strip annotation metadata from standalone grammar-only helper tokens such as particles, auxiliaries, adnominals, common explanatory endings like `んです` / `のだ`, and merged trailing quote-particle forms like `...って` while keeping them tokenized for hover lookup.
-
-### Fixed
-- Launcher: Fixed mpv Lua plugin binary auto-detection on Linux to also search `/usr/bin/subminer` and `/usr/local/bin/subminer` (lowercase), matching the conventional Unix wrapper name used by packaged installs such as the AUR package.
-- Stats: Fixed the in-app stats overlay so it connects to the configured `stats.serverPort` instead of falling back to the default port.
-- Overlay: Fixed subtitle frequency tagging for merged lookup-backed tokens like `陰に` by falling back to exact surface-form Yomitan frequencies when the normalized headword lookup misses.
-- Overlay: Fixed MeCab merged-token position mapping across line breaks so merged content-plus-particle tokens like `陰に` keep their matched Yomitan frequency instead of inheriting shifted POS tags.
-- Overlay: Fixed grouped frequency parsing in both Yomitan and fallback frequency-dictionary lookups so display values like `118,121` use the leading rank instead of collapsing the rank and occurrence count into `118121`.
-- Overlay: Fixed frequency-rank ingestion to ignore Yomitan dictionaries explicitly marked `occurrence-based`, so raw occurrence counts are no longer treated as subtitle rank values.
-- Overlay: Fixed inflected headword frequency tagging to prefer ranks from the selected Yomitan `termsFind` popup entry itself, ordered by configured dictionary priority, so forms like `潜み` use primary-dictionary ranks like `4073` before falling back to lower-priority raw lemma metadata such as `CC100`.
-- Overlay: Fixed annotation-stage frequency filtering so exact kanji noun tokens like `者` keep their matched rank even when MeCab labels them `名詞/非自立`, instead of dropping the highlight after scan-time frequency lookup succeeds.
-- Anki: Fixed repeated character-dictionary startup work by scheduling auto-sync only from mpv media-path changes instead of also re-triggering it from connection and media-title events for the same title.
-- Overlay: Fixed macOS fullscreen overlay stability by keeping the passive visible overlay from stealing focus, re-raising the overlay window when reasserting its macOS topmost level, and tolerating one transient macOS tracker/helper miss before hiding the overlay.
-- Overlay: Kept subtitle tokenization warmup one-shot for the lifetime of the app so later fullscreen/media churn on macOS does not replay the startup warmup gate after the first file is ready.
-- Overlay: Added a bounded macOS tracker loss-grace window so fullscreen enter/leave transitions do not immediately hide and reload the overlay when the helper briefly loses the mpv window.
-- Overlay: Skipped subtitle/tokenization refresh invalidation on character-dictionary auto-sync completion when the dictionary was already current, preventing startup flash/reload loops on unchanged media.
-- Stats: Fixed session stats so known-word counts track real known-word occurrences without collapsing subtitle-line gaps.
-- Stats: Fixed session word totals in session-facing stats views to prefer token counts when available, preventing known words from exceeding total words in the session chart.
-- Stats: Fixed the stats Vocabulary tab blank-screen regression caused by a hook-order crash after vocabulary data finished loading.
-- Anki: Fixed card-mine OSD feedback so the final mine result stops the Anki spinner first, then shows a single-line `✓`/`x` status without being overwritten by a later spinner tick.
-- Stats: Removed the misleading `New words` series from expanded session charts; session detail now shows only the real total-word and known-word lines.
-- Stats: Restored the cross-anime word table behavior in stats vocabulary surfaces so shared vocabulary entries no longer disappear or merge incorrectly across related media.
-- Stats: `subminer stats -b` now runs as a standalone background stats daemon instead of reusing the main SubMiner app process, so the overlay app can still be launched separately for normal video watching.
-- Stats: Dashboard word mining still works against the background daemon by using a short-lived hidden helper for the Yomitan add-note flow.
-- Stats: Load full session timelines by default in stats session detail views so long sessions preserve complete telemetry history instead of being truncated by a fixed sample limit.
-- Stats: Replaced heuristic stats word counts with Yomitan token counts, so session, media, anime, and trend subtitle totals now come directly from parsed subtitle tokens.
-- Stats: Updated stats UI labels and lookup-rate copy to refer to tokens instead of words where those counts are shown.
-- Overlay: Reduced repeated `Overlay loading...` popups on macOS when fullscreen tracker flaps briefly hide and recover the visible overlay.
-- Stats: Scaled expanded session-detail known-word charts to the session's actual percentage range so small changes no longer render as a nearly flat line.
-- Jlpt: Reduced JLPT dictionary startup log noise by summarizing duplicate surface-form collisions instead of logging one line per duplicate entry.
-
 ## v0.6.5 (2026-03-15)
 
 ### Internal

README.md

@@ -1,171 +1,83 @@
 <div align="center">
+  <img src="assets/SubMiner.png" width="169" alt="SubMiner logo">
+  <h1>SubMiner</h1>
+  <strong>Look up words, mine to Anki, and enrich cards with context — without leaving mpv.</strong>
+  <br /><br />
 
-<img src="assets/SubMiner.png" width="160" alt="SubMiner logo">
-
-# SubMiner
-
-## Turn mpv into a sentence-mining workstation.
-
-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)]()
+[![Docs](https://img.shields.io/badge/docs-docs.subminer.moe-blueviolet)](https://docs.subminer.moe)
 
 </div>
 
-## How It Works
+<br />
 
-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.
+<div align="center">
+
+[![SubMiner demo (Animated preview)](./assets/minecard.webp)](./assets/minecard.mp4)
+
+</div>
+
+<br />
+
+## What it does
+
+SubMiner is an Electron overlay that sits on top of mpv. It turns your video player into a full sentence-mining workstation — look up any word with Yomitan, mine it to Anki with one key, and track your immersion progress over time.
 
 ## Features
 
-### Dictionary Lookups
+### Dictionary Lookups While You Watch
 
-Yomitan runs inside the overlay. Trigger a lookup on any word for full dictionary popups — definitions, pitch accent, frequency data — without ever leaving mpv.
+Yomitan runs directly inside the overlay. Hover over any word in the subtitles or navigate with keyboard/controller to get full dictionary popups without pausing or switching windows.
+
+### One-Key Anki Mining
+
+Press a single key to send a word to Anki. SubMiner auto-fills the card with the sentence, audio clip, screenshot, and machine translation — all captured from the exact moment you looked it up.
 
 <div align="center">
-  <img src="docs-site/public/screenshots/yomitan-lookup.png" width="800" alt="Yomitan dictionary popup over annotated subtitles in mpv">
+  <img src="docs-site/public/screenshots/yomitan-lookup.png" width="800" alt="One-key Anki card creation — Yomitan popup with dictionary entry and mine button over annotated subtitles">
 </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.
+Subtitles are annotated in real time with N+1 targeting, frequency-dictionary highlighting, JLPT level tags, and a character name dictionary for anime and manga proper nouns.
 
 <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">
+  <img src="docs-site/public/screenshots/annotations.png" width="800" alt="Subtitle annotations with frequency highlighting, JLPT underlines, known words, N+1 targets, and character names">
+  <br/>
+  <img src="docs-site/public/screenshots/annotations-key.png" width="800" alt="Subtitle annotations with frequency highlighting, JLPT underlines, known words, N+1 targets, and character names">
 </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.
+A local stats dashboard tracks your watch time, anime progress, vocabulary growth, mining throughput, and session history. Drill down into individual sessions or browse your full library.
 
 <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">
+  <img src="docs-site/public/screenshots/stats-overview.png" width="800" alt="Stats dashboard — overview with watch time, cards mined, streaks, and tracking snapshot">
+  <br /><br />
+  <img src="docs-site/public/screenshots/stats-vocabulary.png" width="800" alt="Vocabulary tab — unique words, known words, top repeated words, and unmined word list">
+  <br /><br />
+  <!-- <img src="docs-site/public/screenshots/stats-library.png" width="800" alt="Library tab — anime grid with episode counts and watch time"> -->
 </div>
 
-<br>
+### External Integrations
 
-### 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 — requires <code>alass</code> or <code>ffsubsync</code> on your <code>PATH</code> (optional; subtitle syncing is disabled without them)</td>
-  </tr>
-  <tr>
-    <td><b>WebSocket</b></td>
-    <td>Annotated subtitle feed for external clients (texthooker pages, custom tools)</td>
-  </tr>
-</table>
+- **AniList** — Automatic episode progress tracking
+- **Jellyfin** — Remote playback, cast device mode
+- **Subtitle tools** — Download from Jimaku, sync with alass/ffsubsync
+- **Texthooker & API** — Custom texthooker page and annotated websocket feed for external clients
 
 <div align="center">
-  <img src="docs-site/public/screenshots/texthooker.png" width="800" alt="Texthooker page receiving annotated subtitle lines via WebSocket">
+  <img src="docs-site/public/screenshots/texthooker.png" width="800" alt="Texthooker page with annotated subtitle lines — known words, N+1 targets, character names, and frequency highlighting">
 </div>
 
-<br>
-
----
-
-## Requirements
-
-|                | Required                                | Optional                               |
-| -------------- | --------------------------------------- | -------------------------------------- |
-| **Player**     | [`mpv`](https://mpv.io) with IPC socket | —                                      |
-| **Processing** | `ffmpeg`, `mecab` + `mecab-ipadic`      | `guessit` (AniSkip), `alass` / `ffsubsync` (subtitle sync) |
-| **Media**      | —                                       | `yt-dlp`, `chafa`, `ffmpegthumbnailer` |
-| **Selection**  | —                                       | `fzf` / `rofi`                         |
-
-> [!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.
-
-**Platform-specific:**
-
-| Linux                               | macOS                    | Windows       |
-| ----------------------------------- | ------------------------ | ------------- |
-| `hyprctl` or `xdotool` + `xwininfo` | Accessibility permission | No extra deps |
-
-<details>
-<summary><b>Arch Linux</b></summary>
-
-```bash
-paru -S --needed mpv ffmpeg mecab-git mecab-ipadic
-# Optional
-paru -S --needed yt-dlp fzf rofi chafa ffmpegthumbnailer xdotool xorg-xwininfo
-# Optional: subtitle sync (install at least one for subtitle syncing to work)
-paru -S --needed alass python-ffsubsync
-# X11 / XWAYLAND
-paru -S --needed xdotool xorg-xwininfo
-```
-
-</details>
-
-<details>
-<summary><b>macOS</b></summary>
-
-```bash
-brew install mpv ffmpeg mecab mecab-ipadic
-# Optional
-brew install yt-dlp fzf rofi chafa ffmpegthumbnailer
-# Optional: subtitle sync (install at least one for subtitle syncing to work)
-brew install alass
-pip install ffsubsync
-```
-
-Grant Accessibility permission to SubMiner in **System Settings > Privacy & Security > Accessibility**.
-
-</details>
-
-<details>
-<summary><b>Windows</b></summary>
-
-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>
-
----
-
-## Quick Start
+## Quick start
 
 ### 1. Install
 
-<details>
-<summary><b>Arch Linux (AUR)</b></summary>
+**Arch Linux (AUR):**
+
+Install [`subminer-bin`](https://aur.archlinux.org/packages/subminer-bin) from the AUR. It installs the packaged AppImage plus the `subminer` wrapper:
 
 ```bash
 paru -S subminer-bin
@@ -174,85 +86,84 @@ paru -S subminer-bin
 Or manually:
 
 ```bash
-git clone https://aur.archlinux.org/subminer-bin.git && cd subminer-bin && makepkg -si
+git clone https://aur.archlinux.org/subminer-bin.git
+cd subminer-bin
+makepkg -si
 ```
 
-</details>
-
-<details>
-<summary><b>Linux (AppImage)</b></summary>
+**Linux (AppImage):**
 
 ```bash
-mkdir -p ~/.local/bin
 wget https://github.com/ksyasuda/SubMiner/releases/latest/download/SubMiner.AppImage -O ~/.local/bin/SubMiner.AppImage \
 	&& chmod +x ~/.local/bin/SubMiner.AppImage
 wget https://github.com/ksyasuda/SubMiner/releases/latest/download/subminer -O ~/.local/bin/subminer \
 	&& chmod +x ~/.local/bin/subminer
+
 ```
 
 > [!NOTE]
 > The `subminer` wrapper uses a [Bun](https://bun.sh) shebang. Make sure `bun` is on your `PATH`.
 
-</details>
+**macOS (DMG/ZIP):** download the latest packaged build from [GitHub Releases](https://github.com/ksyasuda/SubMiner/releases/latest) and drag `SubMiner.app` into `/Applications`.
 
-<details>
-<summary><b>macOS</b></summary>
+**Windows (Installer/ZIP):** download the latest `SubMiner-<version>.exe` installer or portable `.zip` from [GitHub Releases](https://github.com/ksyasuda/SubMiner/releases/latest). Keep `mpv` installed and available on `PATH`.
 
-Download the latest DMG or ZIP from [GitHub Releases](https://github.com/ksyasuda/SubMiner/releases/latest) and drag `SubMiner.app` into `/Applications`.
+**From source** — see [docs.subminer.moe/installation#from-source](https://docs.subminer.moe/installation#from-source).
 
-</details>
-
-<details>
-<summary><b>Windows</b></summary>
-
-Download the latest installer or portable `.zip` from [GitHub Releases](https://github.com/ksyasuda/SubMiner/releases/latest). Make sure `mpv` is on your `PATH`.
-
-</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
+### 2. Launch the app once
 
 ```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
+# Linux
+SubMiner.AppImage
 ```
 
----
+On macOS, launch `SubMiner.app`. On Windows, launch `SubMiner.exe` from the Start menu or install directory.
+
+On first launch, SubMiner:
+
+- starts in the tray/background
+- creates the default config directory and `config.jsonc`
+- opens a compact setup popup
+- can install the mpv plugin to the default mpv scripts location for you
+- links directly to Yomitan settings so you can install dictionaries before finishing setup
+
+### 3. Finish setup
+
+- click `Install mpv plugin` if you want the default plugin auto-start flow
+- click `Open Yomitan Settings` and install at least one dictionary
+- click `Refresh status`
+- click `Finish setup`
+
+The mpv plugin step is optional. Yomitan must report at least one installed dictionary before setup can be completed.
+
+### 4. Mine
+
+```bash
+subminer video.mkv # default plugin config auto-starts visible overlay + resumes playback when ready
+subminer --start video.mkv # optional explicit overlay start when plugin auto_start=no
+subminer stats # open the local stats dashboard in your browser
+```
+
+## Requirements
+
+| Required                                   | Optional                                           |
+| ------------------------------------------ | -------------------------------------------------- |
+| `bun` (source builds, Linux `subminer`)    |                                                    |
+| `mpv` with IPC socket                      | `yt-dlp`                                           |
+| `ffmpeg`                                   | `guessit` (better AniSkip title/episode detection) |
+| `mecab` + `mecab-ipadic`                   | `fzf` / `rofi`                                     |
+| Linux: `hyprctl` or `xdotool` + `xwininfo` | `chafa`, `ffmpegthumbnailer`                       |
+| macOS: Accessibility permission            |                                                    |
+
+Windows builds use native window tracking and do not require the Linux compositor helper tools.
 
 ## Documentation
 
-Full guides on configuration, Anki setup, Jellyfin, immersion tracking, and more: **[docs.subminer.moe](https://docs.subminer.moe)**
-
----
+For full guides on configuration, Anki, Jellyfin, immersion tracking/stats, and more, see [docs.subminer.moe](https://docs.subminer.moe). The VitePress source for that site lives in [`docs-site/`](./docs-site/).
 
 ## 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 the shoulders of [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 powered by [Jimaku.cc](https://jimaku.cc). Dictionary lookups via [Yomitan](https://github.com/yomidevs/yomitan), and JLPT tags from [yomitan-jlpt-vocab](https://github.com/stephenmk/yomitan-jlpt-vocab).
 
 ## License
 

backlog/completed/task-195 - Keep-final-card-mine-OSD-result-from-being-overwritten-by-progress-spinner.md

@@ -1,64 +0,0 @@
----
-id: TASK-195
-title: Keep final card-mine OSD result from being overwritten by progress spinner
-status: Done
-assignee:
-  - Codex
-created_date: '2026-03-18 19:40'
-updated_date: '2026-03-18 19:49'
-labels:
-  - anki
-  - ui
-  - bug
-milestone: m-1
-dependencies: []
-references:
-  - src/anki-integration/ui-feedback.ts
-  - src/anki-integration.ts
-  - src/anki-integration/card-creation.ts
-priority: medium
-ordinal: 105610
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-
-When a card mine finishes, the mpv OSD currently tries to show the final status text but the in-flight Anki progress spinner can immediately overwrite it on the next tick. Stop the spinner first, then show a single-line final result with a success/failure marker and the mined-word notification.
-
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-
-<!-- AC:BEGIN -->
-
-- [x] #1 Successful mine/update OSD results render after the spinner is stopped and do not get overwritten by a later spinner tick.
-- [x] #2 Failure results that replace the spinner show an `x` marker and stay visible on the same OSD line.
-- [x] #3 Regression coverage locks the spinner teardown/result-notification ordering.
-<!-- AC:END -->
-
-## Implementation Plan
-
-<!-- SECTION:PLAN:BEGIN -->
-
-1. Add a focused failing regression test around the Anki UI-feedback spinner/result helper.
-2. Add a helper that stops progress before emitting the final OSD result line with `✓`/`x`.
-3. Route mine/update result notifications through that helper, then run targeted verification.
-<!-- SECTION:PLAN:END -->
-
-## Outcome
-
-<!-- SECTION:OUTCOME:BEGIN -->
-
-Added a dedicated Anki UI-feedback result helper that force-clears the in-flight spinner state before emitting the final OSD result line. Successful card-update notifications now render as `✓ Updated card: ...`, and sentence-card creation failures now render as `x Sentence card failed: ...` without a later spinner tick reclaiming the line.
-
-Verification:
-
-- `bun test src/anki-integration/ui-feedback.test.ts`
-- `bun test src/anki-integration/ui-feedback.test.ts src/anki-integration/note-update-workflow.test.ts src/anki-integration.test.ts src/core/services/mining.test.ts src/main/runtime/mining-actions.test.ts`
-- `bun x prettier --check src/anki-integration/ui-feedback.ts src/anki-integration/ui-feedback.test.ts src/anki-integration.ts src/anki-integration/card-creation.ts "backlog/tasks/task-195 - Keep-final-card-mine-OSD-result-from-being-overwritten-by-progress-spinner.md" changes/2026-03-18-mine-osd-spinner-result.md`
-- `bun run changelog:lint`
-- `bash .agents/skills/subminer-change-verification/scripts/verify_subminer_change.sh --lane core src/anki-integration/ui-feedback.ts src/anki-integration/ui-feedback.test.ts src/anki-integration.ts src/anki-integration/card-creation.ts changes/2026-03-18-mine-osd-spinner-result.md`
-- Verifier artifacts: `.tmp/skill-verification/subminer-verify-20260318-194614-uZMrAx/`
-
-<!-- SECTION:OUTCOME:END -->

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-169 - Cut-minor-release-v0.7.0-for-stats-and-runtime-polish.md

@@ -1,80 +0,0 @@
----
-id: TASK-169
-title: Cut minor release v0.7.0 for stats and runtime polish
-status: Done
-assignee:
-  - codex
-created_date: '2026-03-19 17:20'
-updated_date: '2026-03-19 17:31'
-labels:
-  - release
-  - docs
-  - minor
-dependencies:
-  - TASK-168
-references:
-  - package.json
-  - README.md
-  - docs/RELEASING.md
-  - docs-site/changelog.md
-  - CHANGELOG.md
-  - release/release-notes.md
-priority: high
-ordinal: 108000
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-Prepare the next release cut as `v0.7.0`, keeping 0-ver semantics by rolling the accumulated stats/dashboard, launcher, overlay, and stability work into the next minor line instead of a `1.0.0` release.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 Repository version metadata is updated to `0.7.0`.
-- [x] #2 Root release-facing docs are refreshed for the `0.7.0` release cut.
-- [x] #3 `CHANGELOG.md` and `release/release-notes.md` contain the committed `v0.7.0` section and consumed fragments are removed.
-- [x] #4 Public changelog/docs surfaces reflect the new release.
-- [x] #5 Release-prep verification is recorded.
-<!-- AC:END -->
-
-## Implementation Plan
-
-<!-- SECTION:PLAN:BEGIN -->
-1. Bump `package.json` to `0.7.0`.
-2. Refresh release-facing docs: root `README.md`, release guide versioning note, and public docs changelog summary.
-3. Run `bun run changelog:build --version 0.7.0` to commit release artifacts and consume pending fragments.
-4. Run release-prep verification (`changelog`, typecheck, tests, docs build if docs-site changed).
-5. Update this task with notes, verification, and final summary.
-<!-- SECTION:PLAN:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-Bumped `package.json` from `0.6.5` to `0.7.0` and refreshed the root release-facing copy in `README.md` so the release prep explicitly calls out the new stats/dashboard line plus the background stats daemon commands. Updated `docs/RELEASING.md` with the repo's 0-ver versioning policy and an explicit `--date` reminder after the changelog generator initially stamped `2026-03-20` from UTC instead of the intended local release date `2026-03-19`.
-
-Ran `bun run changelog:build --version 0.7.0`, which generated `CHANGELOG.md` and `release/release-notes.md` and removed the queued `changes/*.md` fragments for the accumulated stats, launcher, overlay, JLPT, and stability work. Added a curated `v0.7.0` summary to `docs-site/changelog.md` so the public docs changelog stays aligned with the committed root changelog while remaining user-facing.
-
-Verification:
-- `bash .agents/skills/subminer-change-verification/scripts/classify_subminer_diff.sh`
-- `bun run changelog:lint`
-- `bun run changelog:check --version 0.7.0`
-- `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 minor release `v0.7.0` as the next 0-ver major line. Version metadata, root changelog, generated release notes, README release copy, release-guide policy, and the public docs changelog are now aligned for the release cut.
-
-Docs update required: yes. Completed in `README.md`, `docs/RELEASING.md`, and `docs-site/changelog.md`.
-Changelog fragment required: no new fragment for this task. Existing pending release fragments were consumed into the committed `v0.7.0` changelog section and `release/release-notes.md`.
-
-Release-prep verification passed across changelog validation, config-example verification, typecheck, fast/env tests, full build, and docs-site test/build.
-<!-- SECTION:FINAL_SUMMARY:END -->

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

@@ -1,65 +0,0 @@
----
-id: TASK-177.1
-title: Fix overview lookup rate metric
-status: Done
-assignee:
-  - '@codex'
-created_date: '2026-03-19 17:46'
-updated_date: '2026-03-23 03:22'
-labels:
-  - stats
-  - immersion-tracking
-  - yomitan
-dependencies: []
-references:
-  - stats/src/components/overview/OverviewTab.tsx
-  - stats/src/lib/dashboard-data.ts
-  - stats/src/lib/yomitan-lookup.ts
-  - src/core/services/immersion-tracker/query.ts
-  - src/core/services/stats-server.ts
-parent_task_id: TASK-177
-priority: medium
-ordinal: 132500
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-Update the stats homepage Tracking Snapshot so Lookup Rate reflects lifetime intentional Yomitan lookups normalized by total tokens seen, matching the newer stats semantics already used in session, media, and anime views.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 Overview data exposes the lifetime totals needed to compute global Yomitan lookups per 100 tokens on the homepage
-- [x] #2 The homepage Tracking Snapshot Lookup Rate card shows Yomitan lookup rate as `X / 100 tokens` with tooltip/copy aligned to that meaning
-- [x] #3 Automated tests cover the lifetime totals plumbing and homepage summary/rendering change
-<!-- AC:END -->
-
-## Implementation Plan
-
-<!-- SECTION:PLAN:BEGIN -->
-1. Extend overview lifetime hints/query plumbing to include total tokens seen and total intentional Yomitan lookups from finished sessions.
-2. Add/adjust focused tests first for query hints, stats overview API typing/mocks, and overview summary formatting so the homepage metric fails under old semantics.
-3. Update the overview summary/card to derive Lookup Rate from lifetime Yomitan lookups per 100 tokens and align tooltip/copy with that meaning.
-4. Run focused verification on the touched query, stats-server, and stats UI tests; record results and blockers in the task notes.
-<!-- SECTION:PLAN:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-Extended overview lifetime hints to include total tokens seen and total intentional Yomitan lookups from finished sessions so the homepage can compute a true global lookup rate.
-
-Extracted the homepage Tracking Snapshot into a dedicated presentational component to keep OverviewTab smaller and make the Lookup Rate card copy directly testable.
-
-Focused verification passed for query hints, IPC/stats overview plumbing, stats server overview response, dashboard summary logic, and homepage snapshot rendering.
-
-SubMiner verifier core lane artifact: .tmp/skill-verification/subminer-verify-20260319-105320-7FDlwh. `bun run typecheck` passed there; `bun run test:fast` failed for a pre-existing/unrelated environment issue in scripts/update-aur-package.test.ts because scripts/update-aur-package.sh reported `mapfile: command not found`.
-<!-- SECTION:NOTES:END -->
-
-## Final Summary
-
-<!-- SECTION:FINAL_SUMMARY:BEGIN -->
-Homepage Lookup Rate now uses lifetime intentional Yomitan lookups normalized by lifetime tokens seen, matching the existing session/media/anime semantics instead of the old known-word hit-rate metric. I extended overview query hints and API typings with total token and Yomitan lookup totals, updated the overview summary builder to reuse the shared per-100-token formatter, and replaced the inline Tracking Snapshot block with a dedicated component that renders `X / 100 tokens` plus Yomitan-specific tooltip copy.
-
-Tests added/updated: query hints coverage for the new lifetime totals, stats server and IPC overview fixtures, overview summary assertions, and a dedicated Tracking Snapshot render test for the homepage card text. Focused `bun test` runs passed for those touched areas. Repo-native verifier `--lane core` also passed `bun run typecheck`; its `bun run test:fast` step still fails for the unrelated existing `scripts/update-aur-package.sh: line 71: mapfile: command not found` environment issue.
-<!-- SECTION:FINAL_SUMMARY:END -->

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

@@ -1,63 +0,0 @@
----
-id: TASK-177.2
-title: Count homepage new words by headword
-status: Done
-assignee:
-  - '@codex'
-created_date: '2026-03-19 19:38'
-updated_date: '2026-03-23 03:22'
-labels:
-  - stats
-  - immersion-tracking
-  - vocabulary
-dependencies: []
-references:
-  - src/core/services/immersion-tracker/query.ts
-  - stats/src/components/overview/TrackingSnapshot.tsx
-  - stats/src/lib/dashboard-data.ts
-parent_task_id: TASK-177
-priority: medium
-ordinal: 130500
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-Align the homepage New Words metric with the Known Words semantics by counting distinct headwords first seen in the selected window, so inflected or alternate forms of the same word do not inflate the summary.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 Homepage new-word counts use distinct headwords by earliest first-seen timestamp instead of counting separate word-form rows
-- [x] #2 Homepage tooltip/copy reflects the headword-based semantics
-- [x] #3 Automated tests cover the headword de-duplication behavior and affected overview copy
-<!-- AC:END -->
-
-## Implementation Plan
-
-<!-- SECTION:PLAN:BEGIN -->
-1. Change the new-word aggregate query to group `imm_words` by headword, compute each headword's earliest `first_seen`, and count headwords whose first sighting falls within today/week windows.
-2. Add failing tests first for the aggregate path so multiple rows sharing a headword only contribute once.
-3. Update homepage tooltip/copy to say unique headwords first seen today/week.
-4. Run focused query and stats overview tests, then record verification and any blockers.
-<!-- SECTION:PLAN:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-Updated the new-word aggregate to count distinct headwords by each headword's earliest `first_seen` timestamp, so multiple inflected/form rows for the same headword contribute only once.
-
-Adjusted homepage tooltip copy to say unique headwords first seen today/week, keeping the visible card labels unchanged.
-
-Focused verification passed for the query aggregate and homepage snapshot tests.
-
-SubMiner verifier core lane artifact: .tmp/skill-verification/subminer-verify-20260319-123942-4intgW. `bun run typecheck` passed there; `bun run test:fast` still fails for the unrelated environment issue in scripts/update-aur-package.test.ts (`mapfile: command not found`).
-<!-- SECTION:NOTES:END -->
-
-## Final Summary
-
-<!-- SECTION:FINAL_SUMMARY:BEGIN -->
-Homepage New Words now uses headword-level semantics instead of counting separate `(headword, word, reading)` rows. The aggregate query groups `imm_words` by headword, uses each headword's earliest `first_seen`, and counts headwords first seen today or this week so alternate forms do not inflate the summary. The homepage tooltip copy now explicitly says the metric is based on unique headwords.
-
-Added focused regression coverage for the de-duplication rule in `getQueryHints` and for the updated homepage tooltip text. Targeted `bun test` runs passed for the touched query and stats UI files. Repo verifier `--lane core` again passed `bun run typecheck`; `bun run test:fast` remains blocked by the unrelated existing `scripts/update-aur-package.sh: line 71: mapfile: command not found` failure.
-<!-- SECTION:FINAL_SUMMARY:END -->

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

@@ -1,65 +0,0 @@
----
-id: TASK-177.3
-title: Fix attached stats command flow and browser config
-status: Done
-assignee:
-  - '@codex'
-created_date: '2026-03-19 20:15'
-updated_date: '2026-03-23 03:22'
-labels:
-  - launcher
-  - stats
-  - cli
-dependencies: []
-references:
-  - launcher/commands/stats-command.ts
-  - launcher/commands/command-modules.test.ts
-  - launcher/main.test.ts
-  - src/main/runtime/stats-cli-command.ts
-  - src/main/runtime/stats-cli-command.test.ts
-parent_task_id: TASK-177
-priority: medium
-ordinal: 129500
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-Make `subminer stats` stay attached to the foreground app process instead of routing through daemon startup, while keeping background/stop behavior on the daemon path. Ensure browser opening for stats respects only `stats.autoOpenBrowser` in the normal stats flow.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 Default `subminer stats` forwards through the attached foreground stats command path instead of the daemon-start path
-- [x] #2 `subminer stats --background` and `subminer stats --stop` continue using the daemon control path
-- [x] #3 Normal stats launches do not open a browser when `stats.autoOpenBrowser` is false, and automated tests cover the launcher/runtime regressions
-<!-- AC:END -->
-
-## Implementation Plan
-
-<!-- SECTION:PLAN:BEGIN -->
-1. Add failing launcher tests first so default `stats` expects `--stats` forwarding while `--background` and `--stop` continue to expect daemon control flags.
-2. Add/adjust runtime stats command tests to prove `stats.autoOpenBrowser=false` suppresses browser opening on the normal attached stats path.
-3. Patch launcher forwarding logic in `launcher/commands/stats-command.ts` to choose foreground vs daemon flags correctly without changing cleanup handling.
-4. Run targeted launcher and stats runtime tests, then record verification results and blockers.
-<!-- SECTION:PLAN:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-Confirmed root cause: launcher default `stats` flow always forwarded `--stats-daemon-start` plus `--stats-daemon-open-browser`, which detached the terminal process and bypassed `stats.autoOpenBrowser` because browser opening happened in daemon control instead of the normal stats CLI handler.
-
-Updated launcher forwarding so plain `subminer stats` now uses the attached `--stats` path, while explicit `--background` and `--stop` continue using daemon control flags.
-
-Added launcher regression coverage for the attached/default path and preserved background/stop expectations; added runtime coverage proving `stats.autoOpenBrowser=false` suppresses browser opening on the normal stats path.
-
-Verifier passed for `launcher-plugin` and `runtime-compat` lanes. Artifact: .tmp/skill-verification/subminer-verify-20260319-131703-ZaAaUV.
-<!-- SECTION:NOTES:END -->
-
-## Final Summary
-
-<!-- SECTION:FINAL_SUMMARY:BEGIN -->
-Fixed `subminer stats` so the default command now forwards to the normal attached `--stats` app path instead of the daemon-start path. That keeps the foreground process attached to the terminal as expected, while `subminer stats --background` and `subminer stats --stop` still use daemon control. Because the normal stats CLI path already respects `config.stats.autoOpenBrowser`, this also fixes the unwanted browser-open behavior that previously bypassed config via `--stats-daemon-open-browser`.
-
-Added launcher command and launcher integration regressions for the new forwarding behavior, plus a runtime stats CLI regression that asserts `stats.autoOpenBrowser=false` suppresses browser opening. Verification passed with targeted launcher tests, targeted runtime stats tests, and the SubMiner verifier `launcher-plugin` + `runtime-compat` lanes.
-<!-- SECTION:FINAL_SUMMARY:END -->

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

@@ -1,67 +0,0 @@
----
-id: TASK-182.2
-title: Improve session detail known-word chart scaling
-status: Done
-assignee:
-  - codex
-created_date: '2026-03-19 20:31'
-updated_date: '2026-03-23 03:22'
-labels:
-  - bug
-  - stats
-  - ui
-dependencies: []
-references:
-  - >-
-    /Users/sudacode/projects/japanese/SubMiner/stats/src/components/sessions/SessionDetail.tsx
-  - >-
-    /Users/sudacode/projects/japanese/SubMiner/stats/src/lib/session-detail.test.tsx
-parent_task_id: TASK-182
-ordinal: 128500
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-Adjust the expanded session-detail known-word percentage chart so the vertical range reflects the session's actual percent range instead of always spanning 0-100. Keep the chart easier to read while preserving the percent-based tooltip/legend behavior already used in the stats UI.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 Expanded session detail scales the known/unknown percent chart to the session's observed percent range instead of hard-coding a 0-100 top bound
-- [x] #2 The chart keeps a small headroom above the highest observed known-word percent so the line remains visually readable near the top edge
-- [x] #3 Automated frontend coverage locks the new percent-domain behavior and preserves existing session-detail rendering
-<!-- AC:END -->
-
-## Implementation Plan
-
-<!-- SECTION:PLAN:BEGIN -->
-1. Add a focused frontend regression test for the session-detail ratio chart domain calculation, covering a session whose known-word percentage stays in a narrow band below 100% and expecting a dynamic top bound with headroom.
-2. Update `stats/src/components/sessions/SessionDetail.tsx` to compute a dynamic percent-axis domain and matching ticks for the ratio chart, keeping the lower bound at 0%, adding modest padding above the highest known percentage, rounding to clean tick steps, and capping at 100%.
-3. Apply the computed percent-axis bounds consistently to the right-side Y axis and the session chart pause overlays so the visual framing stays aligned.
-4. Run targeted frontend tests and the SubMiner verification helper on the touched files, then record results and any blockers in the task.
-<!-- SECTION:PLAN:END -->
-
-## Implementation Notes
-
-<!-- SECTION:NOTES:BEGIN -->
-Implemented dynamic known-percentage axis scaling in `stats/src/components/sessions/SessionDetail.tsx`: the ratio chart now keeps a 0% floor, uses the highest observed known percentage plus 5 points of headroom for the top bound, rounds that bound up to clean 10-point ticks, caps at 100%, and enables `allowDataOverflow` so the stacked area chart actually honors the tighter domain.
-
-Added frontend regression coverage in `stats/src/lib/session-detail.test.tsx` for the axis-max helper, covering both a narrow-band session and near-100% cap behavior.
-
-Added user-visible changelog fragment `changes/2026-03-19-session-detail-chart-scaling.md`.
-
-Verification: `bun test stats/src/lib/session-detail.test.tsx` passed; `bun run typecheck` passed; `bun run changelog:lint` passed; `bash .agents/skills/subminer-change-verification/scripts/verify_subminer_change.sh --lane core stats/src/components/sessions/SessionDetail.tsx stats/src/lib/session-detail.test.tsx` ran and passed `typecheck` but failed `bun run test:fast` on a pre-existing unrelated issue in `scripts/update-aur-package.test.ts` / `scripts/update-aur-package.sh` (`mapfile: command not found`). Artifacts: `.tmp/skill-verification/subminer-verify-20260319-134440-JRHAUJ`.
-
-Docs decision: no internal docs update required; the behavior change is localized UI presentation with no API/workflow change. Changelog decision: yes, required and completed because the fix is user-visible.
-<!-- SECTION:NOTES:END -->
-
-## Final Summary
-
-<!-- SECTION:FINAL_SUMMARY:BEGIN -->
-Improved expanded session-detail chart readability by replacing the fixed 0-100 known-word percentage axis with a dynamic top bound based on the session’s highest observed known percentage plus modest headroom, rounded to clean ticks and capped at 100%. The ratio chart now also enables `allowDataOverflow` so Recharts preserves the tighter percent domain even though the stacked known/unknown areas sum to 100%.
-
-Added frontend regression coverage for the new axis-max behavior and a changelog fragment for the user-visible stats fix.
-
-Verification: `bun test stats/src/lib/session-detail.test.tsx`, `bun run typecheck`, and `bun run changelog:lint` passed. The SubMiner verification helper’s `core` lane also passed `typecheck`, but `bun run test:fast` remains red on a pre-existing unrelated bash-compat failure in `scripts/update-aur-package.test.ts` / `scripts/update-aur-package.sh` (`mapfile: command not found`).
-<!-- SECTION:FINAL_SUMMARY:END -->

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

@@ -1,46 +0,0 @@
----
-id: TASK-196
-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
-priority: high
-ordinal: 136500
----
-
-## Description
-
-<!-- SECTION:DESCRIPTION:BEGIN -->
-Investigate and fix file-backed subtitle annotation latency where prefetch should warm upcoming lines but live playback still tokenizes each subtitle line. Likely causes: cache-key mismatch between parsed cue text and mpv `sub-text`, and priority-window selection skipping the currently active cue during mid-line starts/seeks.
-<!-- SECTION:DESCRIPTION:END -->
-
-## Acceptance Criteria
-<!-- AC:BEGIN -->
-- [x] #1 Prefetched subtitle entries are reused when live subtitle text differs only by normalization details such as ASS `\N`, newline collapsing, or surrounding whitespace.
-- [x] #2 Priority-window selection includes the currently active cue when playback starts or seeks into the middle of a cue.
-- [x] #3 Regression tests cover the cache-hit normalization path and active-cue priority-window behavior.
-- [x] #4 Verification covers the touched prefetch/controller lane.
-<!-- AC:END -->
-
-## Implementation Plan
-
-<!-- SECTION:PLAN:BEGIN -->
-1. Add failing regression tests in `subtitle-processing-controller.test.ts` and `subtitle-prefetch.test.ts`.
-2. Normalize cache keys in the subtitle processing controller so prefetch/live paths share keys.
-3. Adjust prefetch priority-window selection to include the active cue.
-4. Run targeted tests, then SubMiner verification lane for touched files.
-<!-- SECTION:PLAN:END -->
-
-## Outcome
-
-<!-- SECTION:OUTCOME:BEGIN -->
-Normalized subtitle cache keys inside the processing controller so prefetched ASS/VTT/live subtitle text variants reuse the same cache entry, and changed priority-window selection to include the currently active cue based on cue end time. Added regression coverage for both paths and verified the change with the `core` lane.
-<!-- SECTION:OUTCOME:END -->