ci: seed config before linux-smoke launch so the server actually serves

With the PyGObject girepository-2.0 fix in place, the linux-smoke step
ran its server-boot assertion for the first time and failed: on a fresh
runner the first-run bootstrap writes a default config and calls
sys.exit(0) ("First run: generated default config ... then restart")
instead of serving, so /api/health never came up and the 15s wait
timed out.

That exit-on-first-run is deliberate product behavior (never silently
start in insecure no-auth mode), so adjust the test rather than the app:
invoke the server once to seed the config (it exits 0 before binding the
port), then launch it for real. /api/health requires no auth, so the
auto-generated token is irrelevant to the check.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

.gitea/workflows/build.yml

@@ -60,6 +60,16 @@ jobs:
         with:
           python-version: '3.11'
 
+      - name: Install native deps for dbus-python + PyGObject
+        run: |
+          # PyGObject >= 3.52 builds against girepository-2.0 (merged into
+          # GLib 2.80), not the old standalone girepository-1.0. ubuntu-latest
+          # (24.04) ships it as libgirepository-2.0-dev.
+          sudo apt-get update
+          sudo apt-get install -y --no-install-recommends \
+            libdbus-1-dev libglib2.0-dev pkg-config \
+            libcairo2-dev libgirepository-2.0-dev
+
       - name: Build Linux distribution
         run: |
           chmod +x build-dist-linux.sh

.gitea/workflows/release.yml

@@ -61,6 +61,10 @@ jobs:
           | Windows (portable) | \`MediaServer-{tag}-win-x64.zip\` |
           | Linux | \`MediaServer-{tag}-linux-x64.tar.gz\` |
           ''').strip())
+          # TODO(macos-runner): re-add the macOS rows below once a macOS
+          # runner is connected to Gitea and the build-macos job is re-enabled.
+          #     | macOS (Apple Silicon) | \`MediaServer-{tag}-macos-arm64.tar.gz\` |
+          #     | macOS (Intel) | \`MediaServer-{tag}-macos-x86_64.tar.gz\`
 
           print(json.dumps('\n\n'.join(sections)))
           ")
@@ -187,6 +191,16 @@ jobs:
         with:
           python-version: '3.11'
 
+      - name: Install native deps for dbus-python + PyGObject
+        run: |
+          # PyGObject >= 3.52 builds against girepository-2.0 (merged into
+          # GLib 2.80), not the old standalone girepository-1.0. ubuntu-latest
+          # (24.04) ships it as libgirepository-2.0-dev.
+          sudo apt-get update
+          sudo apt-get install -y --no-install-recommends \
+            libdbus-1-dev libglib2.0-dev pkg-config \
+            libcairo2-dev libgirepository-2.0-dev
+
       - name: Build Linux distribution
         run: |
           chmod +x build-dist-linux.sh
@@ -226,3 +240,68 @@ jobs:
             -H "Authorization: token $DEPLOY_TOKEN" \
             -H "Content-Type: application/octet-stream" \
             --data-binary "@$FILE"
+
+  # --- Build macOS tarball (best-effort; requires a macos runner) ---
+  # PyObjC wheels are macOS-only, so this job must run on a real Mac.
+  #
+  # TODO(macos-runner): Temporarily disabled via `if: false` because the
+  # Gitea instance currently has no macOS runner attached. To re-enable:
+  #   1. Connect a macOS runner to Gitea
+  #   2. Delete the `if: false` line below
+  #   3. Restore the macOS rows in the Downloads table generated by the
+  #      create-release job (search for the matching TODO(macos-runner)).
+  build-macos:
+    needs: create-release
+    if: false
+    runs-on: macos-latest
+    continue-on-error: true
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: actions/setup-node@v4
+        with:
+          node-version: '20'
+
+      - name: Build frontend
+        run: npm ci && npm run build
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: '3.11'
+
+      - name: Build macOS distribution
+        run: |
+          chmod +x build-dist-macos.sh
+          ./build-dist-macos.sh "${{ gitea.ref_name }}"
+
+      - name: Upload assets to release
+        env:
+          DEPLOY_TOKEN: ${{ secrets.DEPLOY_TOKEN }}
+        run: |
+          RELEASE_ID="${{ needs.create-release.outputs.release_id }}"
+          BASE_URL="${{ gitea.server_url }}/api/v1/repos/${{ gitea.repository }}"
+
+          FILE=$(ls build/MediaServer-*-macos-*.tar.gz | head -1)
+          NAME=$(basename "$FILE")
+
+          EXISTING=$(curl -s "$BASE_URL/releases/$RELEASE_ID/assets" \
+            -H "Authorization: token $DEPLOY_TOKEN")
+          ASSET_ID=$(echo "$EXISTING" | python3 -c "
+          import sys, json
+          assets = json.load(sys.stdin)
+          for a in assets:
+              if a['name'] == '$NAME':
+                  print(a['id'])
+                  break
+          " 2>/dev/null || true)
+
+          if [ -n "$ASSET_ID" ]; then
+            curl -s -X DELETE "$BASE_URL/releases/$RELEASE_ID/assets/$ASSET_ID" \
+              -H "Authorization: token $DEPLOY_TOKEN"
+          fi
+
+          curl -s -X POST \
+            "$BASE_URL/releases/$RELEASE_ID/assets?name=$NAME" \
+            -H "Authorization: token $DEPLOY_TOKEN" \
+            -H "Content-Type: application/octet-stream" \
+            --data-binary "@$FILE"

.gitea/workflows/test.yml

@@ -34,3 +34,73 @@ jobs:
 
       - name: Test
         run: pytest --tb=short -q || test $? -eq 5
+
+  # Linux smoke test: install the linux extra in the same way build-dist-linux.sh
+  # does, then boot the server and hit /api/health. Catches dependency-resolution
+  # and import-time regressions for the Linux distribution path.
+  linux-smoke:
+    if: "!startsWith(github.event.head_commit.message, 'chore: release')"
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: actions/setup-node@v4
+        with:
+          node-version: '20'
+
+      - name: Build frontend
+        run: npm ci && npm run build
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: '3.11'
+
+      - name: Install Linux system deps for dbus-python + PyGObject
+        run: |
+          # PyGObject >= 3.52 builds against girepository-2.0 (merged into
+          # GLib 2.80), not the old standalone girepository-1.0. ubuntu-latest
+          # (24.04) ships it as libgirepository-2.0-dev.
+          sudo apt-get update
+          sudo apt-get install -y --no-install-recommends \
+            libdbus-1-dev libglib2.0-dev pkg-config \
+            libcairo2-dev libgirepository-2.0-dev
+
+      - name: Install with linux extra
+        run: |
+          pip install --upgrade pip
+          pip install ".[linux]"
+
+      - name: Smoke — server boots and /api/health responds
+        run: |
+          # Headless Linux runners have no PulseAudio and no display
+          # server, so we disable the visualizer + update checker, and we
+          # use `dbus-run-session` to give LinuxMediaController a real
+          # session bus to talk to (otherwise dbus.SessionBus() would
+          # raise during startup). This isn't a full MPRIS integration
+          # test — it only proves the dispatcher selects the Linux
+          # controller, all imports resolve, and /api/health returns 200.
+          sudo apt-get install -y --no-install-recommends dbus-x11
+          export MEDIA_SERVER_VISUALIZER_ENABLED=false
+          export MEDIA_SERVER_UPDATE_CHECK_ENABLED=false
+          dbus-run-session -- bash -c '
+            # First run writes a default config (random token) and exits 0
+            # instead of serving, so the server is never left running in
+            # insecure no-auth mode. Run once to seed the config; the real
+            # launch below then finds it and actually boots. /api/health needs
+            # no auth, so the generated token is irrelevant here.
+            python -m media_server.main --no-tray --port 18765 || true
+            python -m media_server.main --no-tray --port 18765 &
+            SERVER_PID=$!
+            for i in $(seq 1 30); do
+              if curl -sf "http://127.0.0.1:18765/api/health" >/dev/null; then
+                echo "Health check passed"
+                kill $SERVER_PID
+                wait $SERVER_PID 2>/dev/null || true
+                exit 0
+              fi
+              sleep 0.5
+            done
+            echo "Server did not respond within 15s"
+            kill $SERVER_PID 2>/dev/null || true
+            exit 1
+          '

README.md

@@ -285,37 +285,44 @@ All connected WebSocket clients receive a `links_changed` notification when link
 
 ## Installation
 
+Dependencies are declared in `pyproject.toml`. Pick the extra that matches
+your OS — the Python deps differ enough between Windows / Linux / macOS
+that there's no single `pip install` line.
+
 ### Installing on Windows
 
 ```bash
-pip install -r requirements.txt
+pip install ".[windows]"
 ```
 
-Required packages: `winsdk`, `pywin32`, `pycaw`, `comtypes`
+Pulls in `winsdk`, `pywin32`, `pycaw`, `comtypes`, `pystray`, etc.
 
 ### Installing on Linux
 
 ```bash
-# Install system dependencies
+# System packages required to build dbus-python + PyGObject from sdist.
-sudo apt-get install python3-dbus python3-gi libdbus-1-dev libglib2.0-dev
+sudo apt-get install -y python3-pip python3-venv \
+    libdbus-1-dev libglib2.0-dev pkg-config
 
-pip install -r requirements.txt
+pip install ".[linux]"
 ```
 
 ### Installing on macOS
 
 ```bash
-pip install -r requirements.txt
+pip install ".[macos]"
 ```
 
-No additional dependencies - uses built-in `osascript`.
+Pulls in `pyobjc-framework-Cocoa` + `pyobjc-framework-Quartz` for the
+foreground-window probe; AppleScript-based media control uses the
+built-in `osascript`.
 
 ### Installing on Android (Termux)
 
 ```bash
 # In Termux
 pkg install python termux-api
-pip install -r requirements.txt
+pip install "."
 ```
 
 Requires Termux and Termux:API apps from F-Droid.
@@ -835,11 +842,19 @@ Install:
 sudo ./service/install_linux.sh install
 ```
 
-Enable and start for your user:
+**Enable user lingering** — required so `/run/user/$UID/bus` (the D-Bus
+session socket needed for MPRIS) exists even when no graphical session
+is active. Without this the server boots but every `/api/media/*` call
+silently returns idle.
 
 ```bash
-sudo systemctl enable media-server@$USER
+sudo loginctl enable-linger $USER
-sudo systemctl start media-server@$USER
+```
+
+Enable and start the templated unit for your user:
+
+```bash
+sudo systemctl enable --now media-server@$USER
 ```
 
 View logs:
@@ -848,6 +863,39 @@ View logs:
 journalctl -u media-server@$USER -f
 ```
 
+**Troubleshooting:**
+
+- *"`/api/media/status` always returns `idle`"* — check the service log for
+  `D-Bus session bus not available`. Most commonly: lingering isn't
+  enabled, or the unit is using the wrong `XDG_RUNTIME_DIR` (`%U` must
+  expand to the user's numeric UID).
+- *"Visualizer permanently unavailable"* — PulseAudio/PipeWire must
+  expose monitor sources. `pactl list sources short | grep monitor`
+  should list at least one entry; if not, install `pipewire-pulse` and
+  restart your session.
+- *"Volume control silently fails"* — `pactl` must be on `PATH` and the
+  user's PulseAudio/PipeWire server must be reachable
+  (`PULSE_RUNTIME_PATH=/run/user/$UID/pulse`).
+- *"Foreground window is always `null`"* — expected under Wayland; the
+  compositor hides window info from unprivileged clients. X11 sessions
+  work normally.
+
+### macOS (LaunchAgent)
+
+The distribution tarball ships an installer:
+
+```bash
+./install-launchagent.sh
+```
+
+This drops `~/Library/LaunchAgents/com.dolgolyov.media-server.plist`,
+starts the service immediately, and re-launches it at every login. Logs
+go to `~/Library/Logs/media-server/{stdout,stderr}.log`. To stop:
+
+```bash
+./uninstall-launchagent.sh
+```
+
 ## Command Line Options
 
 ```text

RELEASE_NOTES.md

@@ -1,55 +1,42 @@
-## v0.2.5 (2026-05-16)
+## v0.4.0 (2026-05-28)
 
-### Security
+Two headline changes since v0.3.1: **Media Server now ships first-class Linux and macOS builds** (no more Windows-only), and the **Windows app icon was redesigned** with a proper multi-resolution ICO so the installer, Start Menu, desktop shortcut, Alt+Tab, and system tray all render sharp.
 
-- **Loopback-by-default + auto-generated token:** Server now binds `127.0.0.1` by default; first-run bootstrap generates a random `api_token` and refuses to bind a non-loopback interface without auth unless explicitly opted in. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
+### Features
-- **Browser path-traversal hardening:** `BrowserService.validate_path` now rejects absolute paths, drive letters, UNC paths, and NUL bytes. `/api/browser/{play,metadata,thumbnail}` require a `folder_id` plus a folder-relative path — arbitrary filesystem reads via the browser API are no longer possible. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
+
-- **Strict input validation on links/scripts:** Pydantic validators reject non-http(s) URLs and any icon outside the `mdi:<slug>` namespace. Create/update/delete on scripts, callbacks, and links is gated by the corresponding `*_management` flags. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
+- **Linux support, production-ready.** New `linux` extra (`dbus-python`, `PyGObject`, `python-xlib`) plus a `build-dist-linux.sh` that emits a portable tarball and a systemd user unit. The unit sets `DBUS_SESSION_BUS_ADDRESS`, `XDG_RUNTIME_DIR`, and `ReadWritePaths` for `~/.config` / `~/.cache` so MPRIS works and audit-log / thumbnail writes aren't blocked by `ProtectHome`. The Linux MPRIS controller now connects to the session bus lazily — a missing or late bus no longer crashes lifespan startup, and the user is logged a one-line hint about `loginctl enable-linger`. ([ddf4a6c](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/ddf4a6c))
-- **Hardened response headers + CORS:** Strict `Content-Security-Policy`, `X-Frame-Options: DENY`, `Referrer-Policy: no-referrer`, `X-Content-Type-Options: nosniff`. CORS locked to `localhost:<port>` + `127.0.0.1:<port>` by default; configurable for trusted origins. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
+- **macOS support.** New `macos` extra (`pyobjc-framework-Cocoa`, `pyobjc-framework-Quartz`), a `build-dist-macos.sh` script, and a per-user LaunchAgent installer producing `MediaServer-vX.Y-macos-{arm64,x86_64}.tar.gz` artifacts. Spotify URL artwork is wired through `MediaController.get_album_art()`. ([ddf4a6c](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/ddf4a6c))
-- **Atomic config writes with restrictive permissions:** `config.yaml` writes go through a temp file + `os.replace` and land with `0o600` on POSIX, so a crash mid-write can never leave a half-written token on disk readable to other users. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
+- **Cross-platform album artwork.** `MediaController.get_album_art()` is now abstract with Linux (`mpris:artUrl`, `file://` + `http(s)://`) and macOS (Spotify URL) implementations; the `/api/media/artwork` endpoint awaits the controller. ([ddf4a6c](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/ddf4a6c))
-- **Subprocess process-group isolation:** Spawned scripts/callbacks now get their own process group (`CREATE_NEW_PROCESS_GROUP` on Windows, `start_new_session=True` on POSIX), so a timeout actually kills the whole tree instead of orphaning child processes. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
+- **Per-OS unavailable reasons on `/api/media/visualizer/status`** so the Web UI can explain *why* the visualizer is off on Linux / macOS instead of just hiding it. ([ddf4a6c](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/ddf4a6c))
-- **Frontend XSS hardening:** Monitor name + details are `escapeHtml`'d, the power button moved to a delegated `data-action` handler, and remote MDI SVGs are parsed and sanitized (strip `<script>`, `<foreignObject>`, `on*` handlers, `javascript:` hrefs) before they touch `innerHTML`. All dynamic URL segments now go through `encodeURIComponent`. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
+- **Startup preflight on Linux** warns when `DBUS_SESSION_BUS_ADDRESS` or `XDG_RUNTIME_DIR` is unset and informs the user when running under Wayland disables the foreground-window probe — so silent loss of features is now diagnosable from the log. ([ddf4a6c](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/ddf4a6c))
-- **CSP-compliant event wiring:** Strict `script-src 'self'` was blocking every inline `onclick`/`onchange`/`oninput`/`onsubmit` in the UI, leaving buttons and forms silently dead. All 53 inline handler attributes in `index.html` were renamed to `data-on*` and a new `wireInlineHandlers()` in `app.js` parses each expression on `DOMContentLoaded` and attaches a real `addEventListener` — supports no-arg calls, string/number/bool/null literals, and the `event` token. No `unsafe-inline` or `unsafe-hashes` needed. ([eaeebb6](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/eaeebb6))
+- **Redesigned app icon ("Beacon").** Replaces the generic Spotify-green circle with a refined squircle + deep-teal diagonal gradient (`#0B3D3B → #1A6B5E`) + warm parchment play triangle (`#F5F1E8`) with a drop shadow, top sheen, and ghosted echo-chevrons that hint at broadcast/stream. ([d798fed](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/d798fed))
+- **Multi-resolution Windows ICO.** `icon.ico` grew from a single 16×16 frame (208 B) to a 10-frame ICO (16/20/24/32/40/48/64/96/128/256 — ~37 KB) so Windows no longer upscales 16×16 into mush for the installer chrome, Start Menu, desktop shortcuts, Alt+Tab, and File Explorer tiles. ([d798fed](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/d798fed))
+- **System tray uses the new icon.** `tray.py` now picks a 64×64 frame from the multi-res ICO; the procedural fallback was reskinned to the same Beacon palette so a missing ICO no longer regresses the tray back to the old Spotify-green circle. ([d798fed](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/d798fed))
 
 ### Bug Fixes
 
-- **WebSocket reconnect robustness:** Close the previous socket before opening a new one, clear the ping interval per-socket, clear `reconnectTimeout` up-front, retry on `online`/`visibilitychange`, and wrap `JSON.parse` in try/catch — eliminates the stale-socket leaks and "stuck offline after sleep" cases. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
+- **`tray._confirm` guarded against `ctypes.windll` on non-Windows** so the new Linux / macOS builds don't crash when the tray prompts for confirmation. ([ddf4a6c](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/ddf4a6c))
-- **Artwork fetch race:** `AbortController` + generation guard so a rapid track change can no longer paint the previous track's artwork over the current one. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
+- **`config.example.yaml` defaults are now cross-platform.** Per-OS commented examples for the on/off scripts, and `on_turn_off` defaults to a harmless `echo` (the previous default silently failed everywhere but Windows). ([ddf4a6c](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/ddf4a6c))
-- **Audio analyzer no longer spins infinitely without a loopback device:** A sticky `_unavailable` flag short-circuits start/stop; cleared by `set_device()` so the user can recover once a device appears. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
-- **Volume short-circuit cache invalidation:** Cache is now busted when the server reports a remote volume change, so the UI no longer ignores volume updates that happened outside the app. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
-- **Browser thumbnail race:** Per-folder generation counter + `isConnected` checks; in-flight fetches are aborted on navigation, so thumbnails from a folder you already left can't paint into the current view. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
-- **Track-skip uses cached title** instead of a full WinRT status round-trip — skip feedback is now instant. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
-- **Browser list column alignment:** `.browser-list` switched to CSS grid + subgrid so header and rows share column tracks, eliminating the misaligned columns when content widths differed between rows. Matching responsive column overrides applied at the parent. Root-folder SVG sizing (hardcoded 24×24 in `browser.js`) now fills the 56px icon box instead of rendering at ~43%. Compact-grid icon fills its thumb wrapper so the emoji centers instead of being stranded top-left. Premature `isConnected` bail removed from `loadThumbnail` — the img element is intentionally detached when called from `renderBrowserGrid/List`, and the post-await checks already handle navigation-away correctly. ([982dda4](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/982dda4))
-
-### Performance
-
-- **Blocking IO off the event loop:** Linux MPRIS/`pactl` calls, `/api/display` DDC/CI handlers, and `browse_directory` are all wrapped in `asyncio.to_thread` — slow SMB shares or laggy monitors can no longer stall the entire async runtime. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
-- **Windows status poll loop reuse:** The 0.5s status poll now caches one asyncio loop per worker thread via `threading.local` instead of `new_event_loop`/`close` on every tick. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
-- **WebSocket broadcast: serialize once:** `broadcast()` serializes JSON a single time and uses `send_text` to fan out to all clients. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
-- **Thumbnail cache cleanup actually runs:** The hourly cleanup task was defined but never scheduled — it is now wired into the lifespan handler so the cache no longer grows unbounded. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
-- **Progress drag listeners attached only while dragging** — no more global `mousemove` handler firing on every cursor twitch. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
-
-### UI/UX
-
-- **Copper accent consistency:** Green leftover focus rings (`rgba(29,185,84,…)`) replaced with copper (`rgba(var(--copper-rgb),…)`) across the UI. Dialogs now have square corners and a copper top hairline so they read as part of the editorial chrome. `.browser-item` is transparent with a copper hover border (was a filled card). ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
-- **Audio device select** uses `var(--sans)` instead of the generic system font so it matches surrounding controls. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
-- **Mobile padding tuned for ≤480px screens.** ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
-- **Accessible breadcrumb home:** Now a real `<button>` with `aria-label`, and `aria-current` is set on the root. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
-- **i18n gaps filled:** `display.msg.power_*`, `execution.*`, `scripts.params.execute`, `callbacks.empty` now have proper en + ru strings. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
 
 ---
 
 ### Development / Internal
 
-#### Quality
+#### Build & Packaging
 
-- All `asyncio.get_event_loop()` in coroutines migrated to `get_running_loop()` (the former is deprecated in Python 3.12+). ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
+- **`scripts/generate-icon.py`** — SVG is the canonical source; `resvg-py` rasterizes every ICO size; Pillow packs the multi-resolution `icon.ico`. Re-run any time the SVG changes. ([d798fed](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/d798fed))
-- `ThreadPoolExecutor`s now shut down cleanly during lifespan teardown. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
+- **Dependency reorganization in `pyproject.toml`** — `screen-brightness-control` and `monitorcontrol` are cross-platform and moved to base deps; new `linux` and `macos` optional-deps groups with `sys_platform` markers. `resvg-py` added to `[dev]` for the icon-generation script. ([ddf4a6c](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/ddf4a6c), [d798fed](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/d798fed))
-- `config_manager` dedup: 12 near-identical CRUD methods collapsed onto generic `_upsert`/`_delete` helpers — about **290 lines removed** with no behavior change. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
+- **`install_linux.sh`** — dropped the dead `requirements.txt` path; now installs via `pip install ".[linux]"` and pre-creates the writable state dirs. ([ddf4a6c](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/ddf4a6c))
-- Service worker no longer pass-throughs every fetch. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
+- **Templated `media-server.service`** updated to match the new dist layout, with proper session-bus env vars and a writable state-dir grant. ([ddf4a6c](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/ddf4a6c))
-- M3U playlist written via `NamedTemporaryFile` so a fixed-path symlink can no longer clobber it. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
+
-- `__version__` prefers live `pyproject.toml` in dev checkouts so `pip install -e .` users see the source-of-truth version, not the stale metadata baked in at install time. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
+#### CI
-- `_broadcast_after_open` hardening: initialize status, swallow per-poll errors, and track background tasks in a strong-ref set with done-callback cleanup so they aren't garbage-collected mid-flight. ([bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40))
+
+- **New `linux-smoke` job** in `.gitea/workflows/test.yml` — installs `.[linux]`, boots the server under `dbus-run-session`, and asserts `/api/health`. Catches dependency-resolution and import-time regressions for the Linux dist path. ([ddf4a6c](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/ddf4a6c))
+- **Release workflow gains apt-deps step** for the Linux build and a best-effort macOS build job that produces the per-arch macOS tarballs. ([ddf4a6c](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/ddf4a6c))
+
+#### Documentation
+
+- **README rewritten for the new extras** — replaces the stale `pip install -r requirements.txt` instructions, adds a systemd-lingering note + troubleshooting section, and a macOS LaunchAgent section. ([ddf4a6c](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/ddf4a6c))
 
 ---
 
@@ -58,8 +45,7 @@
 
 | Hash | Message | Author |
 |------|---------|--------|
-| [982dda4](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/982dda4) | fix(browser): align list columns via subgrid and fix icon sizing | alexei.dolgolyov |
+| [d798fed](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/d798fed) | feat(icon): redesign app icon as "Beacon" and ship multi-resolution ICO | alexei.dolgolyov |
-| [eaeebb6](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/eaeebb6) | fix(csp): replace inline on* handlers with data-on* + JS wiring | alexei.dolgolyov |
+| [ddf4a6c](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/ddf4a6c) | feat: production-ready Linux & macOS support | alexei.dolgolyov |
-| [bcc6d40](https://git.dolgolyov-family.by/alexei.dolgolyov/media-player-server/commit/bcc6d40) | fix: comprehensive security, bug, performance, and UI/UX audit | alexei.dolgolyov |
 
 </details>

build-dist-linux.sh

@@ -16,12 +16,38 @@ BUILD_OUTPUT="build/MediaServer-v${VERSION_CLEAN}-linux-x64"
 clean_dist "${DIST_DIR}" build
 verify_frontend
 
+# --- Sanity check: native deps required for dbus-python + PyGObject ---
+# These compile from sdist on Linux. We don't try to install them via the
+# distro package manager here (the CI image must already have them); fail
+# with a clear message instead of cryptic gcc/pkg-config errors deep
+# inside pip.
+need_pkg() {
+    local pkg="$1"
+    if ! command -v pkg-config >/dev/null 2>&1; then
+        echo "ERROR: pkg-config is required (apt: pkg-config / dnf: pkgconf-pkg-config)" >&2
+        exit 1
+    fi
+    if ! pkg-config --exists "$pkg" 2>/dev/null; then
+        echo "ERROR: $pkg headers missing — install libdbus-1-dev libglib2.0-dev (apt)" >&2
+        echo "       or dbus-devel glib2-devel (dnf) before running this script." >&2
+        exit 1
+    fi
+}
+need_pkg dbus-1
+need_pkg glib-2.0
+
 # --- Create self-contained virtualenv ---
 echo "Creating virtualenv..."
 python3 -m venv "${DIST_DIR}/venv"
 source "${DIST_DIR}/venv/bin/activate"
 pip install --quiet --upgrade pip
-pip install --quiet "."
+# Install with the linux extra so dbus-python / PyGObject / python-xlib land
+# in the venv. Falls back to base deps if the extra is unsupported (shouldn't
+# happen post-0.3.2 but keep the fallback for older tags re-built here).
+if ! pip install --quiet ".[linux]"; then
+    echo "WARN: '.[linux]' extra unavailable; installing base deps only" >&2
+    pip install --quiet "."
+fi
 
 # Remove the installed package (app source is on PYTHONPATH via launcher)
 rm -rf "${DIST_DIR}"/venv/lib/python*/site-packages/media_server*
@@ -47,6 +73,11 @@ LAUNCHER
 chmod +x "${DIST_DIR}/media-server.sh"
 
 # --- Create systemd service installer ---
+# Emits a non-templated unit that runs as the invoking user with the env
+# bits MPRIS / PulseAudio need: DBUS_SESSION_BUS_ADDRESS and XDG_RUNTIME_DIR
+# must be set or DBus.SessionBus() either fails or talks to the wrong bus.
+# We also relax ProtectHome to read-write for ~/.config/media-server +
+# ~/.cache/media-server so audit.log and thumbnail cache writes succeed.
 cat > "${DIST_DIR}/install-service.sh" << 'SERVICE'
 #!/usr/bin/env bash
 set -euo pipefail
@@ -61,6 +92,9 @@ if [ "$EUID" -ne 0 ]; then
 fi
 
 REAL_USER="${SUDO_USER:-$USER}"
+USER_UID="$(id -u "$REAL_USER")"
+USER_HOME="$(getent passwd "$REAL_USER" | cut -d: -f6)"
+RUNTIME_DIR="/run/user/${USER_UID}"
 
 cat > "$SERVICE_FILE" << EOF
 [Unit]
@@ -74,17 +108,36 @@ WorkingDirectory=${SCRIPT_DIR}
 ExecStart=${SCRIPT_DIR}/media-server.sh
 Restart=on-failure
 RestartSec=5
+
+# Required by MPRIS (dbus.SessionBus) and PulseAudio/PipeWire loopback.
 Environment=PYTHONUNBUFFERED=1
+Environment=HOME=${USER_HOME}
+Environment=XDG_RUNTIME_DIR=${RUNTIME_DIR}
+Environment=DBUS_SESSION_BUS_ADDRESS=unix:path=${RUNTIME_DIR}/bus
+
+# Light sandboxing — keep ~/.config/media-server + ~/.cache/media-server
+# writable so audit.log / thumbnails / config CRUD continue to work.
+NoNewPrivileges=true
+PrivateTmp=true
+ProtectSystem=strict
+ProtectHome=read-only
+ReadWritePaths=${USER_HOME}/.config/media-server ${USER_HOME}/.cache/media-server
 
 [Install]
 WantedBy=multi-user.target
 EOF
 
+# Pre-create state dirs so ReadWritePaths actually has something to grant.
+sudo -u "$REAL_USER" mkdir -p \
+    "${USER_HOME}/.config/media-server" \
+    "${USER_HOME}/.cache/media-server"
+
 systemctl daemon-reload
 systemctl enable "${SERVICE_NAME}"
 systemctl start "${SERVICE_NAME}"
 echo "Service '${SERVICE_NAME}' installed and started."
 echo "Check status: systemctl status ${SERVICE_NAME}"
+echo "Tail logs:    journalctl -u ${SERVICE_NAME} -f"
 SERVICE
 chmod +x "${DIST_DIR}/install-service.sh"
 

build-dist-macos.sh

@@ -0,0 +1,142 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+# Build macOS distribution (self-contained venv + tarball)
+# Usage: ./build-dist-macos.sh [VERSION]
+#
+# Must be run on macOS (PyObjC wheels are platform-specific). For CI use
+# the github-hosted macos-latest runner.
+
+source "$(dirname "$0")/build-common.sh"
+
+if [ "$(uname -s)" != "Darwin" ]; then
+    echo "ERROR: build-dist-macos.sh must run on macOS (uname=$(uname -s))" >&2
+    exit 1
+fi
+
+detect_version "${1:-}"
+echo "Building Media Server v${VERSION_CLEAN} for macOS"
+
+# Detect host architecture for archive naming (arm64 = Apple Silicon, x86_64 = Intel).
+ARCH="$(uname -m)"
+DIST_DIR="dist/media-server"
+BUILD_OUTPUT="build/MediaServer-v${VERSION_CLEAN}-macos-${ARCH}"
+
+clean_dist "${DIST_DIR}" build
+verify_frontend
+
+# --- Create self-contained virtualenv ---
+echo "Creating virtualenv..."
+python3 -m venv "${DIST_DIR}/venv"
+# shellcheck disable=SC1091
+source "${DIST_DIR}/venv/bin/activate"
+pip install --quiet --upgrade pip
+if ! pip install --quiet ".[macos]"; then
+    echo "WARN: '.[macos]' extra unavailable; installing base deps only" >&2
+    pip install --quiet "."
+fi
+
+# Remove the installed package (app source is on PYTHONPATH via launcher)
+rm -rf "${DIST_DIR}"/venv/lib/python*/site-packages/media_server*
+rm -rf "${DIST_DIR}"/venv/lib/python*/site-packages/media_server*.dist-info
+
+deactivate
+
+# Trim venv site-packages — macOS native ext is .so, dylibs are .dylib
+MAC_SP=$(echo "${DIST_DIR}"/venv/lib/python*/site-packages)
+cleanup_site_packages "$MAC_SP" "so" "dylib"
+
+copy_app_files "$DIST_DIR"
+
+# --- Launcher ---
+cat > "${DIST_DIR}/media-server.sh" << 'LAUNCHER'
+#!/usr/bin/env bash
+set -euo pipefail
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
+export PYTHONPATH="$SCRIPT_DIR/app"
+# shellcheck disable=SC1091
+source "$SCRIPT_DIR/venv/bin/activate"
+exec python -m media_server.main "$@"
+LAUNCHER
+chmod +x "${DIST_DIR}/media-server.sh"
+
+# --- LaunchAgent installer ---
+# LaunchAgents run as the user, with the user's GUI session — exactly what
+# we want for AppleScript / Music.app / Spotify control. KeepAlive auto-
+# restarts on crash; RunAtLoad starts at login.
+cat > "${DIST_DIR}/install-launchagent.sh" << 'LAUNCHAGENT'
+#!/usr/bin/env bash
+set -euo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
+LABEL="com.dolgolyov.media-server"
+PLIST_DIR="${HOME}/Library/LaunchAgents"
+PLIST="${PLIST_DIR}/${LABEL}.plist"
+LOG_DIR="${HOME}/Library/Logs/media-server"
+
+mkdir -p "$PLIST_DIR" "$LOG_DIR"
+
+cat > "$PLIST" << EOF
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
+<plist version="1.0">
+<dict>
+    <key>Label</key>
+    <string>${LABEL}</string>
+    <key>ProgramArguments</key>
+    <array>
+        <string>${SCRIPT_DIR}/media-server.sh</string>
+        <string>--no-tray</string>
+    </array>
+    <key>WorkingDirectory</key>
+    <string>${SCRIPT_DIR}</string>
+    <key>RunAtLoad</key>
+    <true/>
+    <key>KeepAlive</key>
+    <dict>
+        <key>SuccessfulExit</key>
+        <false/>
+    </dict>
+    <key>StandardOutPath</key>
+    <string>${LOG_DIR}/stdout.log</string>
+    <key>StandardErrorPath</key>
+    <string>${LOG_DIR}/stderr.log</string>
+    <key>EnvironmentVariables</key>
+    <dict>
+        <key>PYTHONUNBUFFERED</key>
+        <string>1</string>
+    </dict>
+</dict>
+</plist>
+EOF
+
+# Reload if already loaded, then load fresh.
+launchctl unload "$PLIST" 2>/dev/null || true
+launchctl load -w "$PLIST"
+
+echo "LaunchAgent '${LABEL}' installed and started."
+echo "Plist:  $PLIST"
+echo "Logs:   $LOG_DIR/{stdout,stderr}.log"
+echo "Stop:   launchctl unload \"$PLIST\""
+echo "Start:  launchctl load -w \"$PLIST\""
+LAUNCHAGENT
+chmod +x "${DIST_DIR}/install-launchagent.sh"
+
+# Convenience uninstaller.
+cat > "${DIST_DIR}/uninstall-launchagent.sh" << 'UNINSTALL'
+#!/usr/bin/env bash
+set -euo pipefail
+LABEL="com.dolgolyov.media-server"
+PLIST="${HOME}/Library/LaunchAgents/${LABEL}.plist"
+launchctl unload "$PLIST" 2>/dev/null || true
+rm -f "$PLIST"
+echo "LaunchAgent '${LABEL}' removed (config preserved under ~/.config/media-server)."
+UNINSTALL
+chmod +x "${DIST_DIR}/uninstall-launchagent.sh"
+
+# --- Package ---
+echo "Creating archive..."
+cp -r "${DIST_DIR}" "${BUILD_OUTPUT}"
+tar -czf "${BUILD_OUTPUT}.tar.gz" -C build "MediaServer-v${VERSION_CLEAN}-macos-${ARCH}"
+
+echo "Build complete: ${BUILD_OUTPUT}.tar.gz"

config.example.yaml

@@ -1,7 +1,13 @@
 # Media Server Configuration
 # Copy this file to config.yaml and customize as needed.
-# By default, authentication is DISABLED (no tokens = open access).
+#
-# To enable auth, uncomment and configure the api_tokens section below.
+# Secure-by-default: the server binds to loopback (127.0.0.1) only and refuses
+# to bind a non-loopback address with no tokens configured.
+#
+# To expose on the LAN you must do ONE of:
+#   1. Configure api_tokens below AND change host to "0.0.0.0", OR
+#   2. Set `allow_lan_without_auth: true` (LAN-open, no auth — insecure on
+#      hostile networks, only acceptable on a trusted home LAN).
 
 # API Tokens - Multiple tokens with friendly labels
 # This allows you to identify which client is making requests in the logs
@@ -11,111 +17,143 @@
 #   web_ui: "your-web-ui-token-here"
 
 # Server settings
-host: "0.0.0.0"
+host: "127.0.0.1"
 port: 8765
+# allow_lan_without_auth: true   # uncomment + change host to 0.0.0.0 for LAN-open mode
+
+# ─── Custom scripts ─────────────────────────────────────────────────────
+#
+# Examples below are platform-specific. Uncomment the block that matches
+# your OS — YAML keys must be unique, so don't ship multiple
+# `lock_screen:` entries.
 
-# Custom scripts
 scripts:
-  lock_screen:
+  # ── Windows ─────────────────────────────────────────────────────────
-    command: "rundll32.exe user32.dll,LockWorkStation"
+  # lock_screen:
-    label: "Lock Screen"
+  #   command: "rundll32.exe user32.dll,LockWorkStation"
-    description: "Lock the workstation"
+  #   label: "Lock Screen"
-    icon: "mdi:lock"
+  #   description: "Lock the workstation"
-    timeout: 5
+  #   icon: "mdi:lock"
-    shell: true
+  #   timeout: 5
+  #   shell: true
+  # sleep:
+  #   command: "rundll32.exe powrprof.dll,SetSuspendState 0,1,0"
+  #   label: "Sleep"
+  #   icon: "mdi:sleep"
+  #   timeout: 10
+  #   shell: true
+  # shutdown:
+  #   command: "shutdown /s /t 0"
+  #   label: "Shutdown"
+  #   icon: "mdi:power"
+  #   timeout: 10
+  #   shell: true
 
-  hibernate:
+  # ── Linux (systemd / xdg) ───────────────────────────────────────────
-    command: "shutdown /h"
+  # lock_screen:
-    label: "Hibernate"
+  #   command: "loginctl lock-session"   # or: xdg-screensaver lock
-    description: "Hibernate the PC"
+  #   label: "Lock Screen"
-    icon: "mdi:power-sleep"
+  #   icon: "mdi:lock"
+  #   timeout: 5
+  #   shell: true
+  # sleep:
+  #   command: "systemctl suspend"
+  #   label: "Sleep"
+  #   icon: "mdi:sleep"
+  #   timeout: 10
+  #   shell: true
+  # shutdown:
+  #   command: "systemctl poweroff"
+  #   label: "Shutdown"
+  #   icon: "mdi:power"
+  #   timeout: 10
+  #   shell: true
+
+  # ── macOS ───────────────────────────────────────────────────────────
+  # lock_screen:
+  #   command: "pmset displaysleepnow"
+  #   label: "Lock Screen"
+  #   icon: "mdi:lock"
+  #   timeout: 5
+  #   shell: true
+  # sleep:
+  #   command: "pmset sleepnow"
+  #   label: "Sleep"
+  #   icon: "mdi:sleep"
+  #   timeout: 5
+  #   shell: true
+  # shutdown:
+  #   command: "osascript -e 'tell application \"System Events\" to shut down'"
+  #   label: "Shutdown"
+  #   icon: "mdi:power"
+  #   timeout: 10
+  #   shell: true
+
+  # Cross-platform smoke test (always defined so first-run users see the UI populate).
+  example_script:
+    command: "echo Hello from Media Server!"
+    description: "Example script - echoes a message"
     timeout: 10
     shell: true
 
-  sleep:
+# Media folder management from Web UI (default: false)
-    command: "rundll32.exe powrprof.dll,SetSuspendState 0,1,0"
-    label: "Sleep"
-    description: "Put PC to sleep"
-    icon: "mdi:sleep"
-    timeout: 10
-    shell: true
-
-  shutdown:
-    command: "shutdown /s /t 0"
-    label: "Shutdown"
-    description: "Shutdown the PC immediately"
-    icon: "mdi:power"
-    timeout: 10
-    shell: true
-
-  restart:
-    command: "shutdown /r /t 0"
-    label: "Restart"
-    description: "Restart the PC immediately"
-    icon: "mdi:restart"
-    timeout: 10
-    shell: true
-
-# Media folder management from Web UI (default: true)
 # When enabled, media folders can be added, edited, and deleted from the Settings tab.
-# Set to false to disable folder management from the UI.
+# media_folders_management: true
-# media_folders_management: false
+
+# ─── Callback scripts (executed by integration events) ──────────────────
+#
+# Callbacks are OS-agnostic — if you want side effects, fill in commands
+# below. The defaults are no-op echos so first-run logs don't error out.
 
-# Callback scripts (executed after media actions)
-# All callbacks are optional - if not defined, the action runs without callback
 callbacks:
-  # Media control callbacks (run after successful action)
   on_play:
     command: "echo Play triggered"
     timeout: 10
     shell: true
-
   on_pause:
     command: "echo Pause triggered"
     timeout: 10
     shell: true
-
   on_stop:
     command: "echo Stop triggered"
     timeout: 10
     shell: true
-
   on_next:
     command: "echo Next track"
     timeout: 10
     shell: true
-
   on_previous:
     command: "echo Previous track"
     timeout: 10
     shell: true
-
   on_volume:
     command: "echo Volume changed"
     timeout: 10
     shell: true
-
   on_mute:
     command: "echo Mute toggled"
     timeout: 10
     shell: true
-
   on_seek:
     command: "echo Seek triggered"
     timeout: 10
     shell: true
 
-  # Turn on/off/toggle (callback-only actions, no default behavior)
+  # Turn on/off/toggle (callback-only actions, no default behavior).
+  # The Windows-only example used to ship as the default for `on_turn_off`,
+  # which silently failed on Linux/macOS. Defaults are now no-ops — pick a
+  # platform-appropriate command below.
   on_turn_on:
     command: "echo Turn on callback"
     timeout: 10
     shell: true
-
   on_turn_off:
-    command: "rundll32.exe user32.dll,LockWorkStation"
+    # Windows: "rundll32.exe user32.dll,LockWorkStation"
+    # Linux:   "loginctl lock-session"
+    # macOS:   "pmset displaysleepnow"
+    command: "echo Turn off callback"
     timeout: 5
     shell: true
-
   on_toggle:
     command: "echo Toggle callback"
     timeout: 10

media_server/auth.py

@@ -13,6 +13,8 @@ security = HTTPBearer(auto_error=False)
 
 # Context variable to store current request's token label
 token_label_var: ContextVar[str] = ContextVar("token_label", default="unknown")
+# Per-request correlation ID — generated in middleware if upstream didn't send one.
+request_id_var: ContextVar[str] = ContextVar("request_id", default="-")
 
 
 def auth_enabled() -> bool:
@@ -29,12 +31,42 @@ def get_token_label(token: str) -> Optional[str]:
     Returns:
         The label for the token, or None if invalid
     """
-    for label, stored_token in settings.api_tokens.items():
+    for label, spec in settings.api_tokens.items():
-        if secrets.compare_digest(stored_token, token):
+        if secrets.compare_digest(spec.token, token):
             return label
     return None
 
 
+def token_has_scope(label: str, required: str) -> bool:
+    """Whether the token identified by `label` grants `required` scope."""
+    spec = settings.api_tokens.get(label)
+    if spec is None:
+        # Unknown label = no auth or anonymous; treat as full access only
+        # when auth is disabled entirely (matches existing behaviour).
+        return not auth_enabled()
+    return spec.grants(required)
+
+
+def require_scope(scope: str):
+    """Build a FastAPI dependency that enforces the given scope.
+
+    Use as ``Depends(require_scope("admin"))`` on management endpoints. When
+    auth is disabled the dependency is a no-op (anonymous access).
+    """
+
+    async def _checker(label: str = Depends(verify_token)) -> str:
+        if not auth_enabled():
+            return label
+        if not token_has_scope(label, scope):
+            raise HTTPException(
+                status_code=status.HTTP_403_FORBIDDEN,
+                detail=f"Token '{label}' lacks required scope: {scope}",
+            )
+        return label
+
+    return _checker
+
+
 async def verify_token(
     request: Request,
     credentials: HTTPAuthorizationCredentials = Depends(security),

media_server/config.py

@@ -7,12 +7,49 @@ from pathlib import Path
 from typing import Optional
 
 import yaml
-from pydantic import BaseModel, Field
+from pydantic import BaseModel, Field, field_validator
 from pydantic_settings import BaseSettings, SettingsConfigDict
 
 logger = logging.getLogger(__name__)
 
 
+# Token scopes form a strict hierarchy: admin > control > read. Helper utility
+# used by both auth.py and the validator below.
+SCOPE_HIERARCHY: dict[str, frozenset[str]] = {
+    "read": frozenset({"read"}),
+    "control": frozenset({"read", "control"}),
+    "admin": frozenset({"read", "control", "admin"}),
+}
+ALL_SCOPES: frozenset[str] = frozenset(SCOPE_HIERARCHY.keys())
+
+
+class TokenSpec(BaseModel):
+    """Per-token authentication entry with explicit scopes."""
+
+    token: str = Field(..., min_length=8, description="Secret token value")
+    scopes: list[str] = Field(
+        default_factory=lambda: ["admin"],
+        description="Granted scopes (subset of read|control|admin).",
+    )
+
+    @field_validator("scopes")
+    @classmethod
+    def _validate_scopes(cls, v: list[str]) -> list[str]:
+        if not v:
+            raise ValueError("scopes must list at least one of read|control|admin")
+        unknown = set(v) - ALL_SCOPES
+        if unknown:
+            raise ValueError(f"unknown scopes: {sorted(unknown)}; valid={sorted(ALL_SCOPES)}")
+        return v
+
+    def grants(self, required: str) -> bool:
+        """Whether this token grants the requested scope (with hierarchy expansion)."""
+        granted: set[str] = set()
+        for s in self.scopes:
+            granted |= SCOPE_HIERARCHY.get(s, frozenset({s}))
+        return required in granted
+
+
 class MediaFolderConfig(BaseModel):
     """Configuration for a media folder."""
 
@@ -48,6 +85,13 @@ class ScriptParameterConfig(BaseModel):
     options: Optional[list[str]] = Field(
         default=None, description="Allowed values (select type only)"
     )
+    pattern: Optional[str] = Field(
+        default=None,
+        description=(
+            "Optional regex (Python flavour) that string-typed values must match."
+            " Use to harden parameters that flow into shell=true scripts."
+        ),
+    )
 
 
 class ScriptConfig(BaseModel):
@@ -108,19 +152,84 @@ class Settings(BaseSettings):
         ),
     )
 
+    # Reverse-proxy deployment: when serving the API behind nginx/Caddy/Traefik,
+    # uvicorn must trust the X-Forwarded-* headers from the proxy so that the
+    # `Origin` allow-list, request URLs, and logs reflect the public-facing
+    # values. Off by default — only enable when there's a real proxy in front
+    # (otherwise clients can spoof their own IP).
+    proxy_headers: bool = Field(
+        default=False,
+        description="Honor X-Forwarded-For / X-Forwarded-Proto from upstream proxy.",
+    )
+    forwarded_allow_ips: str = Field(
+        default="127.0.0.1",
+        description=(
+            "Comma-separated IPs / CIDRs that uvicorn should trust X-Forwarded-* from."
+            " Use '*' to trust all (only safe when bound to a private interface)."
+        ),
+    )
+
+    # HTTPS / TLS. Both must be set together to enable TLS; if only one is set
+    # the server refuses to start. Use `mkcert` or letsencrypt to generate the
+    # pair; the server reads them at startup.
+    ssl_certfile: Optional[str] = Field(
+        default=None,
+        description="Path to TLS certificate (PEM). Pair with ssl_keyfile.",
+    )
+    ssl_keyfile: Optional[str] = Field(
+        default=None,
+        description="Path to TLS private key (PEM). Pair with ssl_certfile.",
+    )
+    ssl_keyfile_password: Optional[str] = Field(
+        default=None,
+        description="Optional password for the private key if encrypted.",
+    )
+
     # Admin-grade operations (script / callback / link / folder create/update/delete).
     # When True the same token used for read/play can also persist arbitrary shell
-    # commands. Disable to make the API read+execute only.
+    # commands. Default False so a single leaked token cannot escalate to RCE; opt
-    scripts_management: bool = Field(default=True, description="Allow scripts CRUD via API")
+    # in explicitly to manage scripts/callbacks/links via the Web UI.
-    callbacks_management: bool = Field(default=True, description="Allow callbacks CRUD via API")
+    scripts_management: bool = Field(default=False, description="Allow scripts CRUD via API")
-    links_management: bool = Field(default=True, description="Allow links CRUD via API")
+    callbacks_management: bool = Field(default=False, description="Allow callbacks CRUD via API")
+    links_management: bool = Field(default=False, description="Allow links CRUD via API")
 
-    # Authentication (empty = auth disabled, anyone can access the API)
+    # Authentication (empty = auth disabled, anyone can access the API).
-    api_tokens: dict[str, str] = Field(
+    #
+    # Each entry can be either:
+    #   • a bare string (legacy form, treated as scopes = ["admin"] for back-compat), OR
+    #   • a mapping with explicit scopes, e.g.
+    #         "ha":     {token: "<token>", scopes: ["read", "control"]}
+    #         "kiosk":  {token: "<token>", scopes: ["read"]}
+    #         "ops":    {token: "<token>", scopes: ["admin"]}
+    #
+    # Available scopes:
+    #   read    — GET /api/* (status, list, browse) but no state-changing calls.
+    #   control — read + media transport, display/audio, script EXECUTE, callback EXECUTE.
+    #   admin   — control + CRUD on scripts/callbacks/links/folders.
+    #
+    # Validation normalises both forms to TokenSpec at load time.
+    api_tokens: dict[str, TokenSpec] = Field(
         default_factory=dict,
-        description="Named API tokens for access control (label: token pairs). Empty = no auth.",
+        description=(
+            "Named API tokens. Value can be a bare token string (= admin scope) or"
+            " a {token, scopes} mapping. See TokenSpec for scope definitions."
+        ),
     )
 
+    @field_validator("api_tokens", mode="before")
+    @classmethod
+    def _normalise_tokens(cls, v):
+        """Accept legacy `label: <bare-token>` form and promote to TokenSpec."""
+        if not isinstance(v, dict):
+            return v
+        out: dict[str, dict | TokenSpec] = {}
+        for label, entry in v.items():
+            if isinstance(entry, str):
+                out[label] = {"token": entry, "scopes": ["admin"]}
+            else:
+                out[label] = entry
+        return out
+
     # Media controller settings
     poll_interval: float = Field(
         default=1.0, description="Media status poll interval in seconds"
@@ -156,7 +265,7 @@ class Settings(BaseSettings):
         description="Media folders available for browsing in the media browser",
     )
     media_folders_management: bool = Field(
-        default=True,
+        default=False,
         description="Allow adding, editing, and deleting media folders from the Web UI",
     )
 
@@ -263,8 +372,11 @@ def generate_default_config(path: Optional[Path] = None) -> Path:
     config = {
         "host": "127.0.0.1",
         "port": 8765,
+        # Default token grants "admin" scope (full access). To create a
+        # read-only or control-only token, add a second entry:
+        #   ha_readonly: {token: "<token>", scopes: ["read"]}
         "api_tokens": {
-            "default": default_token,
+            "default": {"token": default_token, "scopes": ["admin"]},
         },
         "poll_interval": 1.0,
         "log_level": "INFO",
@@ -298,8 +410,16 @@ def _write_yaml_atomic(path: Path, data: dict) -> None:
 
 
 def _restrict_config_perms(path: Path) -> None:
-    """On POSIX, ensure config file is readable only by owner (0600)."""
+    """Ensure config file is readable only by its owner.
+
+    POSIX → ``chmod 0600``. On Windows the default NTFS ACL leaves the file
+    readable by every interactive user on the machine (Users group has Read),
+    which is bad given the file stores plaintext API tokens. Use ``icacls`` to
+    grant exclusive access to the current user + SYSTEM + Administrators and
+    strip inheritance.
+    """
     if os.name == "nt":
+        _restrict_config_perms_windows(path)
         return
     try:
         os.chmod(path, 0o600)
@@ -308,5 +428,31 @@ def _restrict_config_perms(path: Path) -> None:
         logger.debug("Could not chmod %s", path, exc_info=True)
 
 
+def _restrict_config_perms_windows(path: Path) -> None:
+    """Apply restrictive NTFS ACL to a config file (Windows only)."""
+    import subprocess
+
+    try:
+        username = os.environ.get("USERNAME") or os.environ.get("USER")
+        if not username:
+            logger.debug("Cannot detect current user; skipping icacls hardening")
+            return
+        # Disable inheritance and remove every existing ACE, then grant access
+        # only to current user, SYSTEM, and Administrators. /Q suppresses
+        # progress output; /C lets per-file errors not abort the batch.
+        subprocess.run(
+            ["icacls", str(path), "/inheritance:r"],
+            check=False, capture_output=True, timeout=5,
+        )
+        for principal in (username, "SYSTEM", "Administrators"):
+            subprocess.run(
+                ["icacls", str(path), "/grant:r", f"{principal}:(R,W)"],
+                check=False, capture_output=True, timeout=5,
+            )
+    except (FileNotFoundError, subprocess.TimeoutExpired, OSError):
+        # `icacls` missing or sandboxed — leave the default ACL in place.
+        logger.debug("icacls hardening failed for %s", path, exc_info=True)
+
+
 # Global settings instance
 settings = Settings.load_from_yaml()

media_server/main.py

@@ -15,13 +15,14 @@ from fastapi.responses import FileResponse
 from fastapi.staticfiles import StaticFiles
 
 from . import __version__
-from .auth import get_token_label, token_label_var
+from .auth import get_token_label, request_id_var, token_label_var
 from .config import generate_default_config, get_config_dir, settings
 from .routes import (
     audio_router,
     browser_router,
     callbacks_router,
     display_router,
+    foreground_router,
     health_router,
     links_router,
     media_router,
@@ -32,10 +33,34 @@ from .services.websocket_manager import ws_manager
 
 
 class TokenLabelFilter(logging.Filter):
-    """Add token label to log records."""
+    """Add token label + request_id to log records."""
 
     def filter(self, record):
         record.token_label = token_label_var.get("unknown")
+        record.request_id = request_id_var.get("-")
+        return True
+
+
+class _StripTokenQueryFilter(logging.Filter):
+    """Strip `token=...` from query strings before they hit the access log.
+
+    uvicorn's default access log format includes the full request line, so
+    `/api/media/artwork?token=SECRET` would otherwise be persisted verbatim
+    in stdout/journald/file sinks.
+    """
+
+    import re as _re
+
+    _TOKEN_RE = _re.compile(r"([?&])token=[^&\s\"']+")
+
+    def filter(self, record):  # type: ignore[override]
+        if isinstance(record.args, tuple):
+            record.args = tuple(
+                self._TOKEN_RE.sub(r"\1token=REDACTED", a) if isinstance(a, str) else a
+                for a in record.args
+            )
+        if isinstance(record.msg, str) and "token=" in record.msg:
+            record.msg = self._TOKEN_RE.sub(r"\1token=REDACTED", record.msg)
         return True
 
 
@@ -48,17 +73,34 @@ def setup_logging():
 
     logging.basicConfig(
         level=getattr(logging, settings.log_level.upper()),
-        format="%(asctime)s - %(name)s - [%(token_label)s] - %(levelname)s - %(message)s",
+        format=(
+            "%(asctime)s - %(name)s - [%(token_label)s] [%(request_id)s]"
+            " - %(levelname)s - %(message)s"
+        ),
         handlers=[handler],
     )
 
     # Suppress noisy third-party loggers
     logging.getLogger("screen_brightness_control").setLevel(logging.ERROR)
 
+    # Make sure the uvicorn access log never persists tokens leaked into the
+    # query string (the artwork + WS endpoints accept `?token=` for browser
+    # compatibility — see verify_token_or_query).
+    strip_filter = _StripTokenQueryFilter()
+    for name in ("uvicorn.access", "uvicorn"):
+        logging.getLogger(name).addFilter(strip_filter)
+
 
 @asynccontextmanager
 async def lifespan(app: FastAPI):
-    """Application lifespan handler."""
+    """Application lifespan handler.
+
+    All long-lived resources started during startup are kept in local refs and
+    torn down in a `finally:` so a partial-startup failure cannot orphan tasks
+    or thread pools.
+    """
+    import asyncio
+
     setup_logging()
     logger = logging.getLogger(__name__)
     logger.info(f"Media Server starting on {settings.host}:{settings.port}")
@@ -70,92 +112,149 @@ async def lifespan(app: FastAPI):
     else:
         logger.warning("No API tokens configured — authentication is DISABLED")
 
-    # Start WebSocket status monitor
+    # Linux preflight: most MPRIS / PulseAudio failures are environmental
-    controller = get_media_controller()
+    # (no DBUS_SESSION_BUS_ADDRESS, missing XDG_RUNTIME_DIR, systemd service
-    await ws_manager.start_status_monitor(controller.get_status)
+    # started before logind). Surface that early so the failure mode is a
-    logger.info("WebSocket status monitor started")
+    # warning at boot instead of silent "/api/media/status returns idle".
-
+    import os
-    # Start update checker
-    update_checker = None
-    if settings.update_check_enabled:
-        from .services.gitea_release_provider import GiteaReleaseProvider
-        from .services.update_checker import UpdateChecker
-
-        provider = GiteaReleaseProvider()
-        update_checker = UpdateChecker(provider, __version__)
-        await update_checker.start(settings.update_check_interval)
-        # Store globally so health endpoint can access cached result
-        app.state.update_checker = update_checker
-
-    # Schedule periodic thumbnail cache cleanup so the 500 MB cap is actually
-    # enforced. Runs once at startup and then hourly until shutdown.
-    from .services.thumbnail_service import ThumbnailService
-
-    async def _thumbnail_cleanup_loop() -> None:
-        while True:
-            try:
-                await asyncio.to_thread(ThumbnailService.cleanup_cache)
-            except Exception as e:
-                logger.warning("Thumbnail cache cleanup failed: %s", e)
-            try:
-                await asyncio.sleep(3600)
-            except asyncio.CancelledError:
-                break
-
-    import asyncio
-    cleanup_task = asyncio.create_task(_thumbnail_cleanup_loop())
-
-    # Register audio visualizer (capture starts on-demand when clients subscribe)
-    analyzer = None
-    if settings.visualizer_enabled:
-        from .services.audio_analyzer import get_audio_analyzer
-
-        analyzer = get_audio_analyzer(
-            num_bins=settings.visualizer_bins,
-            target_fps=settings.visualizer_fps,
-            device_name=settings.visualizer_device,
-        )
-        if analyzer.available:
-            await ws_manager.start_audio_monitor(analyzer)
-            logger.info("Audio visualizer available (capture on-demand)")
-        else:
-            logger.info("Audio visualizer unavailable (install soundcard + numpy)")
-
-    yield
-
-    # Stop update checker
-    if update_checker is not None:
-        await update_checker.stop()
-
-    # Cancel periodic thumbnail cleanup
-    cleanup_task.cancel()
-    try:
-        await cleanup_task
-    except asyncio.CancelledError:
-        pass
-
-    # Stop audio visualizer
-    await ws_manager.stop_audio_monitor()
-    if analyzer and analyzer.running:
-        analyzer.stop()
-
-    # Stop WebSocket status monitor
-    await ws_manager.stop_status_monitor()
-
-    # Shut down dedicated thread pools so pending scripts don't leak threads
-    from .routes.callbacks import shutdown_callback_executor
-    from .routes.scripts import shutdown_script_executor
-
-    shutdown_script_executor()
-    shutdown_callback_executor()
-
-    # Clean up platform-specific resources
     import platform as _platform
-    if _platform.system() == "Windows":
+    if _platform.system() == "Linux":
-        from .services.windows_media import shutdown_executor
+        missing = [
-        shutdown_executor()
+            v for v in ("DBUS_SESSION_BUS_ADDRESS", "XDG_RUNTIME_DIR")
+            if not os.environ.get(v)
+        ]
+        if missing:
+            logger.warning(
+                "Linux preflight: %s not set — MPRIS / PulseAudio may be unavailable."
+                " Under systemd, run `loginctl enable-linger <user>` and ensure the"
+                " service unit sets DBUS_SESSION_BUS_ADDRESS + XDG_RUNTIME_DIR.",
+                ", ".join(missing),
+            )
+        if os.environ.get("WAYLAND_DISPLAY"):
+            logger.info(
+                "Wayland session detected — foreground-window probe is intentionally"
+                " disabled (Wayland hides window info from unprivileged clients)."
+            )
 
-    logger.info("Media Server shutting down")
+    update_checker = None
+    cleanup_task: asyncio.Task | None = None
+    analyzer = None
+    status_monitor_started = False
+
+    try:
+        # Start WebSocket status monitor
+        controller = get_media_controller()
+        await ws_manager.start_status_monitor(controller.get_status)
+        status_monitor_started = True
+        logger.info("WebSocket status monitor started")
+
+        # Start update checker
+        if settings.update_check_enabled:
+            from .services.gitea_release_provider import GiteaReleaseProvider
+            from .services.update_checker import UpdateChecker
+
+            provider = GiteaReleaseProvider()
+            update_checker = UpdateChecker(provider, __version__)
+            await update_checker.start(settings.update_check_interval)
+            # Store globally so health endpoint can access cached result
+            app.state.update_checker = update_checker
+
+        # Schedule periodic thumbnail cache cleanup so the 500 MB cap is actually
+        # enforced. Runs once at startup and then hourly until shutdown.
+        from .services.thumbnail_service import ThumbnailService
+
+        async def _thumbnail_cleanup_loop() -> None:
+            while True:
+                try:
+                    await asyncio.to_thread(ThumbnailService.cleanup_cache)
+                except Exception as e:
+                    logger.warning("Thumbnail cache cleanup failed: %s", e)
+                try:
+                    await asyncio.sleep(3600)
+                except asyncio.CancelledError:
+                    break
+
+        cleanup_task = asyncio.create_task(_thumbnail_cleanup_loop())
+
+        # Register audio visualizer (capture starts on-demand when clients subscribe)
+        if settings.visualizer_enabled:
+            from .services.audio_analyzer import get_audio_analyzer
+
+            analyzer = get_audio_analyzer(
+                num_bins=settings.visualizer_bins,
+                target_fps=settings.visualizer_fps,
+                device_name=settings.visualizer_device,
+            )
+            if analyzer.available:
+                await ws_manager.start_audio_monitor(analyzer)
+                logger.info("Audio visualizer available (capture on-demand)")
+            else:
+                logger.info("Audio visualizer unavailable (install soundcard + numpy)")
+
+        yield
+    finally:
+        # Stop update checker
+        if update_checker is not None:
+            try:
+                await update_checker.stop()
+            except Exception:
+                logger.exception("Error stopping update checker")
+
+        # Cancel periodic thumbnail cleanup
+        if cleanup_task is not None:
+            cleanup_task.cancel()
+            try:
+                await cleanup_task
+            except asyncio.CancelledError:
+                pass
+            except Exception:
+                logger.exception("Error awaiting thumbnail cleanup task")
+
+        # Stop audio visualizer
+        try:
+            await ws_manager.stop_audio_monitor()
+        except Exception:
+            logger.exception("Error stopping audio monitor")
+        if analyzer and analyzer.running:
+            try:
+                analyzer.stop()
+            except Exception:
+                logger.exception("Error stopping audio analyzer")
+
+        # Stop WebSocket status monitor
+        if status_monitor_started:
+            try:
+                await ws_manager.stop_status_monitor()
+            except Exception:
+                logger.exception("Error stopping status monitor")
+
+        # Shut down dedicated thread pools so pending scripts don't leak threads
+        try:
+            from .routes.callbacks import shutdown_callback_executor
+            from .routes.scripts import shutdown_script_executor
+
+            shutdown_script_executor()
+            shutdown_callback_executor()
+        except Exception:
+            logger.exception("Error shutting down script/callback executors")
+
+        # Flush audit log writer
+        try:
+            from .services.audit_log import shutdown_audit_log
+            shutdown_audit_log()
+        except Exception:
+            logger.exception("Error flushing audit log")
+
+        # Clean up platform-specific resources
+        import platform as _platform
+        if _platform.system() == "Windows":
+            try:
+                from .services.windows_media import shutdown_executor
+                shutdown_executor()
+            except Exception:
+                logger.exception("Error shutting down windows_media executor")
+
+        logger.info("Media Server shutting down")
 
 
 def create_app() -> FastAPI:
@@ -172,7 +271,15 @@ def create_app() -> FastAPI:
 
     # CORS — restrict to same-origin by default; users that integrate the API
     # from another origin (e.g. Home Assistant on a different host) can set
-    # cors_origins in config.yaml.
+    # cors_origins in config.yaml. Refuse "*" outright: combined with the
+    # admin endpoints this would let any origin in the universe run
+    # arbitrary shell. If users genuinely need every origin, they can list
+    # them explicitly.
+    if any(o.strip() == "*" for o in settings.cors_origins):
+        raise RuntimeError(
+            "cors_origins must not contain '*' — list exact origins instead. "
+            "This protects the script-execution endpoints from any-origin abuse."
+        )
     cors_origins = settings.cors_origins or [
         f"http://localhost:{settings.port}",
         f"http://127.0.0.1:{settings.port}",
@@ -185,6 +292,23 @@ def create_app() -> FastAPI:
         allow_headers=["Authorization", "Content-Type"],
     )
 
+    # Request correlation ID — accept upstream X-Request-ID if it's a sane
+    # ASCII id, otherwise mint a fresh UUID4. Emitted on the response so
+    # clients can quote it back in bug reports.
+    import re
+    import uuid as _uuid
+
+    _REQ_ID_RE = re.compile(r"^[A-Za-z0-9._\-]{1,128}$")
+
+    @app.middleware("http")
+    async def request_id_middleware(request: Request, call_next):
+        incoming = request.headers.get("x-request-id", "")
+        req_id = incoming if _REQ_ID_RE.match(incoming) else _uuid.uuid4().hex[:16]
+        request_id_var.set(req_id)
+        response = await call_next(request)
+        response.headers["X-Request-ID"] = req_id
+        return response
+
     # Security headers — strict CSP for the bundled UI, disallow framing, hide referrer.
     @app.middleware("http")
     async def security_headers_middleware(request: Request, call_next):
@@ -199,6 +323,9 @@ def create_app() -> FastAPI:
                 "style-src 'self' 'unsafe-inline'; "
                 "font-src 'self' data:; "
                 "frame-ancestors 'none'; "
+                "form-action 'self'; "
+                "worker-src 'self'; "
+                "manifest-src 'self'; "
                 "base-uri 'self'"
             ),
         )
@@ -207,32 +334,63 @@ def create_app() -> FastAPI:
         response.headers.setdefault("Referrer-Policy", "no-referrer")
         return response
 
-    # Add token logging middleware
+    # Add token logging middleware + auth-failure rate limit
+    from fastapi.responses import JSONResponse
+
+    from .services.rate_limit import check as ratelimit_check
+    from .services.rate_limit import get_peer
+
     @app.middleware("http")
     async def token_logging_middleware(request: Request, call_next):
-        """Extract token label and set in context for logging."""
+        """Extract token label, set in context, and rate-limit failed auths."""
         if not settings.api_tokens:
             token_label_var.set("anonymous")
         else:
             token_label = "unknown"
+            token_present = False
+            token_valid = False
 
             # Try Authorization header
             auth_header = request.headers.get("authorization", "")
             if auth_header.startswith("Bearer "):
+                token_present = True
                 token = auth_header[7:]
                 label = get_token_label(token)
                 if label:
                     token_label = label
+                    token_valid = True
 
             # Try query parameter (for artwork endpoint)
             elif "token" in request.query_params:
+                token_present = True
                 token = request.query_params["token"]
                 label = get_token_label(token)
                 if label:
                     token_label = label
+                    token_valid = True
 
             token_label_var.set(token_label)
 
+            # Brute-force gate: a peer that produces a wrong/missing token gets
+            # 5 failures per minute before being throttled. Static-asset
+            # requests (GET /static/*, /, /sw.js) and the docs endpoint are
+            # exempt — they're served unauthenticated by design.
+            if token_present and not token_valid:
+                path = request.url.path
+                if not (
+                    path == "/" or path == "/sw.js"
+                    or path.startswith("/static/")
+                    or path.startswith("/docs") or path.startswith("/openapi")
+                    or path.startswith("/redoc")
+                ):
+                    allowed, retry_after = ratelimit_check("auth", get_peer(request))
+                    if not allowed:
+                        return JSONResponse(
+                            status_code=429,
+                            content={"detail": "Too many authentication failures"},
+                            headers={"Retry-After": str(int(retry_after or 60))},
+                        )
+
         response = await call_next(request)
         return response
 
@@ -241,6 +399,7 @@ def create_app() -> FastAPI:
     app.include_router(browser_router)
     app.include_router(callbacks_router)
     app.include_router(display_router)
+    app.include_router(foreground_router)
     app.include_router(health_router)
     app.include_router(links_router)
     app.include_router(media_router)
@@ -264,6 +423,11 @@ def create_app() -> FastAPI:
         async def serve_ui():
             """Serve the Web UI."""
             return FileResponse(static_dir / "index.html")
+    else:
+        logging.getLogger(__name__).warning(
+            "static_dir not found at %s — Web UI disabled (API only)",
+            static_dir,
+        )
 
     return app
 
@@ -314,55 +478,86 @@ def main():
         print(f"Config directory: {get_config_dir()}")
         if settings.api_tokens:
             print("\nAPI Tokens:")
-            for label, token in settings.api_tokens.items():
+            for label, spec in settings.api_tokens.items():
-                print(f"  {label:20} {token}")
+                scope_str = ",".join(spec.scopes)
+                print(f"  {label:20} {spec.token}  [scopes: {scope_str}]")
         else:
             print("\nAuthentication is DISABLED (no tokens configured)")
         return
 
+    # Stderr is invisible when launched via wscript / pythonw (Start Menu shortcut,
+    # autostart). Mirror pre-uvicorn failures to a file in the config dir so the
+    # next silent boot failure is diagnosable.
+    def _fatal(msg: str, exit_code: int = 1) -> None:
+        print(msg, file=sys.stderr)
+        try:
+            log_path = get_config_dir() / "startup-errors.log"
+            from datetime import datetime
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(f"[{datetime.now().isoformat(timespec='seconds')}] {msg}\n")
+        except OSError:
+            pass
+        sys.exit(exit_code)
+
     # First-run bootstrap: if no config has ever been written, generate one
     # with a random token instead of starting in the insecure "no-auth" mode.
     config_path = get_config_dir() / "config.yaml"
     if not config_path.exists() and not settings.api_tokens:
         try:
             generate_default_config(config_path)
-            print(
+            _fatal(
                 f"\nFirst run: generated default config at {config_path}.\n"
                 "Run --show-token to retrieve the API token, then restart.",
-                file=sys.stderr,
+                exit_code=0,
             )
-            sys.exit(0)
         except OSError as e:
             print(f"WARNING: could not bootstrap config: {e}", file=sys.stderr)
 
     # Refuse to bind a non-loopback address with no tokens, unless explicitly opted in.
     non_loopback = args.host not in ("127.0.0.1", "localhost", "::1")
     if non_loopback and not settings.api_tokens and not settings.allow_lan_without_auth:
-        print(
+        _fatal(
             "ERROR: refusing to bind a non-loopback address with no api_tokens configured.\n"
             "Either set api_tokens in config.yaml, bind to 127.0.0.1,"
-            " or set allow_lan_without_auth: true in config.yaml to override.",
+            " or set allow_lan_without_auth: true in config.yaml to override."
-            file=sys.stderr,
         )
-        sys.exit(1)
 
     # Check if port is available before starting
     with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as sock:
         try:
             sock.bind((args.host if args.host != "0.0.0.0" else "127.0.0.1", args.port))
         except OSError:
-            print(
+            _fatal(
                 f"ERROR: Port {args.port} is already in use. "
                 f"Another instance of Media Server may be running.\n"
-                f"Stop the other process or use --port to pick a different port.",
+                f"Stop the other process or use --port to pick a different port."
-                file=sys.stderr,
             )
-            sys.exit(1)
 
     from .tray import PYSTRAY_AVAILABLE, TrayManager
 
     use_tray = PYSTRAY_AVAILABLE and not args.no_tray
 
+    # Validate TLS pair consistency before either path so we don't fail late.
+    if bool(settings.ssl_certfile) ^ bool(settings.ssl_keyfile):
+        _fatal(
+            "ERROR: ssl_certfile and ssl_keyfile must both be set, or both unset."
+        )
+
+    def _uvicorn_kwargs() -> dict:
+        kw: dict = {
+            "host": args.host,
+            "port": args.port,
+            "log_level": settings.log_level.lower(),
+            "proxy_headers": settings.proxy_headers,
+            "forwarded_allow_ips": settings.forwarded_allow_ips,
+        }
+        if settings.ssl_certfile and settings.ssl_keyfile:
+            kw["ssl_certfile"] = settings.ssl_certfile
+            kw["ssl_keyfile"] = settings.ssl_keyfile
+            if settings.ssl_keyfile_password:
+                kw["ssl_keyfile_password"] = settings.ssl_keyfile_password
+        return kw
+
     if use_tray:
         import asyncio
         import threading
@@ -370,9 +565,7 @@ def main():
         # Run uvicorn in a background thread so tray owns the main thread message loop
         uv_config = uvicorn.Config(
             "media_server.main:app",
-            host=args.host,
+            **_uvicorn_kwargs(),
-            port=args.port,
-            log_level=settings.log_level.lower(),
         )
         server = uvicorn.Server(uv_config)
 
@@ -410,9 +603,8 @@ def main():
     else:
         uvicorn.run(
             "media_server.main:app",
-            host=args.host,
-            port=args.port,
             reload=False,
+            **_uvicorn_kwargs(),
         )
 
 

media_server/routes/__init__.py

@@ -4,6 +4,7 @@ from .audio import router as audio_router
 from .browser import router as browser_router
 from .callbacks import router as callbacks_router
 from .display import router as display_router
+from .foreground import router as foreground_router
 from .health import router as health_router
 from .links import router as links_router
 from .media import router as media_router
@@ -14,6 +15,7 @@ __all__ = [
     "browser_router",
     "callbacks_router",
     "display_router",
+    "foreground_router",
     "health_router",
     "links_router",
     "media_router",

media_server/routes/browser.py

@@ -36,12 +36,20 @@ def _spawn_background(coro) -> asyncio.Task:
 
 
 def _require_folder_management() -> None:
-    """Raise 403 if media folder management is disabled in config."""
+    """Raise 403 if media folder management is disabled OR caller lacks admin scope."""
     if not settings.media_folders_management:
         raise HTTPException(
             status_code=403,
             detail="Media folder management is disabled. Set media_folders_management: true in config.yaml to enable.",
         )
+    from ..auth import auth_enabled, token_has_scope, token_label_var
+    if auth_enabled():
+        label = token_label_var.get("unknown")
+        if not token_has_scope(label, "admin"):
+            raise HTTPException(
+                status_code=403,
+                detail=f"Token '{label}' lacks required scope: admin",
+            )
 
 
 async def _broadcast_after_open(controller, label: str, max_wait: float = 2.0) -> None:

media_server/routes/callbacks.py

@@ -8,12 +8,14 @@ import time
 from concurrent.futures import ThreadPoolExecutor
 from typing import Any
 
-from fastapi import APIRouter, Depends, HTTPException, status
+from fastapi import APIRouter, Depends, HTTPException, Request, status
 from pydantic import BaseModel, Field
 
 from ..auth import verify_token
 from ..config import CallbackConfig, settings
 from ..config_manager import config_manager
+from ..services.rate_limit import check as ratelimit_check
+from ..services.rate_limit import get_peer
 
 router = APIRouter(prefix="/api/callbacks", tags=["callbacks"])
 logger = logging.getLogger(__name__)
@@ -28,6 +30,7 @@ def shutdown_callback_executor() -> None:
 
 
 def _require_callbacks_management() -> None:
+    """Authorise a callbacks-CRUD operation. Operator flag + per-token admin scope."""
     if not settings.callbacks_management:
         raise HTTPException(
             status_code=status.HTTP_403_FORBIDDEN,
@@ -36,6 +39,14 @@ def _require_callbacks_management() -> None:
                 " in config.yaml to enable."
             ),
         )
+    from ..auth import auth_enabled, token_has_scope, token_label_var
+    if auth_enabled():
+        label = token_label_var.get("unknown")
+        if not token_has_scope(label, "admin"):
+            raise HTTPException(
+                status_code=status.HTTP_403_FORBIDDEN,
+                detail=f"Token '{label}' lacks required scope: admin",
+            )
 
 
 class CallbackInfo(BaseModel):
@@ -122,6 +133,7 @@ async def list_callbacks(_: str = Depends(verify_token)) -> list[CallbackInfo]:
 @router.post("/execute/{callback_name}")
 async def execute_callback(
     callback_name: str,
+    http_request: Request,
     _: str = Depends(verify_token),
 ) -> CallbackExecuteResponse:
     """Execute a callback for debugging purposes.
@@ -132,6 +144,16 @@ async def execute_callback(
     Returns:
         Execution result including stdout, stderr, and exit code
     """
+    # Rate-limit callback execution per peer (10/min) — callbacks also run
+    # subprocesses and need the same protection as scripts.
+    allowed, retry_after = ratelimit_check("execute", get_peer(http_request))
+    if not allowed:
+        raise HTTPException(
+            status_code=429,
+            detail="Too many callback executions, slow down",
+            headers={"Retry-After": str(int(retry_after or 60))},
+        )
+
     # Validate callback name
     _validate_callback_name(callback_name)
 
@@ -146,6 +168,8 @@ async def execute_callback(
 
     logger.info(f"Executing callback for debugging: {callback_name}")
 
+    from ..services.audit_log import record_script_execution
+
     try:
         # Execute in dedicated thread pool to not block the default executor
         loop = asyncio.get_running_loop()
@@ -159,6 +183,15 @@ async def execute_callback(
             ),
         )
 
+        record_script_execution(
+            kind="callback",
+            name=callback_name,
+            exit_code=result["exit_code"],
+            duration=result.get("execution_time"),
+            stdout=result.get("stdout"),
+            stderr=result.get("stderr"),
+        )
+
         return CallbackExecuteResponse(
             success=result["exit_code"] == 0,
             callback=callback_name,
@@ -170,6 +203,13 @@ async def execute_callback(
 
     except Exception as e:
         logger.error(f"Callback execution error: {e}")
+        record_script_execution(
+            kind="callback",
+            name=callback_name,
+            exit_code=None,
+            duration=None,
+            error=str(e),
+        )
         return CallbackExecuteResponse(
             success=False,
             callback=callback_name,

media_server/routes/foreground.py

@@ -0,0 +1,26 @@
+"""Foreground (topmost) window/process API."""
+
+import asyncio
+import logging
+
+from fastapi import APIRouter, Depends
+
+from ..auth import verify_token
+from ..services.foreground_service import get_foreground_info
+
+logger = logging.getLogger(__name__)
+
+router = APIRouter(prefix="/api/foreground", tags=["foreground"])
+
+
+@router.get("")
+async def get_foreground(
+    refresh: bool = False, _: str = Depends(verify_token)
+) -> dict:
+    """Return metadata about the foreground window and owning process.
+
+    The probe is cached for ~500ms server-side; pass ``?refresh=1`` to bypass
+    the cache for one-shot queries.
+    """
+    info = await asyncio.to_thread(get_foreground_info, refresh)
+    return info.to_dict()

media_server/routes/links.py

@@ -39,11 +39,20 @@ def _validate_icon(icon: str) -> str:
 
 
 def _require_links_management() -> None:
+    """Authorise a links-CRUD operation. Operator flag + per-token admin scope."""
     if not settings.links_management:
         raise HTTPException(
             status_code=status.HTTP_403_FORBIDDEN,
             detail="Links management is disabled. Set links_management: true in config.yaml to enable.",
         )
+    from ..auth import auth_enabled, token_has_scope, token_label_var
+    if auth_enabled():
+        label = token_label_var.get("unknown")
+        if not token_has_scope(label, "admin"):
+            raise HTTPException(
+                status_code=status.HTTP_403_FORBIDDEN,
+                detail=f"Token '{label}' lacks required scope: admin",
+            )
 
 
 class LinkInfo(BaseModel):

media_server/routes/media.py

@@ -3,13 +3,22 @@
 import asyncio
 import logging
 
-from fastapi import APIRouter, Depends, HTTPException, Query, WebSocket, WebSocketDisconnect, status
+from fastapi import (
+    APIRouter,
+    Depends,
+    HTTPException,
+    Query,
+    Request,
+    WebSocket,
+    WebSocketDisconnect,
+    status,
+)
 from fastapi.responses import Response
 
 from ..auth import verify_token, verify_token_or_query
 from ..config import settings
 from ..models import MediaStatus, SeekRequest, VolumeRequest
-from ..services import get_current_album_art, get_media_controller
+from ..services import get_current_album_art_async, get_media_controller
 from ..services.websocket_manager import ws_manager
 
 logger = logging.getLogger(__name__)
@@ -17,19 +26,28 @@ logger = logging.getLogger(__name__)
 router = APIRouter(prefix="/api/media", tags=["media"])
 
 
+# Strong refs to background tasks so the asyncio GC can't drop them before
+# they run. Mirrors the pattern used in routes/browser.py.
+_background_callback_tasks: set[asyncio.Task] = set()
+
+
 def _run_callback(callback_name: str) -> None:
     """Fire-and-forget a callback if configured. Failures are logged but don't block."""
     if not settings.callbacks or callback_name not in settings.callbacks:
         return
 
     async def _execute():
+        # Use the dedicated callback executor (not the default loop pool) so a
+        # misbehaving callback can't starve the rest of the app's sync tasks.
+        from ..services.audit_log import record_script_execution
+        from .callbacks import _callback_executor
         from .scripts import _run_script
 
         try:
             callback = settings.callbacks[callback_name]
             loop = asyncio.get_running_loop()
             result = await loop.run_in_executor(
-                None,
+                _callback_executor,
                 lambda: _run_script(
                     command=callback.command,
                     timeout=callback.timeout,
@@ -37,6 +55,14 @@ def _run_callback(callback_name: str) -> None:
                     working_dir=callback.working_dir,
                 ),
             )
+            record_script_execution(
+                kind="event-callback",
+                name=callback_name,
+                exit_code=result["exit_code"],
+                duration=result.get("execution_time"),
+                stdout=result.get("stdout"),
+                stderr=result.get("stderr"),
+            )
             if result["exit_code"] != 0:
                 logger.warning(
                     "Callback %s failed with exit code %s: %s",
@@ -46,8 +72,18 @@ def _run_callback(callback_name: str) -> None:
                 )
         except Exception as e:
             logger.error("Callback %s error: %s", callback_name, e)
+            from ..services.audit_log import record_script_execution as _rec
+            _rec(
+                kind="event-callback",
+                name=callback_name,
+                exit_code=None,
+                duration=None,
+                error=str(e),
+            )
 
-    asyncio.create_task(_execute())
+    task = asyncio.create_task(_execute())
+    _background_callback_tasks.add(task)
+    task.add_done_callback(_background_callback_tasks.discard)
 
 
 @router.get("/status", response_model=MediaStatus)
@@ -242,41 +278,91 @@ async def toggle(_: str = Depends(verify_token)) -> dict:
 
 
 @router.get("/artwork")
-async def get_artwork(_: str = Depends(verify_token_or_query)) -> Response:
+async def get_artwork(
+    request: Request,
+    _: str = Depends(verify_token_or_query),
+) -> Response:
     """Get the current album artwork.
 
-    Returns:
+    Returns the bytes with a content-derived ETag so the browser can serve a
-        The album art image as PNG/JPEG
+    304 when the same track is re-requested.
     """
-    art_bytes = get_current_album_art()
+    art_bytes = await get_current_album_art_async()
     if art_bytes is None:
         raise HTTPException(
             status_code=status.HTTP_404_NOT_FOUND,
             detail="No album artwork available",
         )
 
-    # Try to detect image type from magic bytes
+    # Detect image type from magic bytes
-    content_type = "image/png"  # Default
     if art_bytes[:3] == b"\xff\xd8\xff":
         content_type = "image/jpeg"
     elif art_bytes[:8] == b"\x89PNG\r\n\x1a\n":
         content_type = "image/png"
-    elif art_bytes[:4] == b"RIFF" and art_bytes[8:12] == b"WEBP":
+    elif art_bytes[:4] == b"RIFF" and len(art_bytes) > 12 and art_bytes[8:12] == b"WEBP":
         content_type = "image/webp"
+    elif art_bytes[:2] == b"BM":
+        content_type = "image/bmp"
+    else:
+        content_type = "application/octet-stream"
 
-    return Response(content=art_bytes, media_type=content_type)
+    # Content-derived ETag (blake2b-128 — non-crypto cache key, ruff S324-safe)
+    import hashlib
+
+    etag = '"' + hashlib.blake2b(art_bytes, digest_size=16).hexdigest() + '"'
+
+    if request.headers.get("if-none-match") == etag:
+        return Response(status_code=status.HTTP_304_NOT_MODIFIED, headers={"ETag": etag})
+
+    return Response(
+        content=art_bytes,
+        media_type=content_type,
+        headers={
+            "ETag": etag,
+            "Cache-Control": "private, max-age=0, must-revalidate",
+        },
+    )
 
 
 @router.get("/visualizer/status")
 async def visualizer_status(_: str = Depends(verify_token)) -> dict:
-    """Check if audio visualizer is available and running."""
+    """Check if audio visualizer is available and running.
+
+    ``available`` is True only when both numpy + soundcard import.
+    ``unavailable_reason`` carries a short human-readable hint when False
+    or when a usable loopback device was not found — useful on Linux where
+    PulseAudio/PipeWire may not expose monitor sources to a headless
+    systemd-as-user service.
+    """
+    import platform as _platform
+
     from ..services.audio_analyzer import get_audio_analyzer
 
     analyzer = get_audio_analyzer()
+    reason: str | None = None
+    if not analyzer.available:
+        reason = "soundcard or numpy is not installed"
+    elif getattr(analyzer, "_unavailable", False):
+        if _platform.system() == "Linux":
+            reason = (
+                "No loopback audio device found. On Linux this requires a "
+                "running PulseAudio/PipeWire session with monitor sources "
+                "(systemd user service: ensure DBUS_SESSION_BUS_ADDRESS + "
+                "XDG_RUNTIME_DIR are set)."
+            )
+        elif _platform.system() == "Darwin":
+            reason = (
+                "No loopback audio device found. macOS does not expose system "
+                "loopback by default — install BlackHole or Soundflower."
+            )
+        else:
+            reason = "No loopback audio device found"
+
     return {
         "available": analyzer.available,
         "running": analyzer.running,
         "current_device": analyzer.current_device,
+        "unavailable_reason": reason,
     }
 
 
@@ -323,12 +409,17 @@ async def set_visualizer_device(
 @router.websocket("/ws")
 async def websocket_endpoint(
     websocket: WebSocket,
-    token: str | None = Query(None, description="API authentication token"),
+    token: str | None = Query(None, description="API authentication token (legacy)"),
 ) -> None:
     """WebSocket endpoint for real-time media status updates.
 
-    Authentication is done via query parameter since WebSocket
+    Authentication is accepted from two sources, in priority order:
-    doesn't support custom headers in the browser.
+      1. ``Sec-WebSocket-Protocol`` subprotocol of the form
+         ``media-server.token.<TOKEN>``. This is the preferred path because
+         the token never lands in the URL, request logs, or browser history.
+         The browser WebSocket API supports custom subprotocols natively.
+      2. ``?token=<TOKEN>`` query parameter (legacy, kept for back-compat
+         with older clients and the HA integration).
 
     Messages sent to client:
     - {"type": "status", "data": {...}} - Initial status on connect
@@ -339,11 +430,54 @@ async def websocket_endpoint(
     - {"type": "ping"} - Keepalive, server responds with {"type": "pong"}
     - {"type": "get_status"} - Request current status
     """
+    # Pull token from subprotocol if present. WebSocket spec lets either side
+    # negotiate exactly one subprotocol back; we accept the token one and
+    # answer with the same string so browsers consider the negotiation
+    # successful.
+    subprotocol_token: str | None = None
+    accept_subprotocol: str | None = None
+    raw_protocols = websocket.headers.get("sec-websocket-protocol", "")
+    for proto in (p.strip() for p in raw_protocols.split(",") if p.strip()):
+        if proto.startswith("media-server.token."):
+            subprotocol_token = proto[len("media-server.token."):]
+            accept_subprotocol = proto
+            break
+    effective_token = subprotocol_token or token
+    # Origin check — block CSWSH from third-party LAN pages. Accept the same
+    # set of origins as CORS plus the default localhost loopback, AND any
+    # same-origin connection (where Origin matches the request's Host header).
+    # Same-origin is inherently safe from CSWSH because CSWSH is a *cross*-
+    # origin attack — without this, binding to 0.0.0.0 and accessing the UI
+    # via a LAN IP would have its WebSocket rejected by the browser-sent
+    # Origin, which the static allowlist can't anticipate.
+    allowed_origins = set(
+        settings.cors_origins
+        or [
+            f"http://localhost:{settings.port}",
+            f"http://127.0.0.1:{settings.port}",
+        ]
+    )
+    origin = websocket.headers.get("origin")
+    # Same-origin connections from native apps may omit Origin entirely; only
+    # reject when an Origin is present AND not in the allow-list.
+    if origin is not None and origin not in allowed_origins:
+        host_header = websocket.headers.get("host", "")
+        # Origin uses http/https; match against both scheme variants of Host
+        # so HTTPS deployments without an explicit cors_origins still work.
+        same_origin_candidates = (
+            {f"http://{host_header}", f"https://{host_header}"}
+            if host_header
+            else set()
+        )
+        if origin not in same_origin_candidates:
+            await websocket.close(code=4003, reason="Origin not allowed")
+            return
+
     # Verify token
     from ..auth import auth_enabled, get_token_label, token_label_var
 
     if auth_enabled():
-        label = get_token_label(token) if token else None
+        label = get_token_label(effective_token) if effective_token else None
         if label is None:
             await websocket.close(code=4001, reason="Invalid authentication token")
             return
@@ -351,16 +485,25 @@ async def websocket_endpoint(
     else:
         token_label_var.set("anonymous")
 
-    await ws_manager.connect(websocket)
+    # Accept with the negotiated subprotocol if one was used. Starlette's
+    # connect() calls accept() with no subprotocol — we need to accept first
+    # explicitly to echo the subprotocol back, then hand off to the manager.
+    if accept_subprotocol is not None:
+        await websocket.accept(subprotocol=accept_subprotocol)
+        await ws_manager.connect(websocket, already_accepted=True)
+    else:
+        await ws_manager.connect(websocket)
 
     try:
         while True:
             # Wait for messages from client (for keepalive/ping)
             data = await websocket.receive_json()
 
-            if data.get("type") == "ping":
+            msg_type = data.get("type") if isinstance(data, dict) else None
+
+            if msg_type == "ping":
                 await websocket.send_json({"type": "pong"})
-            elif data.get("type") == "get_status":
+            elif msg_type == "get_status":
                 # Allow manual status request
                 controller = get_media_controller()
                 status_data = await controller.get_status()
@@ -368,15 +511,20 @@ async def websocket_endpoint(
                     "type": "status",
                     "data": status_data.model_dump(),
                 })
-            elif data.get("type") == "volume":
+            elif msg_type == "volume":
-                # Low-latency volume control via WebSocket
+                # Low-latency volume control via WebSocket. Coerce, clamp, and
-                volume = data.get("volume")
+                # never drop the socket on a single bad message — that would
-                if volume is not None:
+                # turn the WS into a one-shot DoS for any holder of a token.
-                    controller = get_media_controller()
+                try:
-                    await controller.set_volume(int(volume))
+                    volume = int(data.get("volume"))
-            elif data.get("type") == "enable_visualizer":
+                except (TypeError, ValueError):
+                    continue
+                volume = max(0, min(100, volume))
+                controller = get_media_controller()
+                await controller.set_volume(volume)
+            elif msg_type == "enable_visualizer":
                 await ws_manager.subscribe_visualizer(websocket)
-            elif data.get("type") == "disable_visualizer":
+            elif msg_type == "disable_visualizer":
                 await ws_manager.unsubscribe_visualizer(websocket)
 
     except WebSocketDisconnect:

media_server/routes/scripts.py

@@ -10,12 +10,14 @@ import time
 from concurrent.futures import ThreadPoolExecutor
 from typing import Any
 
-from fastapi import APIRouter, Depends, HTTPException, status
+from fastapi import APIRouter, Depends, HTTPException, Request, status
 from pydantic import BaseModel, Field
 
 from ..auth import verify_token
 from ..config import ScriptConfig, ScriptParameterConfig, settings
 from ..config_manager import config_manager
+from ..services.rate_limit import check as ratelimit_check
+from ..services.rate_limit import get_peer
 from ..services.websocket_manager import ws_manager
 
 router = APIRouter(prefix="/api/scripts", tags=["scripts"])
@@ -31,6 +33,12 @@ def shutdown_script_executor() -> None:
 
 
 def _require_scripts_management() -> None:
+    """Authorise a scripts-CRUD operation.
+
+    Two gates: the operator-level `scripts_management` flag in config.yaml,
+    AND the per-token `admin` scope check (read from request-context). Either
+    failure → 403.
+    """
     if not settings.scripts_management:
         raise HTTPException(
             status_code=status.HTTP_403_FORBIDDEN,
@@ -39,6 +47,14 @@ def _require_scripts_management() -> None:
                 " in config.yaml to enable."
             ),
         )
+    from ..auth import auth_enabled, token_has_scope, token_label_var
+    if auth_enabled():
+        label = token_label_var.get("unknown")
+        if not token_has_scope(label, "admin"):
+            raise HTTPException(
+                status_code=status.HTTP_403_FORBIDDEN,
+                detail=f"Token '{label}' lacks required scope: admin",
+            )
 
 
 class ScriptExecuteRequest(BaseModel):
@@ -215,6 +231,28 @@ def _validate_params(
             # string — just convert to str
             value = str(value)
 
+        # Optional regex constraint, validated against the *string form* of the
+        # value. This is the only practical defence for string parameters that
+        # flow into shell=true scripts via env vars (Windows cmd.exe expands
+        # `%VAR%` after argument parsing, so embedded `&`/`|`/`%` would inject
+        # commands). Authors of shell scripts should ALWAYS define a pattern.
+        if pdef.pattern:
+            try:
+                if not re.fullmatch(pdef.pattern, str(value)):
+                    raise HTTPException(
+                        status_code=status.HTTP_400_BAD_REQUEST,
+                        detail=(
+                            f"Parameter '{pname}' value {value!r} does not match"
+                            f" required pattern: {pdef.pattern}"
+                        ),
+                    )
+            except re.error as e:
+                # Bad pattern in config — fail closed.
+                raise HTTPException(
+                    status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+                    detail=f"Parameter '{pname}' has invalid pattern: {e}",
+                ) from e
+
         env_vars[f"SCRIPT_PARAM_{pname.upper()}"] = str(value)
 
     return env_vars
@@ -223,6 +261,7 @@ def _validate_params(
 @router.post("/execute/{script_name}")
 async def execute_script(
     script_name: str,
+    http_request: Request,
     request: ScriptExecuteRequest | None = None,
     _: str = Depends(verify_token),
 ) -> ScriptExecuteResponse:
@@ -235,6 +274,16 @@ async def execute_script(
     Returns:
         Execution result including stdout, stderr, and exit code
     """
+    # Rate-limit script execution per peer so a leaked token can't be used to
+    # spam the shell-exec endpoint.
+    allowed, retry_after = ratelimit_check("execute", get_peer(http_request))
+    if not allowed:
+        raise HTTPException(
+            status_code=429,
+            detail="Too many script executions, slow down",
+            headers={"Retry-After": str(int(retry_after or 60))},
+        )
+
     # Check if script exists
     if script_name not in settings.scripts:
         raise HTTPException(
@@ -249,6 +298,8 @@ async def execute_script(
 
     logger.info(f"Executing script: {script_name}")
 
+    from ..services.audit_log import record_script_execution
+
     try:
         # Execute in dedicated thread pool to not block the default executor
         loop = asyncio.get_running_loop()
@@ -263,6 +314,15 @@ async def execute_script(
             ),
         )
 
+        record_script_execution(
+            kind="script",
+            name=script_name,
+            exit_code=result["exit_code"],
+            duration=result.get("execution_time"),
+            stdout=result.get("stdout"),
+            stderr=result.get("stderr"),
+        )
+
         return ScriptExecuteResponse(
             success=result["exit_code"] == 0,
             script=script_name,
@@ -274,6 +334,13 @@ async def execute_script(
 
     except Exception as e:
         logger.error(f"Script execution error: {e}")
+        record_script_execution(
+            kind="script",
+            name=script_name,
+            exit_code=None,
+            duration=None,
+            error=str(e),
+        )
         return ScriptExecuteResponse(
             success=False,
             script=script_name,
@@ -313,9 +380,21 @@ def _run_script(
     else:
         popen_kwargs["start_new_session"] = True
 
+    # When shell=False, the user-provided command string is split via shlex
+    # (POSIX rules — also works for Windows args without backslashes). This
+    # disables shell metacharacter expansion entirely, so SCRIPT_PARAM_* env
+    # vars referenced as $FOO / %FOO% will be treated as literal text by the
+    # process, not interpreted by a shell. Use shell=false for any script
+    # whose params come from external input.
+    if shell:
+        run_command: str | list[str] = command
+    else:
+        import shlex
+        run_command = shlex.split(command, posix=(sys.platform != "win32"))
+
     try:
         result = subprocess.run(
-            command,
+            run_command,
             shell=shell,
             cwd=working_dir,
             capture_output=True,

media_server/service/install_linux.sh

@@ -57,25 +57,38 @@ install_service() {
     # Create installation directory
     mkdir -p "$INSTALL_DIR"
 
-    # Copy source files
+    # Resolve the source-tree root (two levels up from this script:
-    cp -r "$(dirname "$0")/../"* "$INSTALL_DIR/"
+    # media_server/service/install_linux.sh → repo root).
+    SOURCE_ROOT="$(cd "$(dirname "$0")/../.." && pwd)"
+
+    # Copy the source tree (pyproject.toml + media_server/ package)
+    cp -r "$SOURCE_ROOT/." "$INSTALL_DIR/"
 
     # Create virtual environment
     echo_info "Creating Python virtual environment..."
     python3 -m venv "$INSTALL_DIR/venv"
 
-    # Install Python dependencies
+    # Install Python dependencies from pyproject.toml (with linux extra).
+    # cd into the install dir so pip resolves `.[linux]` against the local
+    # pyproject — passing a path-with-extras (`$INSTALL_DIR[linux]`) trips
+    # the requirement-spec parser on some pip versions.
     echo_info "Installing Python dependencies..."
     "$INSTALL_DIR/venv/bin/pip" install --upgrade pip
-    "$INSTALL_DIR/venv/bin/pip" install -r "$INSTALL_DIR/requirements.txt"
+    (cd "$INSTALL_DIR" && "$INSTALL_DIR/venv/bin/pip" install ".[linux]")
 
-    # Install systemd service file
+    # Install systemd service file (templated unit)
     echo_info "Installing systemd service..."
-    cp "$INSTALL_DIR/service/media-server.service" "$SERVICE_FILE"
+    cp "$INSTALL_DIR/media_server/service/media-server.service" "$SERVICE_FILE"
 
     # Reload systemd
     systemctl daemon-reload
 
+    # Pre-create writable state dirs so the unit's ReadWritePaths grant
+    # actually has something to grant.
+    sudo -u "$SUDO_USER" mkdir -p \
+        "/home/$SUDO_USER/.config/media-server" \
+        "/home/$SUDO_USER/.cache/media-server"
+
     # Generate config if not exists
     if [[ ! -f "/home/$SUDO_USER/.config/media-server/config.yaml" ]]; then
         echo_info "Generating configuration file..."

media_server/service/media-server.service

@@ -3,34 +3,38 @@ Description=Media Server - REST API for controlling system media playback
 After=network.target sound.target
 Wants=sound.target
 
+# Templated unit. Enable as:
+#   sudo systemctl enable --now media-server@$USER
+# %i is the user name supplied after the '@'; %U is the matching numeric UID.
+
 [Service]
 Type=simple
 User=%i
 Group=%i
 
-# Environment variables (optional - can also use config file)
+# Working directory (override via drop-in if you install elsewhere)
-# Environment=MEDIA_SERVER_HOST=0.0.0.0
-# Environment=MEDIA_SERVER_PORT=8765
-# Environment=MEDIA_SERVER_API_TOKEN=your-secret-token
-
-# Working directory
 WorkingDirectory=/opt/media-server
 
-# Start command - adjust path to your Python environment
+# Start command — adjust to match where you installed the venv. --no-tray
-ExecStart=/opt/media-server/venv/bin/python -m media_server.main
+# avoids pulling pystray into a headless service environment.
+ExecStart=/opt/media-server/venv/bin/python -m media_server.main --no-tray
 
-# Restart policy
+Restart=on-failure
-Restart=always
 RestartSec=10
 
-# Security hardening
+# Required for MPRIS (dbus.SessionBus) and PulseAudio/PipeWire loopback.
+Environment=PYTHONUNBUFFERED=1
+Environment=XDG_RUNTIME_DIR=/run/user/%U
+Environment=DBUS_SESSION_BUS_ADDRESS=unix:path=/run/user/%U/bus
+
+# Light sandboxing. ProtectHome=read-only by itself blocks the app's own
+# audit.log / thumbnail cache writes — ReadWritePaths re-opens just the
+# two state dirs the server owns.
 NoNewPrivileges=true
+PrivateTmp=true
 ProtectSystem=strict
 ProtectHome=read-only
-PrivateTmp=true
+ReadWritePaths=/home/%i/.config/media-server /home/%i/.cache/media-server
-
-# Required for D-Bus access (MPRIS)
-Environment=DBUS_SESSION_BUS_ADDRESS=unix:path=/run/user/%U/bus
 
 [Install]
 WantedBy=multi-user.target

media_server/services/__init__.py

@@ -65,7 +65,12 @@ def get_media_controller() -> "MediaController":
 
 
 def get_current_album_art() -> bytes | None:
-    """Get the current album art bytes (Windows only for now)."""
+    """Get the current album art bytes (synchronous, Windows-cached path).
+
+    Windows pre-populates a module-level cache via the WinRT polling thread,
+    so this stays sync. For Linux/macOS the controller fetches on demand —
+    use ``get_current_album_art_async()`` from FastAPI handlers instead.
+    """
     system = platform.system()
     if system == "Windows":
         from .windows_media import get_current_album_art as _get_art
@@ -73,6 +78,22 @@ def get_current_album_art() -> bytes | None:
     return None
 
 
+async def get_current_album_art_async() -> bytes | None:
+    """Cross-platform album art fetch. Awaits the controller's impl.
+
+    Falls back to the sync Windows cache when running on Windows so we don't
+    pay an extra coroutine hop for the existing path.
+    """
+    system = platform.system()
+    if system == "Windows":
+        return get_current_album_art()
+    controller = get_media_controller()
+    try:
+        return await controller.get_album_art()
+    except Exception:  # noqa: BLE001 — art is best-effort; never break the route
+        return None
+
+
 def get_audio_devices() -> list[dict[str, str]]:
     """Get list of available audio output devices (Windows only for now)."""
     system = platform.system()
@@ -82,4 +103,9 @@ def get_audio_devices() -> list[dict[str, str]]:
     return []
 
 
-__all__ = ["get_media_controller", "get_current_album_art", "get_audio_devices"]
+__all__ = [
+    "get_media_controller",
+    "get_current_album_art",
+    "get_current_album_art_async",
+    "get_audio_devices",
+]

media_server/services/audit_log.py

@@ -0,0 +1,120 @@
+"""Append-only audit log for sensitive actions (script + callback execution).
+
+Writes a single JSONL line per event to ``<config_dir>/audit.log``. The log is
+write-only from the app's perspective — it never reads back, and rotation is
+left to the operator (the file size is dominated by stdout/stderr truncation,
+which is already capped at 10 KB per stream in `_run_script`).
+
+Designed to be cheap: the write goes through a small background thread so the
+hot path never blocks on disk I/O, and a failure to write is logged at WARNING
+but never raised to callers.
+"""
+
+from __future__ import annotations
+
+import json
+import logging
+import queue
+import threading
+import time
+from typing import Any
+
+from ..auth import token_label_var
+from ..config import get_config_dir
+
+logger = logging.getLogger(__name__)
+
+# Cap on stdout/stderr inside the audit record so a chatty script doesn't
+# explode the log. Mirrors the 10k cap used by _run_script.
+_OUTPUT_CAP = 2000
+
+_audit_queue: "queue.Queue[dict[str, Any] | None]" = queue.Queue(maxsize=1000)
+_audit_thread: threading.Thread | None = None
+_audit_lock = threading.Lock()
+
+
+def _ensure_writer_started() -> None:
+    global _audit_thread
+    with _audit_lock:
+        if _audit_thread is not None and _audit_thread.is_alive():
+            return
+        _audit_thread = threading.Thread(
+            target=_audit_writer_loop,
+            name="audit-log",
+            daemon=True,
+        )
+        _audit_thread.start()
+
+
+def _audit_writer_loop() -> None:
+    log_path = get_config_dir() / "audit.log"
+    while True:
+        try:
+            record = _audit_queue.get()
+        except Exception:
+            return
+        if record is None:
+            return
+        try:
+            line = json.dumps(record, ensure_ascii=False, default=str)
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(line + "\n")
+        except OSError as e:
+            logger.warning("Failed to write audit record: %s", e)
+
+
+def _truncate(value: str | None) -> str | None:
+    if value is None:
+        return None
+    if len(value) <= _OUTPUT_CAP:
+        return value
+    return value[:_OUTPUT_CAP] + f"\n…[truncated, {len(value) - _OUTPUT_CAP} chars]"
+
+
+def record_script_execution(
+    *,
+    kind: str,
+    name: str,
+    exit_code: int | None,
+    duration: float | None,
+    stdout: str | None = None,
+    stderr: str | None = None,
+    error: str | None = None,
+) -> None:
+    """Append a single audit record. Never raises."""
+    _ensure_writer_started()
+    try:
+        record = {
+            "ts": time.time(),
+            "iso": time.strftime("%Y-%m-%dT%H:%M:%S", time.gmtime()),
+            "token_label": token_label_var.get("unknown"),
+            "kind": kind,
+            "name": name,
+            "exit_code": exit_code,
+            "duration_s": round(duration, 4) if duration is not None else None,
+            "success": exit_code == 0 if exit_code is not None else False,
+            "stdout": _truncate(stdout),
+            "stderr": _truncate(stderr),
+            "error": error,
+        }
+        _audit_queue.put_nowait(record)
+    except queue.Full:
+        # Backpressure: drop oldest record to make room. We'd rather lose an
+        # old entry than block the script that just ran.
+        try:
+            _audit_queue.get_nowait()
+            _audit_queue.put_nowait(record)
+        except queue.Empty:
+            pass
+    except Exception as e:
+        logger.warning("Failed to enqueue audit record: %s", e)
+
+
+def shutdown_audit_log() -> None:
+    """Flush the audit queue on app shutdown."""
+    try:
+        _audit_queue.put_nowait(None)
+    except queue.Full:
+        pass
+    if _audit_thread is not None:
+        _audit_thread.join(timeout=2)

media_server/services/browser_url_service.py

@@ -0,0 +1,296 @@
+"""Extract page-level metadata from a focused desktop web browser.
+
+The browser's window title is the reliable signal — every major browser
+formats it as ``"<page title> - <Browser Name>"``, so stripping the suffix
+gives us the page title for free.
+
+URL extraction was attempted via UI Automation (UIA), but Chromium-based
+browsers (Chrome/Edge/Brave/Vivaldi) keep their accessibility tree dormant
+unless a screen reader is active or ``--force-renderer-accessibility`` is
+set — neither is something we want to require from end users. The UIA
+machinery is still here behind a feature flag in case a future caller
+opts into the accessibility-flag path; by default we just return the
+page title and leave ``url=None``.
+
+Other platforms (macOS via AppleScript, Linux via AT-SPI) are out of scope
+for this iteration.
+"""
+
+from __future__ import annotations
+
+import logging
+import os
+import platform
+import threading
+from dataclasses import dataclass
+
+logger = logging.getLogger(__name__)
+
+# UIA URL extraction is opt-in because Chromium browsers keep their
+# accessibility tree dormant unless the user starts the browser with
+# ``--force-renderer-accessibility`` (or a screen reader is running).
+# Without that, `FindAll` throws and we'd burn 5s per probe retrying.
+# Set MEDIA_SERVER_BROWSER_UIA=1 to enable; default off.
+_UIA_ENABLED = os.environ.get("MEDIA_SERVER_BROWSER_UIA", "").lower() in (
+    "1", "true", "yes", "on"
+)
+
+
+# Known browser executables (lowercase, .exe-stripped). Used to decide
+# whether to spend the UIA query budget on this foreground process.
+BROWSER_PROCESS_HINTS: frozenset[str] = frozenset({
+    "chrome",
+    "msedge",
+    "firefox",
+    "brave",
+    "opera",
+    "vivaldi",
+    "yandex",
+    "browser",  # Yandex Browser sometimes reports as browser.exe
+    "arc",
+    "thorium",
+})
+
+
+@dataclass(frozen=True)
+class BrowserPageInfo:
+    url: str | None = None
+    page_title: str | None = None
+
+
+_EMPTY = BrowserPageInfo()
+
+
+def is_browser_process(process_name: str | None) -> bool:
+    """Return True when ``process_name`` looks like a supported browser."""
+    if not process_name:
+        return False
+    base = process_name.lower()
+    if base.endswith(".exe"):
+        base = base[:-4]
+    return base in BROWSER_PROCESS_HINTS
+
+
+def _strip_browser_suffix(title: str | None, process_name: str | None) -> str | None:
+    """Pull the page title out of the browser's window title.
+
+    Most browsers format their window title as ``"<page> - <Browser Name>"``.
+    We strip the trailing suffix so consumers get the page title alone. If
+    the suffix can't be matched, return the raw title unchanged.
+    """
+    if not title:
+        return None
+    suffixes = (
+        " - Google Chrome",
+        " — Google Chrome",
+        " - Microsoft Edge",
+        " - Microsoft Edge",
+        " — Mozilla Firefox",
+        " - Mozilla Firefox",
+        " - Brave",
+        " - Opera",
+        " - Vivaldi",
+        " - Yandex",
+    )
+    for s in suffixes:
+        if title.endswith(s):
+            return title[: -len(s)].strip() or None
+    return title
+
+
+# ─── UIA lookup (Windows) ───────────────────────────────────────────
+
+# UIA control type / property constants we need. Avoiding the full
+# UIAutomationClient typelib generation — those constants are stable.
+_UIA_EditControlTypeId = 50004
+_UIA_ControlTypePropertyId = 30003
+_UIA_ValueValuePropertyId = 30045
+_UIA_NamePropertyId = 30005
+_UIA_ValuePatternId = 10002
+_TreeScope_Descendants = 4
+_PropertyConditionFlags_IgnoreCase = 1
+
+
+# Lazy import + per-thread COM init.
+_uia_lock = threading.Lock()
+_uia_singleton = None
+_uia_load_error: str | None = None
+_uia_thread_local = threading.local()
+
+
+def _ensure_com() -> None:
+    """Initialise COM on the current thread (idempotent per thread)."""
+    if getattr(_uia_thread_local, "initialised", False):
+        return
+    try:
+        import comtypes  # type: ignore
+
+        # COINIT_APARTMENTTHREADED is required by UIA; comtypes' default
+        # CoInitializeEx already passes that flag.
+        comtypes.CoInitialize()
+        _uia_thread_local.initialised = True
+    except Exception as e:
+        logger.debug("CoInitialize failed: %s", e)
+
+
+def _get_uia():
+    """Return the IUIAutomation singleton, or None if unavailable."""
+    global _uia_singleton, _uia_load_error
+    if _uia_singleton is not None:
+        return _uia_singleton
+    if _uia_load_error is not None:
+        return None
+    with _uia_lock:
+        if _uia_singleton is not None:
+            return _uia_singleton
+        try:
+            import comtypes.client  # type: ignore
+
+            # CLSID for CUIAutomation. Using GetActiveObject would fail,
+            # so we cocreate. comtypes.client.CreateObject keeps the COM
+            # plumbing tidy.
+            _uia_singleton = comtypes.client.CreateObject(
+                "{ff48dba4-60ef-4201-aa87-54103eef594e}",
+                interface=comtypes.client.GetModule(
+                    "UIAutomationCore.dll"
+                ).IUIAutomation,
+            )
+            return _uia_singleton
+        except Exception as e:
+            _uia_load_error = str(e)
+            logger.info("UIA unavailable; browser URL extraction disabled: %s", e)
+            return None
+
+
+def _find_address_bar_value(hwnd: int) -> str | None:
+    """Walk the UIA tree under ``hwnd`` looking for the URL Edit control.
+
+    Strategy: find every descendant Edit control, then pick the first one
+    whose Name contains an address-bar hint, or — failing that — the first
+    one whose value parses as a URL-ish string. Browsers expose extra Edit
+    controls (search bars, find-in-page) so name matching is the reliable
+    signal; the URL-ish fallback covers locale variants we haven't seen.
+    """
+    _ensure_com()
+    uia = _get_uia()
+    if uia is None:
+        return None
+
+    try:
+        element = uia.ElementFromHandle(hwnd)
+        if not element:
+            return None
+
+        # Build a condition matching ControlType=Edit, then enumerate.
+        edit_condition = uia.CreatePropertyCondition(
+            _UIA_ControlTypePropertyId, _UIA_EditControlTypeId
+        )
+        edits = element.FindAll(_TreeScope_Descendants, edit_condition)
+        count = edits.Length if edits else 0
+        if count == 0:
+            return None
+
+        # Hints (lowercase) used to identify the address bar by its Name
+        # property. Covers en-US plus a few common locales / browsers.
+        name_hints = (
+            "address",          # Chrome/Edge: "Address and search bar"
+            "адрес",            # Chrome ru: "Адресная строка и строка поиска"
+            "адресная",
+            "search with",      # Firefox: "Search with Google or enter address"
+            "поиск или ввод",   # Firefox ru
+            "url",
+            "location",
+        )
+
+        # First pass: name-based match (high confidence).
+        candidates: list[tuple[int, str]] = []
+        for i in range(count):
+            edit = edits.GetElement(i)
+            try:
+                name = (edit.CurrentName or "").lower()
+            except Exception:
+                name = ""
+            try:
+                value = edit.GetCurrentPropertyValue(_UIA_ValueValuePropertyId)
+            except Exception:
+                value = None
+            if value is None:
+                continue
+            value_str = str(value)
+            for h in name_hints:
+                if h in name:
+                    return value_str
+            candidates.append((i, value_str))
+
+        # Second pass: URL-ish fallback. Pick the first candidate that
+        # looks like a URL; this catches browser/locale combos we haven't
+        # listed above.
+        for _i, v in candidates:
+            lv = v.lower()
+            if (
+                lv.startswith("http://")
+                or lv.startswith("https://")
+                or lv.startswith("about:")
+                or lv.startswith("chrome://")
+                or lv.startswith("edge://")
+                or lv.startswith("brave://")
+                or lv.startswith("file://")
+                or lv.startswith("ftp://")
+            ):
+                return v
+
+        return None
+    except Exception as e:
+        logger.debug("UIA address-bar lookup failed: %s", e)
+        return None
+
+
+# ─── Per-(hwnd, title) cache ────────────────────────────────────────
+
+_cache_lock = threading.Lock()
+_cache_key: tuple[int | None, str | None] = (None, None)
+_cache_value: BrowserPageInfo = _EMPTY
+
+
+def get_browser_page(
+    *,
+    hwnd: int | None,
+    process_name: str | None,
+    window_title: str | None,
+) -> BrowserPageInfo:
+    """Return the URL + page title for the foreground browser tab, if any.
+
+    Callers pass the already-resolved foreground HWND/title/process_name so
+    this service doesn't re-walk Win32 to find them. Returns ``_EMPTY`` for
+    non-browser processes or when UIA can't resolve the URL.
+    """
+    if not is_browser_process(process_name):
+        return _EMPTY
+    if platform.system() != "Windows":
+        # macOS/Linux paths not implemented in this iteration.
+        return _EMPTY
+    if not hwnd:
+        return _EMPTY
+
+    global _cache_key, _cache_value
+    key = (hwnd, window_title)
+    with _cache_lock:
+        if key == _cache_key and _cache_value is not _EMPTY:
+            return _cache_value
+
+    url = _find_address_bar_value(hwnd) if _UIA_ENABLED else None
+    page_title = _strip_browser_suffix(window_title, process_name)
+    info = BrowserPageInfo(url=url, page_title=page_title)
+
+    with _cache_lock:
+        _cache_key = key
+        _cache_value = info
+    return info
+
+
+def reset_cache() -> None:
+    """Reset the cache. Useful in tests."""
+    global _cache_key, _cache_value
+    with _cache_lock:
+        _cache_key = (None, None)
+        _cache_value = _EMPTY

media_server/services/display_service.py

@@ -192,10 +192,11 @@ _CACHE_TTL = 5.0  # seconds
 # Per-monitor cache of static capabilities (option lists + support flags).
 # DDC/CI capability discovery is the slow part — it only changes when a
 # monitor is replaced or rewired, so we probe it once per monitor and reuse
-# it across refreshes. Cleared on explicit `rediscover` or when the monitor
+# it across refreshes. Keyed by a stable identity tuple
-# count changes (cheap stale-detection for hot-plug events).
+# (manufacturer, model, edid_hash) so that hot-plug swaps where the new
-_static_cache: dict[int, dict] = {}
+# topology has the same number of monitors but different devices still
-_static_cache_monitor_count: int = -1
+# refresh the cache for the new monitor instead of serving stale capabilities.
+_static_cache: dict[tuple, dict] = {}
 
 
 def _enum_name(value, enum_cls=None) -> str | None:
@@ -353,7 +354,7 @@ def list_monitors(force_refresh: bool = False, rediscover: bool = False) -> list
             next probe re-runs DDC/CI capability discovery. Use after hot-plug
             or when a monitor's reported capabilities change.
     """
-    global _monitor_cache, _cache_time, _static_cache_monitor_count
+    global _monitor_cache, _cache_time
 
     if (
         not force_refresh
@@ -372,12 +373,11 @@ def list_monitors(force_refresh: bool = False, rediscover: bool = False) -> list
         info_list = sbc.list_monitors_info()
         brightnesses = sbc.get_brightness()
 
-        # Invalidate the static cache on explicit rediscover OR on topology
+        # Explicit rediscover wipes the whole cache; otherwise rely on stable
-        # change (hot-plug / disconnect). Both indicate the cached probe is
+        # per-monitor keys (manufacturer|model|edid_hash) so a hot-plug swap
-        # potentially stale.
+        # invalidates the entry for the missing monitor automatically.
-        if rediscover or len(info_list) != _static_cache_monitor_count:
+        if rediscover:
             _static_cache.clear()
-            _static_cache_monitor_count = len(info_list)
 
         mc = _load_monitorcontrol()
         ddc_monitors = []
@@ -387,6 +387,9 @@ def list_monitors(force_refresh: bool = False, rediscover: bool = False) -> list
             except Exception:
                 pass
 
+        import hashlib
+
+        seen_keys: set[tuple] = set()
         for i, info in enumerate(info_list):
             name = info.get("name", f"Monitor {i}")
             model = info.get("model", "")
@@ -400,6 +403,21 @@ def list_monitors(force_refresh: bool = False, rediscover: bool = False) -> list
             edid = info.get("edid", "")
             resolution = _parse_edid_resolution(edid) if edid else None
 
+            # Stable cache key — EDID hash is unique per physical monitor.
+            # Fall back to (manufacturer, model, serial-ish) when EDID is
+            # missing, then to the legacy index as a last resort.
+            if edid:
+                edid_hash = hashlib.blake2b(
+                    edid.encode("utf-8") if isinstance(edid, str) else bytes(edid),
+                    digest_size=8,
+                ).hexdigest()
+                cache_key: tuple = ("edid", edid_hash)
+            elif manufacturer or model:
+                cache_key = ("mm", manufacturer, model, name)
+            else:
+                cache_key = ("idx", i)
+            seen_keys.add(cache_key)
+
             static: dict = {}
             dynamic: dict = {}
 
@@ -409,13 +427,13 @@ def list_monitors(force_refresh: bool = False, rediscover: bool = False) -> list
             if power_supported and i < len(ddc_monitors):
                 try:
                     with ddc_monitors[i] as mon:
-                        if i not in _static_cache:
+                        if cache_key not in _static_cache:
-                            _static_cache[i] = _probe_static_open(mon, mc, i)
+                            _static_cache[cache_key] = _probe_static_open(mon, mc, i)
-                        static = _static_cache[i]
+                        static = _static_cache[cache_key]
                         dynamic = _probe_dynamic_open(mon, mc, i, static)
                 except Exception as e:
                     logger.debug("Monitor %d: DDC/CI session failed: %s", i, e)
-                    static = _static_cache.get(i, {})
+                    static = _static_cache.get(cache_key, {})
 
             monitors.append(MonitorInfo(
                 id=i,
@@ -439,6 +457,12 @@ def list_monitors(force_refresh: bool = False, rediscover: bool = False) -> list
                 available_picture_modes=static.get("available_picture_modes", []),
                 picture_mode_supported=static.get("picture_mode_supported", False),
             ))
+        # Evict cache entries for monitors that disappeared from this scan so
+        # the next hot-plug of a different monitor with the same identity
+        # tuple (e.g. same model) doesn't hit a stale entry first.
+        for stale_key in list(_static_cache.keys()):
+            if stale_key not in seen_keys:
+                _static_cache.pop(stale_key, None)
     except Exception as e:
         logger.error("Failed to enumerate monitors: %s", e)
 

media_server/services/foreground_service.py

@@ -0,0 +1,543 @@
+"""Foreground (topmost) window/process tracking.
+
+Reports the process that currently owns the foreground window, plus useful
+metadata (window title, executable path, monitor index, whether the window
+covers a full monitor, process start time).
+
+All probes happen behind a short TTL cache so the WebSocket status poll and
+per-entity HA polls don't pay the OS call cost on every tick.
+
+Windows uses the Win32 API via ``ctypes`` (no extra dependency) and falls back
+gracefully when individual probes fail. Linux/macOS implementations are
+best-effort and return ``available=False`` when the required tooling is
+missing, so the rest of the stack keeps working.
+"""
+
+from __future__ import annotations
+
+import logging
+import platform
+import threading
+import time
+from dataclasses import asdict, dataclass, field
+
+logger = logging.getLogger(__name__)
+
+_CACHE_TTL = 0.5  # seconds — fast enough for WebSocket broadcast loop
+
+
+@dataclass(frozen=True)
+class ForegroundInfo:
+    """Snapshot of the foreground window/process."""
+
+    available: bool
+    pid: int | None = None
+    process_name: str | None = None
+    executable_path: str | None = None
+    window_title: str | None = None
+    window_handle: int | None = None
+    is_fullscreen: bool = False
+    is_minimized: bool = False
+    monitor_id: int | None = None
+    monitor_geometry: dict[str, int] | None = None
+    window_geometry: dict[str, int] | None = None
+    started_at: float | None = None
+    platform: str = field(default_factory=lambda: platform.system())
+    error: str | None = None
+    # Populated only when the foreground process is a recognised web
+    # browser. ``browser_page_title`` is derived from the window title
+    # (suffix stripped); ``browser_url`` requires UIA to succeed.
+    is_browser: bool = False
+    browser_url: str | None = None
+    browser_page_title: str | None = None
+
+    def to_dict(self) -> dict:
+        return asdict(self)
+
+
+_UNAVAILABLE = ForegroundInfo(available=False)
+
+
+class _Cache:
+    """Single-slot TTL cache shared across callers."""
+
+    def __init__(self) -> None:
+        self._lock = threading.Lock()
+        self._value: ForegroundInfo | None = None
+        self._fetched_at: float = 0.0
+
+    def get(self, ttl: float, fetch) -> ForegroundInfo:
+        with self._lock:
+            now = time.monotonic()
+            if self._value is not None and (now - self._fetched_at) < ttl:
+                return self._value
+        # Fetch outside the lock — OS calls can take tens of ms.
+        value = fetch()
+        with self._lock:
+            self._value = value
+            self._fetched_at = time.monotonic()
+        return value
+
+    def invalidate(self) -> None:
+        with self._lock:
+            self._value = None
+            self._fetched_at = 0.0
+
+
+_cache = _Cache()
+
+# Win32 handles + signatures are declared once at module load (when running on
+# Windows). The TTL cache fires this hundreds of times per minute; redoing the
+# DLL load + ~10 argtype assignments per call was the largest chunk of probe
+# cost. Keep these guarded behind a lazy init so non-Windows platforms don't
+# pay the import.
+_WIN32_INITIALIZED = False
+_win32_user32 = None
+_win32_kernel32 = None
+_win32_psapi = None
+
+
+def _init_win32_apis() -> None:
+    """Declare ctypes argtypes/restype on every Win32 call we make.
+
+    CRITICAL: ctypes defaults to `c_int` (32-bit) for HANDLE/HWND/HMONITOR
+    which silently truncates 64-bit pointer values on x64 — that corrupts the
+    handle so `CloseHandle()` can either fail or close the wrong kernel
+    object, and pointer-equality comparisons (monitor index lookup) miss.
+    """
+    global _WIN32_INITIALIZED, _win32_user32, _win32_kernel32, _win32_psapi
+    if _WIN32_INITIALIZED:
+        return
+
+    import ctypes
+    import ctypes.wintypes as wt
+
+    user32 = ctypes.WinDLL("user32", use_last_error=True)
+    kernel32 = ctypes.WinDLL("kernel32", use_last_error=True)
+    psapi = ctypes.WinDLL("psapi", use_last_error=True)
+
+    user32.GetForegroundWindow.restype = wt.HWND
+    user32.GetWindowThreadProcessId.argtypes = [wt.HWND, ctypes.POINTER(wt.DWORD)]
+    user32.GetWindowThreadProcessId.restype = wt.DWORD
+    user32.GetWindowTextLengthW.argtypes = [wt.HWND]
+    user32.GetWindowTextLengthW.restype = ctypes.c_int
+    user32.GetWindowTextW.argtypes = [wt.HWND, wt.LPWSTR, ctypes.c_int]
+    user32.GetWindowTextW.restype = ctypes.c_int
+    user32.IsIconic.argtypes = [wt.HWND]
+    user32.IsIconic.restype = wt.BOOL
+    user32.GetWindowRect.argtypes = [wt.HWND, ctypes.POINTER(wt.RECT)]
+    user32.GetWindowRect.restype = wt.BOOL
+    user32.MonitorFromWindow.argtypes = [wt.HWND, wt.DWORD]
+    user32.MonitorFromWindow.restype = wt.HMONITOR
+    user32.GetMonitorInfoW.argtypes = [wt.HMONITOR, ctypes.c_void_p]
+    user32.GetMonitorInfoW.restype = wt.BOOL
+
+    kernel32.OpenProcess.argtypes = [wt.DWORD, wt.BOOL, wt.DWORD]
+    kernel32.OpenProcess.restype = wt.HANDLE
+    kernel32.CloseHandle.argtypes = [wt.HANDLE]
+    kernel32.CloseHandle.restype = wt.BOOL
+    kernel32.QueryFullProcessImageNameW.argtypes = [
+        wt.HANDLE, wt.DWORD, wt.LPWSTR, ctypes.POINTER(wt.DWORD)
+    ]
+    kernel32.QueryFullProcessImageNameW.restype = wt.BOOL
+    kernel32.GetProcessTimes.argtypes = [
+        wt.HANDLE,
+        ctypes.POINTER(wt.FILETIME),
+        ctypes.POINTER(wt.FILETIME),
+        ctypes.POINTER(wt.FILETIME),
+        ctypes.POINTER(wt.FILETIME),
+    ]
+    kernel32.GetProcessTimes.restype = wt.BOOL
+
+    psapi.GetModuleFileNameExW.argtypes = [wt.HANDLE, wt.HMODULE, wt.LPWSTR, wt.DWORD]
+    psapi.GetModuleFileNameExW.restype = wt.DWORD
+
+    _win32_user32, _win32_kernel32, _win32_psapi = user32, kernel32, psapi
+    _WIN32_INITIALIZED = True
+
+
+def _probe_windows() -> ForegroundInfo:
+    """Probe foreground window state on Windows via Win32 API."""
+    import ctypes
+    import ctypes.wintypes as wt
+
+    _init_win32_apis()
+    user32 = _win32_user32
+    kernel32 = _win32_kernel32
+    psapi = _win32_psapi
+
+    hwnd = user32.GetForegroundWindow()
+    if not hwnd:
+        return ForegroundInfo(available=True, error="no foreground window")
+
+    # PID + window thread.
+    pid = wt.DWORD(0)
+    user32.GetWindowThreadProcessId(hwnd, ctypes.byref(pid))
+    pid_val = int(pid.value) if pid.value else None
+
+    # Window title — Unicode.
+    length = user32.GetWindowTextLengthW(hwnd)
+    title_buf = ctypes.create_unicode_buffer(length + 1)
+    user32.GetWindowTextW(hwnd, title_buf, length + 1)
+    window_title = title_buf.value or None
+
+    # Minimized flag.
+    is_minimized = bool(user32.IsIconic(hwnd))
+
+    # Window rect (screen coords).
+    rect = wt.RECT()
+    window_geometry: dict[str, int] | None = None
+    if user32.GetWindowRect(hwnd, ctypes.byref(rect)):
+        window_geometry = {
+            "left": int(rect.left),
+            "top": int(rect.top),
+            "right": int(rect.right),
+            "bottom": int(rect.bottom),
+            "width": int(rect.right - rect.left),
+            "height": int(rect.bottom - rect.top),
+        }
+
+    # Monitor under the window + its geometry.
+    monitor_geometry: dict[str, int] | None = None
+    monitor_id: int | None = None
+    is_fullscreen = False
+    try:
+        MONITOR_DEFAULTTONEAREST = 2
+
+        class MONITORINFO(ctypes.Structure):
+            _fields_ = [
+                ("cbSize", wt.DWORD),
+                ("rcMonitor", wt.RECT),
+                ("rcWork", wt.RECT),
+                ("dwFlags", wt.DWORD),
+            ]
+
+        hmon = user32.MonitorFromWindow(hwnd, MONITOR_DEFAULTTONEAREST)
+        if hmon:
+            mi = MONITORINFO()
+            mi.cbSize = ctypes.sizeof(mi)
+            if user32.GetMonitorInfoW(hmon, ctypes.byref(mi)):
+                monitor_geometry = {
+                    "left": int(mi.rcMonitor.left),
+                    "top": int(mi.rcMonitor.top),
+                    "right": int(mi.rcMonitor.right),
+                    "bottom": int(mi.rcMonitor.bottom),
+                    "width": int(mi.rcMonitor.right - mi.rcMonitor.left),
+                    "height": int(mi.rcMonitor.bottom - mi.rcMonitor.top),
+                }
+                # Fullscreen heuristic: window rect equals monitor rect AND
+                # not minimized. Many media players (VLC, browser fullscreen)
+                # set themselves to exactly the monitor bounds.
+                if window_geometry and not is_minimized:
+                    is_fullscreen = (
+                        window_geometry["left"] == monitor_geometry["left"]
+                        and window_geometry["top"] == monitor_geometry["top"]
+                        and window_geometry["right"] == monitor_geometry["right"]
+                        and window_geometry["bottom"] == monitor_geometry["bottom"]
+                    )
+
+            # Resolve monitor index by enumerating displays in order. Coerce
+            # both the foreground hmon and the per-enum hmon to int so the
+            # equality compare uses 64-bit values consistently regardless of
+            # how ctypes represents the handle internally.
+            try:
+                indexed: list[int] = []
+
+                def _cb(hm, _hdc, _rect, _data):
+                    indexed.append(int(hm) if hm else 0)
+                    return True
+
+                MONITORENUMPROC = ctypes.WINFUNCTYPE(
+                    ctypes.c_int,
+                    wt.HMONITOR,
+                    wt.HDC,
+                    ctypes.POINTER(wt.RECT),
+                    wt.LPARAM,
+                )
+                user32.EnumDisplayMonitors.argtypes = [
+                    wt.HDC, ctypes.POINTER(wt.RECT), MONITORENUMPROC, wt.LPARAM
+                ]
+                user32.EnumDisplayMonitors.restype = wt.BOOL
+                user32.EnumDisplayMonitors(None, None, MONITORENUMPROC(_cb), 0)
+                target = int(hmon) if hmon else 0
+                if target and target in indexed:
+                    monitor_id = indexed.index(target)
+            except Exception as e:
+                logger.debug("Monitor index resolution failed: %s", e)
+    except Exception as e:
+        logger.debug("Monitor info probe failed: %s", e)
+
+    # Process executable path + start time.
+    executable_path: str | None = None
+    process_name: str | None = None
+    started_at: float | None = None
+    if pid_val:
+        PROCESS_QUERY_LIMITED_INFORMATION = 0x1000
+        h_proc = kernel32.OpenProcess(
+            PROCESS_QUERY_LIMITED_INFORMATION, False, pid_val
+        )
+        if h_proc:
+            try:
+                # Image filename — full path. QueryFullProcessImageNameW works
+                # across 32/64-bit boundaries, unlike GetModuleFileNameExW.
+                buf = ctypes.create_unicode_buffer(1024)
+                size = wt.DWORD(len(buf))
+                if kernel32.QueryFullProcessImageNameW(
+                    h_proc, 0, buf, ctypes.byref(size)
+                ):
+                    executable_path = buf.value or None
+                else:
+                    # Fallback via psapi. Return value is the length copied
+                    # into the buffer (0 on failure); ignoring it would leave
+                    # `executable_path` as an empty string from the freshly
+                    # allocated buffer instead of None.
+                    written = psapi.GetModuleFileNameExW(h_proc, None, buf, len(buf))
+                    if written:
+                        executable_path = buf.value or None
+                    else:
+                        logger.debug(
+                            "QueryFullProcessImageNameW + psapi fallback both "
+                            "failed for pid=%s (err=%d)",
+                            pid_val,
+                            ctypes.get_last_error(),
+                        )
+
+                if executable_path:
+                    import os
+                    process_name = os.path.basename(executable_path)
+
+                # Process creation time (FILETIME, 100ns ticks since 1601).
+                creation = wt.FILETIME()
+                exit_t = wt.FILETIME()
+                kernel_t = wt.FILETIME()
+                user_t = wt.FILETIME()
+                if kernel32.GetProcessTimes(
+                    h_proc,
+                    ctypes.byref(creation),
+                    ctypes.byref(exit_t),
+                    ctypes.byref(kernel_t),
+                    ctypes.byref(user_t),
+                ):
+                    ticks = (creation.dwHighDateTime << 32) | creation.dwLowDateTime
+                    # Convert to Unix epoch seconds (1601-01-01 → 1970-01-01).
+                    if ticks:
+                        started_at = (ticks - 116444736000000000) / 10_000_000
+            finally:
+                kernel32.CloseHandle(h_proc)
+
+    return ForegroundInfo(
+        available=True,
+        pid=pid_val,
+        process_name=process_name,
+        executable_path=executable_path,
+        window_title=window_title,
+        window_handle=int(hwnd) if hwnd else None,
+        is_fullscreen=is_fullscreen,
+        is_minimized=is_minimized,
+        monitor_id=monitor_id,
+        monitor_geometry=monitor_geometry,
+        window_geometry=window_geometry,
+        started_at=started_at,
+    )
+
+
+def _probe_macos() -> ForegroundInfo:
+    """Best-effort probe on macOS via AppKit (PyObjC).
+
+    Returns ``available=False`` when PyObjC is not installed — we don't take
+    a hard dependency on it because the typical macOS install path uses pip
+    + the standalone wheel.
+    """
+    try:
+        from AppKit import NSWorkspace  # type: ignore
+        from Quartz import (  # type: ignore
+            CGWindowListCopyWindowInfo,
+            kCGNullWindowID,
+            kCGWindowListOptionOnScreenOnly,
+        )
+    except Exception:
+        return ForegroundInfo(available=False, error="AppKit/Quartz not available")
+
+    try:
+        ws = NSWorkspace.sharedWorkspace()
+        app = ws.frontmostApplication()
+        if app is None:
+            return ForegroundInfo(available=True, error="no frontmost app")
+
+        pid = int(app.processIdentifier())
+        process_name = str(app.localizedName() or "")
+        bundle_url = app.bundleURL()
+        executable_path = str(bundle_url.path()) if bundle_url else None
+        started_at = None
+        launch_date = app.launchDate()
+        if launch_date is not None:
+            started_at = float(launch_date.timeIntervalSince1970())
+
+        # Window title — frontmost on-screen window owned by this PID.
+        window_title: str | None = None
+        try:
+            windows = CGWindowListCopyWindowInfo(
+                kCGWindowListOptionOnScreenOnly, kCGNullWindowID
+            )
+            for w in windows or []:
+                if int(w.get("kCGWindowOwnerPID", -1)) == pid:
+                    name = w.get("kCGWindowName")
+                    if name:
+                        window_title = str(name)
+                        break
+        except Exception as e:
+            logger.debug("CGWindowListCopyWindowInfo failed: %s", e)
+
+        return ForegroundInfo(
+            available=True,
+            pid=pid,
+            process_name=process_name,
+            executable_path=executable_path,
+            window_title=window_title,
+            started_at=started_at,
+        )
+    except Exception as e:
+        logger.debug("macOS foreground probe failed: %s", e)
+        return ForegroundInfo(available=False, error=str(e))
+
+
+def _probe_linux() -> ForegroundInfo:
+    """Best-effort probe on Linux via Xlib (X11 only).
+
+    Wayland sessions intentionally hide window/process info from unprivileged
+    clients, so this returns ``available=False`` on Wayland. The caller still
+    gets a structured response and can render "unavailable" in the UI.
+    """
+    import os
+
+    if os.environ.get("WAYLAND_DISPLAY"):
+        return ForegroundInfo(
+            available=False, error="Wayland session — foreground probe unavailable"
+        )
+
+    try:
+        from Xlib import X, display  # type: ignore  # noqa: F401
+    except Exception:
+        return ForegroundInfo(available=False, error="python-xlib not installed")
+
+    try:
+        d = display.Display()
+        root = d.screen().root
+        NET_ACTIVE_WINDOW = d.intern_atom("_NET_ACTIVE_WINDOW")
+        NET_WM_PID = d.intern_atom("_NET_WM_PID")
+        NET_WM_NAME = d.intern_atom("_NET_WM_NAME")
+        UTF8_STRING = d.intern_atom("UTF8_STRING")
+
+        active = root.get_full_property(NET_ACTIVE_WINDOW, X.AnyPropertyType)
+        if not active or not active.value:
+            return ForegroundInfo(available=True, error="no active window")
+        win_id = int(active.value[0])
+        win = d.create_resource_object("window", win_id)
+
+        pid_prop = win.get_full_property(NET_WM_PID, X.AnyPropertyType)
+        pid_val = int(pid_prop.value[0]) if pid_prop and pid_prop.value else None
+
+        name_prop = win.get_full_property(NET_WM_NAME, UTF8_STRING)
+        window_title = (
+            name_prop.value.decode("utf-8", "replace") if name_prop and name_prop.value else None
+        )
+
+        process_name: str | None = None
+        executable_path: str | None = None
+        started_at: float | None = None
+        if pid_val:
+            try:
+                exe = os.readlink(f"/proc/{pid_val}/exe")
+                executable_path = exe
+                process_name = os.path.basename(exe)
+            except OSError as e:
+                logger.debug("readlink /proc/%d/exe failed: %s", pid_val, e)
+            try:
+                started_at = os.stat(f"/proc/{pid_val}").st_ctime
+            except OSError as e:
+                logger.debug("stat /proc/%d failed: %s", pid_val, e)
+
+        return ForegroundInfo(
+            available=True,
+            pid=pid_val,
+            process_name=process_name,
+            executable_path=executable_path,
+            window_title=window_title,
+            window_handle=win_id,
+            started_at=started_at,
+        )
+    except Exception as e:
+        logger.debug("Linux foreground probe failed: %s", e)
+        return ForegroundInfo(available=False, error=str(e))
+
+
+def _enrich_browser(info: ForegroundInfo) -> ForegroundInfo:
+    """If ``info`` describes a focused browser, attach URL + page title.
+
+    The UIA lookup is wrapped in its own try/except so a failure here can't
+    take down the rest of the foreground probe.
+    """
+    try:
+        from . import browser_url_service as bus
+    except Exception as e:
+        logger.debug("browser_url_service unavailable: %s", e)
+        return info
+
+    if not info.available or not bus.is_browser_process(info.process_name):
+        return info
+
+    try:
+        page = bus.get_browser_page(
+            hwnd=info.window_handle,
+            process_name=info.process_name,
+            window_title=info.window_title,
+        )
+    except Exception as e:
+        logger.debug("Browser URL enrichment failed: %s", e)
+        return info
+
+    # ``dataclasses.replace`` keeps the frozen-dataclass contract.
+    from dataclasses import replace
+    return replace(
+        info,
+        is_browser=True,
+        browser_url=page.url,
+        browser_page_title=page.page_title,
+    )
+
+
+def _probe() -> ForegroundInfo:
+    system = platform.system()
+    try:
+        if system == "Windows":
+            info = _probe_windows()
+        elif system == "Darwin":
+            info = _probe_macos()
+        elif system == "Linux":
+            info = _probe_linux()
+        else:
+            return ForegroundInfo(
+                available=False, error=f"unsupported platform: {system}"
+            )
+        return _enrich_browser(info)
+    except Exception as e:
+        logger.warning("Foreground probe crashed: %s", e)
+        return ForegroundInfo(available=False, error=str(e))
+
+
+def get_foreground_info(force_refresh: bool = False) -> ForegroundInfo:
+    """Return the current foreground window/process snapshot.
+
+    Args:
+        force_refresh: bypass the short TTL cache. WebSocket broadcast loop
+            should leave this False; the REST endpoint accepts ?refresh=1
+            for callers that want a fresh probe.
+    """
+    if force_refresh:
+        _cache.invalidate()
+    return _cache.get(_CACHE_TTL, _probe)
+
+
+def reset_cache() -> None:
+    """Reset the cache. Useful in tests."""
+    _cache.invalidate()

media_server/services/gitea_release_provider.py

@@ -2,6 +2,7 @@
 
 import json
 import logging
+import re
 import urllib.error
 import urllib.request
 from typing import Optional
@@ -15,6 +16,11 @@ _DEFAULT_BASE_URL = "https://git.dolgolyov-family.by"
 _DEFAULT_OWNER = "alexei.dolgolyov"
 _DEFAULT_REPO = "media-player-server"
 
+# Restrictive tag whitelist — prevents a hostile Gitea response (or MITM) from
+# injecting `..`, slashes, or URL-altering characters into the release URL we
+# broadcast to clients. SemVer + pre-release suffix only.
+_TAG_RE = re.compile(r"^v?\d+\.\d+\.\d+(?:[\w.\-+]{0,32})?$")
+
 
 class GiteaReleaseProvider(ReleaseProvider):
     """Fetches the latest release from a Gitea repository."""
@@ -53,6 +59,9 @@ class GiteaReleaseProvider(ReleaseProvider):
                 continue
 
             tag = release.get("tag_name", "")
+            if not isinstance(tag, str) or not _TAG_RE.match(tag):
+                logger.warning("Rejecting malformed release tag from upstream: %r", tag)
+                continue
             version = tag.lstrip("v")
             if not version:
                 continue

media_server/services/linux_media.py

@@ -2,14 +2,23 @@
 
 import asyncio
 import logging
+import os
 import subprocess
+import threading
 from typing import Any, Optional
+from urllib.parse import unquote, urlparse
 
 from ..models import MediaState, MediaStatus
 from .media_controller import MediaController
 
 logger = logging.getLogger(__name__)
 
+# Cap remote artwork downloads so a hostile / huge Spotify image can't
+# blow up RAM. Real album art rarely exceeds ~2 MB; 8 MB is a comfortable
+# upper bound that also covers loss-less PNGs.
+_MAX_ART_BYTES = 8 * 1024 * 1024
+_ART_FETCH_TIMEOUT = 5.0  # seconds
+
 # Linux-specific imports
 try:
     import dbus
@@ -35,13 +44,54 @@ class LinuxMediaController(MediaController):
                 "Linux media control requires dbus-python package. "
                 "Install with: sudo apt-get install python3-dbus"
             )
-        DBusGMainLoop(set_as_default=True)
+        # The session-bus connection is deferred until first use. Connecting
-        self._bus = dbus.SessionBus()
+        # in __init__ raised during app startup whenever the user's session
+        # bus wasn't ready yet — common under systemd (service starts
+        # before logind set up /run/user/<uid>/bus), under SSH-without-X11,
+        # and in headless CI. Failing here killed the whole lifespan; now
+        # MPRIS calls return "idle" until the bus appears, and other
+        # endpoints (health, scripts, browser, …) keep working.
+        self._bus_lock = threading.Lock()
+        self._bus = None  # type: ignore[assignment]
+        self._bus_init_logged = False
+        # Cached art bytes keyed by the mpris:artUrl currently in flight.
+        # Lock guards the swap from the status thread vs the artwork handler.
+        self._art_lock = threading.Lock()
+        self._art_url: Optional[str] = None
+        self._art_bytes: Optional[bytes] = None
+
+    def _get_bus(self):
+        """Lazily connect to the session bus; returns None if unavailable."""
+        if self._bus is not None:
+            return self._bus
+        with self._bus_lock:
+            if self._bus is not None:
+                return self._bus
+            try:
+                DBusGMainLoop(set_as_default=True)
+                self._bus = dbus.SessionBus()
+                logger.info("Connected to D-Bus session bus")
+                return self._bus
+            except Exception as e:
+                # Log once at INFO to avoid log spam if every status poll fails.
+                if not self._bus_init_logged:
+                    logger.info(
+                        "D-Bus session bus not available (%s). "
+                        "MPRIS calls will return 'idle' until DBUS_SESSION_BUS_ADDRESS"
+                        " is set and the bus is reachable. Under systemd, ensure"
+                        " `loginctl enable-linger <user>` is set.",
+                        e,
+                    )
+                    self._bus_init_logged = True
+                return None
 
     def _get_active_player(self) -> Optional[str]:
         """Find an active MPRIS media player on the bus."""
+        bus = self._get_bus()
+        if bus is None:
+            return None
         try:
-            bus_names = self._bus.list_names()
+            bus_names = bus.list_names()
             mpris_players = [
                 name for name in bus_names if name.startswith(self.MPRIS_PREFIX)
             ]
@@ -181,7 +231,15 @@ class LinuxMediaController(MediaController):
             if artists:
                 status.artist = str(artists[0]) if isinstance(artists, list) else str(artists)
             status.album = str(metadata.get("xesam:album", "")) or None
-            status.album_art_url = str(metadata.get("mpris:artUrl", "")) or None
+            art_url = str(metadata.get("mpris:artUrl", "")) or None
+            status.album_art_url = art_url
+            # Invalidate cached bytes when the track changes. Real fetch
+            # happens lazily in get_album_art() so the status hot path
+            # never blocks on HTTP.
+            with self._art_lock:
+                if art_url != self._art_url:
+                    self._art_url = art_url
+                    self._art_bytes = None
             length = metadata.get("mpris:length", 0)
             if length:
                 status.duration = int(length) / 1_000_000
@@ -273,3 +331,61 @@ class LinuxMediaController(MediaController):
         except Exception as e:
             logger.error(f"Failed to open file {file_path}: {e}")
             return False
+
+    def _fetch_art_sync(self, url: str) -> Optional[bytes]:
+        """Resolve an ``mpris:artUrl`` to raw bytes (file://, http(s)://).
+
+        Other schemes (data:, ftp:, …) are rejected — we only support the
+        two cases real-world MPRIS providers use. The HTTP path is capped
+        at _MAX_ART_BYTES and the file path is read with a size guard so a
+        symlink to /dev/zero can't OOM the server.
+        """
+        try:
+            parsed = urlparse(url)
+        except ValueError:
+            return None
+
+        scheme = parsed.scheme.lower()
+        if scheme == "file":
+            path = unquote(parsed.path)
+            try:
+                size = os.stat(path).st_size
+                if size <= 0 or size > _MAX_ART_BYTES:
+                    return None
+                with open(path, "rb") as f:
+                    return f.read(_MAX_ART_BYTES)
+            except OSError as e:
+                logger.debug("Could not read local art %s: %s", path, e)
+                return None
+
+        if scheme in ("http", "https"):
+            import urllib.error
+            import urllib.request
+
+            req = urllib.request.Request(url, headers={"User-Agent": "media-server/0.x"})
+            try:
+                with urllib.request.urlopen(req, timeout=_ART_FETCH_TIMEOUT) as resp:
+                    # Cap reads to defend against unbounded responses.
+                    return resp.read(_MAX_ART_BYTES + 1)[:_MAX_ART_BYTES] or None
+            except (urllib.error.URLError, urllib.error.HTTPError, OSError) as e:
+                logger.debug("Could not fetch remote art %s: %s", url, e)
+                return None
+
+        logger.debug("Unsupported art URL scheme: %s", scheme)
+        return None
+
+    async def get_album_art(self) -> Optional[bytes]:
+        """Return cached MPRIS art, fetching on first access per track."""
+        with self._art_lock:
+            url = self._art_url
+            cached = self._art_bytes
+        if cached is not None:
+            return cached
+        if not url:
+            return None
+        data = await asyncio.to_thread(self._fetch_art_sync, url)
+        # Store even on None so we don't re-hammer a 404 every second.
+        with self._art_lock:
+            if url == self._art_url:
+                self._art_bytes = data
+        return data

media_server/services/macos_media.py

@@ -3,6 +3,7 @@
 import asyncio
 import logging
 import subprocess
+import threading
 from typing import Optional
 
 from ..models import MediaState, MediaStatus
@@ -10,10 +11,20 @@ from .media_controller import MediaController
 
 logger = logging.getLogger(__name__)
 
+# Cap remote artwork downloads (Spotify's artwork url is http(s)://).
+_MAX_ART_BYTES = 8 * 1024 * 1024
+_ART_FETCH_TIMEOUT = 5.0  # seconds
+
 
 class MacOSMediaController(MediaController):
     """Media controller for macOS using osascript and system commands."""
 
+    def __init__(self) -> None:
+        # Cached art bytes keyed by the active art URL.
+        self._art_lock = threading.Lock()
+        self._art_url: Optional[str] = None
+        self._art_bytes: Optional[bytes] = None
+
     def _run_osascript(self, script: str) -> Optional[str]:
         """Run an AppleScript and return the output."""
         try:
@@ -193,12 +204,60 @@ class MacOSMediaController(MediaController):
             status.album = info.get("album")
             status.duration = info.get("duration")
             status.position = info.get("position")
-            status.album_art_url = info.get("art_url")
+            art_url = info.get("art_url")
+            status.album_art_url = art_url
+            # Track changes invalidate the cached image bytes — actual
+            # fetch happens lazily in get_album_art().
+            with self._art_lock:
+                if art_url != self._art_url:
+                    self._art_url = art_url
+                    self._art_bytes = None
         else:
             status.state = MediaState.IDLE
 
         return status
 
+    def _fetch_art_sync(self, url: str) -> Optional[bytes]:
+        """Resolve a Spotify/Music art URL (http(s)://) to bytes.
+
+        File-scheme URLs aren't expected on macOS (AppleScript apps return
+        artwork as remote URLs), so only http(s) is supported.
+        """
+        try:
+            from urllib.parse import urlparse
+            parsed = urlparse(url)
+        except ValueError:
+            return None
+
+        if parsed.scheme.lower() not in ("http", "https"):
+            return None
+
+        import urllib.error
+        import urllib.request
+
+        req = urllib.request.Request(url, headers={"User-Agent": "media-server/0.x"})
+        try:
+            with urllib.request.urlopen(req, timeout=_ART_FETCH_TIMEOUT) as resp:
+                return resp.read(_MAX_ART_BYTES + 1)[:_MAX_ART_BYTES] or None
+        except (urllib.error.URLError, urllib.error.HTTPError, OSError) as e:
+            logger.debug("Could not fetch macOS art %s: %s", url, e)
+            return None
+
+    async def get_album_art(self) -> Optional[bytes]:
+        """Return cached art bytes, fetching once per track URL."""
+        with self._art_lock:
+            url = self._art_url
+            cached = self._art_bytes
+        if cached is not None:
+            return cached
+        if not url:
+            return None
+        data = await asyncio.to_thread(self._fetch_art_sync, url)
+        with self._art_lock:
+            if url == self._art_url:
+                self._art_bytes = data
+        return data
+
     async def play(self) -> bool:
         """Resume playback using media key simulation."""
         # Use system media key
@@ -264,8 +323,12 @@ class MacOSMediaController(MediaController):
 
     async def set_volume(self, volume: int) -> bool:
         """Set system volume."""
-        result = self._run_osascript(f"set volume output volume {volume}")
+        # osascript returns empty string on success and None on failure (the
-        return result is not None or True  # osascript returns empty on success
+        # _run_osascript helper catches subprocess errors). The previous
+        # `result is not None or True` always returned True regardless of
+        # outcome — surface real failures so the route can return 503.
+        result = self._run_osascript(f"set volume output volume {int(volume)}")
+        return result is not None
 
     async def toggle_mute(self) -> bool:
         """Toggle mute state."""

media_server/services/media_controller.py

@@ -106,3 +106,12 @@ class MediaController(ABC):
             True if successful, False otherwise
         """
         pass
+
+    async def get_album_art(self) -> bytes | None:
+        """Return the current album art bytes, or ``None`` when unavailable.
+
+        Default impl returns ``None`` — controllers that can produce art
+        (Windows via SMTC thumbnail, Linux via mpris:artUrl, macOS via the
+        Spotify/Music artwork-url field) override this.
+        """
+        return None

media_server/services/rate_limit.py

@@ -0,0 +1,95 @@
+"""In-process token-bucket rate limiter.
+
+Light enough for a single-process app: one dict keyed by ``(bucket, peer)``
+guarded by a thread lock. No extra dependency, no Redis. Good enough for
+defeating credential-stuffing and runaway clients on a LAN; not a substitute
+for an upstream WAF in a public deployment.
+
+Buckets:
+    auth        — failed-auth attempts, 5/min/peer (used in auth middleware)
+    execute     — script + callback execute calls, 10/min/peer (LAN-friendly)
+    default     — generic POST/DELETE writes, 60/min/peer
+"""
+
+from __future__ import annotations
+
+import logging
+import threading
+import time
+from dataclasses import dataclass
+from typing import Optional
+
+logger = logging.getLogger(__name__)
+
+
+@dataclass
+class BucketConfig:
+    capacity: float       # max tokens (= burst size)
+    refill_per_sec: float # tokens added per second
+
+
+# Defaults — tuned for "trusted LAN" use; operator can override via Settings.
+BUCKETS: dict[str, BucketConfig] = {
+    "auth":    BucketConfig(capacity=5,  refill_per_sec=5 / 60),     # 5/min
+    "execute": BucketConfig(capacity=10, refill_per_sec=10 / 60),    # 10/min
+    "default": BucketConfig(capacity=60, refill_per_sec=60 / 60),    # 60/min
+}
+
+
+_state: dict[tuple[str, str], tuple[float, float]] = {}
+_lock = threading.Lock()
+_LAST_CLEANUP = 0.0
+
+
+def _evict_stale_locked(now: float) -> None:
+    """Drop entries whose buckets are full (= idle for capacity / refill seconds)."""
+    global _LAST_CLEANUP
+    if now - _LAST_CLEANUP < 60:
+        return
+    _LAST_CLEANUP = now
+    stale = []
+    for key, (tokens, last) in _state.items():
+        bucket = BUCKETS.get(key[0])
+        if bucket is None:
+            continue
+        if tokens >= bucket.capacity and (now - last) > 3600:
+            stale.append(key)
+    for key in stale:
+        _state.pop(key, None)
+
+
+def check(bucket: str, peer: str) -> tuple[bool, Optional[float]]:
+    """Try to consume one token from ``(bucket, peer)``.
+
+    Returns:
+        (allowed, retry_after_seconds). When allowed=True retry_after is None.
+        When allowed=False, retry_after is the seconds to wait for one more token.
+    """
+    cfg = BUCKETS.get(bucket) or BUCKETS["default"]
+    now = time.monotonic()
+    with _lock:
+        _evict_stale_locked(now)
+        tokens, last = _state.get((bucket, peer), (cfg.capacity, now))
+        elapsed = max(0.0, now - last)
+        tokens = min(cfg.capacity, tokens + elapsed * cfg.refill_per_sec)
+        if tokens >= 1:
+            tokens -= 1
+            _state[(bucket, peer)] = (tokens, now)
+            return True, None
+        deficit = 1 - tokens
+        retry = deficit / cfg.refill_per_sec if cfg.refill_per_sec > 0 else 60
+        _state[(bucket, peer)] = (tokens, now)
+        return False, retry
+
+
+def get_peer(request) -> str:
+    """Best-effort peer identifier from a Starlette request.
+
+    Honors X-Forwarded-For (only when settings.proxy_headers is True, which is
+    already enforced by uvicorn's middleware) so a reverse-proxied install
+    still rate-limits per real client.
+    """
+    client = getattr(request, "client", None)
+    if client and client.host:
+        return client.host
+    return "unknown"

media_server/services/thumbnail_service.py

@@ -26,12 +26,23 @@ class ThumbnailService:
     def get_cache_dir() -> Path:
         """Get the thumbnail cache directory path.
 
-        Returns:
+        Returns user-writable platform cache dir so installs under
-            Path to the cache directory (project-local).
+        ``%PROGRAMFILES%`` / ``/opt`` work without elevated permissions.
+        Mirrors the platform branching of ``config.get_config_dir``.
         """
-        # Store cache in project directory: media-server/.cache/thumbnails/
+        import os
-        project_root = Path(__file__).parent.parent.parent
+
-        cache_dir = project_root / ".cache" / "thumbnails"
+        if os.name == "nt":
+            # %LOCALAPPDATA% so the cache survives roaming-profile sync.
+            base = Path(os.environ.get("LOCALAPPDATA")
+                        or os.environ.get("APPDATA")
+                        or Path.home() / "AppData" / "Local")
+            cache_dir = base / "media-server" / "cache" / "thumbnails"
+        else:
+            # XDG_CACHE_HOME convention; falls back to ~/.cache.
+            xdg = os.environ.get("XDG_CACHE_HOME")
+            base = Path(xdg) if xdg else Path.home() / ".cache"
+            cache_dir = base / "media-server" / "thumbnails"
 
         cache_dir.mkdir(parents=True, exist_ok=True)
         return cache_dir

media_server/services/websocket_manager.py

@@ -19,6 +19,9 @@ class ConnectionManager:
         self._active_connections: set[WebSocket] = set()
         self._lock = asyncio.Lock()
         self._last_status: dict[str, Any] | None = None
+        self._last_foreground: dict[str, Any] | None = None
+        self._foreground_poll_interval: float = 1.0
+        self._last_foreground_poll: float = 0.0
         self._get_status_func: Callable[[], Coroutine[Any, Any, Any]] | None = None
         self._broadcast_task: asyncio.Task | None = None
         self._poll_interval: float = 0.5  # Internal poll interval for change detection
@@ -30,9 +33,15 @@ class ConnectionManager:
         self._audio_task: asyncio.Task | None = None
         self._audio_analyzer = None
 
-    async def connect(self, websocket: WebSocket) -> None:
+    async def connect(self, websocket: WebSocket, already_accepted: bool = False) -> None:
-        """Accept a new WebSocket connection."""
+        """Accept a new WebSocket connection.
-        await websocket.accept()
+
+        ``already_accepted=True`` is for callers that needed to call
+        ``websocket.accept(subprotocol=...)`` themselves (token-via-subprotocol
+        auth path).
+        """
+        if not already_accepted:
+            await websocket.accept()
         async with self._lock:
             self._active_connections.add(websocket)
         logger.info(
@@ -54,6 +63,18 @@ class ConnectionManager:
             except Exception as e:
                 logger.debug("Failed to send initial status: %s", e)
 
+        # Push a fresh foreground snapshot on connect so the UI can render
+        # the tile immediately instead of waiting for the next change.
+        try:
+            from .foreground_service import get_foreground_info
+
+            fg = await asyncio.to_thread(get_foreground_info)
+            fg_dict = fg.to_dict()
+            self._last_foreground = fg_dict
+            await websocket.send_json({"type": "foreground", "data": fg_dict})
+        except Exception as e:
+            logger.debug("Failed to send initial foreground snapshot: %s", e)
+
     async def disconnect(self, websocket: WebSocket) -> None:
         """Remove a WebSocket connection. Stops audio capture if last visualizer subscriber."""
         should_stop = False
@@ -115,6 +136,35 @@ class ConnectionManager:
         await self.broadcast(message)
         logger.info("Broadcast sent: links_changed")
 
+    def foreground_changed(
+        self, old: dict[str, Any] | None, new: dict[str, Any]
+    ) -> bool:
+        """Detect a meaningful change in the foreground process snapshot.
+
+        The probe also returns ``window_geometry`` which jitters on every
+        pixel of cursor drag — comparing the whole dict would flood clients.
+        We only diff the fields a user (or HA automation) would actually act
+        on. ``window_geometry``/``monitor_geometry``/``started_at`` are still
+        delivered in the payload, but they don't drive broadcast cadence.
+        """
+        if old is None:
+            return True
+        diff_fields = (
+            "pid",
+            "process_name",
+            "executable_path",
+            "window_title",
+            "is_fullscreen",
+            "is_minimized",
+            "monitor_id",
+            "available",
+            "error",
+        )
+        for f in diff_fields:
+            if old.get(f) != new.get(f):
+                return True
+        return False
+
     async def subscribe_visualizer(self, websocket: WebSocket) -> None:
         """Subscribe a client to audio visualizer data. Starts capture on first subscriber."""
         should_start = False
@@ -314,6 +364,10 @@ class ConnectionManager:
         get_status_func: Callable[[], Coroutine[Any, Any, Any]],
     ) -> None:
         """Background loop that polls for status changes and broadcasts."""
+        # Foreground tracker is imported lazily so unit tests of the WS
+        # manager don't drag in platform-specific probe code.
+        from .foreground_service import get_foreground_info
+
         while self._running:
             try:
                 # Only poll if we have connected clients
@@ -340,6 +394,28 @@ class ConnectionManager:
                     # Update cached status even without broadcast
                     self._last_status = status_dict
 
+                # Foreground process — poll at a coarser interval than media
+                # status. Broadcasts only fire on a real change, so a quiet
+                # desktop costs nothing.
+                now = time.time()
+                if (
+                    now - self._last_foreground_poll
+                ) >= self._foreground_poll_interval:
+                    self._last_foreground_poll = now
+                    try:
+                        fg = await asyncio.to_thread(get_foreground_info)
+                        fg_dict = fg.to_dict()
+                        if self.foreground_changed(self._last_foreground, fg_dict):
+                            self._last_foreground = fg_dict
+                            await self.broadcast(
+                                {"type": "foreground_update", "data": fg_dict}
+                            )
+                            logger.debug("Broadcast sent: foreground change")
+                        else:
+                            self._last_foreground = fg_dict
+                    except Exception as e:
+                        logger.debug("Foreground poll failed: %s", e)
+
                 await asyncio.sleep(self._poll_interval)
 
             except asyncio.CancelledError:

media_server/services/windows_media.py

@@ -31,8 +31,15 @@ def _thread_loop() -> asyncio.AbstractEventLoop:
         _thread_local.loop = loop
     return loop
 
-# Global storage for current album art (as bytes)
+# Global storage for current album art (as bytes). Guarded by _art_lock so the
+# WinRT polling thread and the FastAPI handler thread don't race on swap.
 _current_album_art_bytes: bytes | None = None
+_art_lock = threading.Lock()
+
+# Identity of the track whose art is currently in _current_album_art_bytes.
+# Used to gate the expensive WinRT thumbnail.open_read_async() so the bytes
+# aren't re-decoded on every 500ms status poll.
+_current_album_art_key: tuple | None = None
 
 # Lock protecting _position_cache and _track_skip_pending from concurrent access
 _position_lock = threading.Lock()
@@ -56,8 +63,9 @@ _track_skip_pending = {
 
 
 def get_current_album_art() -> bytes | None:
-    """Get the current album art bytes."""
+    """Get the current album art bytes (thread-safe snapshot)."""
-    return _current_album_art_bytes
+    with _art_lock:
+        return _current_album_art_bytes
 
 # Windows-specific imports
 try:
@@ -379,28 +387,48 @@ def _sync_get_media_status() -> dict[str, Any]:
                 except Exception as e:
                     logger.debug(f"Timeline parse error: {e}")
 
-            # Try to get album art (requires media_props)
+            # Try to get album art (requires media_props). Gated by track key so
+            # the WinRT IPC + bytes copy only runs when the track actually
+            # changes; otherwise we just preserve the existing cached bytes.
             if media_props:
-                try:
+                track_key = (
-                    thumbnail = media_props.thumbnail
+                    getattr(media_props, "title", "") or "",
-                    if thumbnail:
+                    getattr(media_props, "artist", "") or "",
-                        stream = loop.run_until_complete(thumbnail.open_read_async())
+                    getattr(media_props, "album_title", "") or "",
-                        if stream:
+                )
-                            size = stream.size
+                global _current_album_art_bytes, _current_album_art_key
-                            if size > 0 and size < 10 * 1024 * 1024:  # Max 10MB
+                if track_key == _current_album_art_key and _current_album_art_bytes:
-                                from winsdk.windows.storage.streams import DataReader
+                    # Same track — reuse cached art bytes without touching WinRT.
-                                reader = DataReader(stream)
+                    result["album_art_url"] = "/api/media/artwork"
-                                loop.run_until_complete(reader.load_async(size))
+                else:
-                                buffer = bytearray(size)
+                    try:
-                                reader.read_bytes(buffer)
+                        thumbnail = media_props.thumbnail
-                                reader.close()
+                        if thumbnail:
-                                stream.close()
+                            stream = loop.run_until_complete(thumbnail.open_read_async())
+                            if stream:
+                                size = stream.size
+                                if size > 0 and size < 10 * 1024 * 1024:  # Max 10MB
+                                    from winsdk.windows.storage.streams import DataReader
+                                    reader = DataReader(stream)
+                                    loop.run_until_complete(reader.load_async(size))
+                                    buffer = bytearray(size)
+                                    reader.read_bytes(buffer)
+                                    reader.close()
+                                    stream.close()
 
-                                global _current_album_art_bytes
+                                    with _art_lock:
-                                _current_album_art_bytes = bytes(buffer)
+                                        _current_album_art_bytes = bytes(buffer)
-                                result["album_art_url"] = "/api/media/artwork"
+                                        _current_album_art_key = track_key
-                except Exception as e:
+                                    result["album_art_url"] = "/api/media/artwork"
-                    logger.debug(f"Failed to get album art: {e}")
+                        else:
+                            # No thumbnail on this track — drop stale bytes so
+                            # the ETag flips and clients don't keep showing the
+                            # previous album's cover.
+                            with _art_lock:
+                                _current_album_art_bytes = None
+                                _current_album_art_key = track_key
+                    except Exception as e:
+                        logger.debug(f"Failed to get album art: {e}")
 
             result["source"] = session.source_app_user_model_id
 

media_server/static/css/styles.css

@@ -9321,3 +9321,321 @@ body.is-fullscreen-player .now-playing .vu-meter {
     body.is-fullscreen-player .fs-bloom #fs-bloom-art { animation: none !important; }
     :root[data-theme="light"] body.is-fullscreen-player .fs-bloom { opacity: 0.22; }
 }
+
+/* ════════════════════════════════════════════════════════════════
+   FOREGROUND container — editorial process plate
+   ════════════════════════════════════════════════════════════════ */
+.foreground-container {
+    background: transparent;
+    border: 0;
+    padding: 0;
+    box-shadow: none;
+    margin-top: 28px;
+}
+
+.foreground-stage {
+    min-height: 360px;
+}
+
+/* Match the inter-section gap used between .settings-section blocks
+   in the Settings tab — keeps cadence consistent across tabs. */
+.display-container > * + * {
+    margin-top: 16px;
+}
+
+.foreground-card {
+    position: relative;
+    display: block;
+    padding: clamp(24px, 3vw, 40px) clamp(24px, 3vw, 40px) 28px;
+    border: 1px solid var(--rule);
+    border-top: 2px solid var(--copper);
+    background:
+        radial-gradient(120% 80% at 0% 0%, rgba(var(--copper-rgb), 0.05), transparent 60%),
+        var(--bg-paper);
+    box-shadow:
+        0 1px 0 var(--bg-paper),
+        0 28px 60px -28px rgba(0, 0, 0, 0.45),
+        0 8px 20px -10px rgba(0, 0, 0, 0.25);
+}
+.foreground-card[data-fullscreen="1"] {
+    border-top-color: var(--copper-hi);
+    box-shadow:
+        0 1px 0 var(--bg-paper),
+        0 28px 60px -28px rgba(0, 0, 0, 0.55),
+        0 0 0 1px rgba(var(--copper-rgb), 0.18),
+        0 0 60px -12px var(--copper-glow);
+}
+
+.foreground-card .fg-kicker {
+    display: flex;
+    align-items: center;
+    gap: 14px;
+    font-family: var(--mono);
+    font-size: 10px;
+    letter-spacing: 0.32em;
+    text-transform: uppercase;
+    color: var(--copper);
+    margin-bottom: 22px;
+}
+.foreground-card .fg-kicker::before,
+.foreground-card .fg-kicker::after {
+    content: "";
+    height: 1px;
+    background: var(--copper);
+    opacity: 0.6;
+    flex: 0 0 24px;
+}
+.foreground-card .fg-kicker::after { flex: 1 0 auto; }
+
+.foreground-card .fg-process {
+    font-family: var(--serif);
+    font-weight: 400;
+    font-size: clamp(34px, 4.4vw, 56px);
+    line-height: 1.02;
+    letter-spacing: -0.02em;
+    font-variation-settings: 'opsz' 144;
+    color: var(--ink);
+    margin: 0 0 10px;
+    word-break: break-word;
+    overflow-wrap: anywhere;
+    transition: color 180ms var(--ease, ease);
+}
+.foreground-card .fg-process:hover {
+    color: var(--copper-hi);
+}
+
+.foreground-card .fg-window-title {
+    font-family: var(--serif);
+    font-style: italic;
+    font-size: 20px;
+    font-weight: 300;
+    color: var(--ink-soft);
+    font-variation-settings: 'opsz' 60;
+    margin-bottom: 22px;
+    line-height: 1.35;
+    overflow: hidden;
+    text-overflow: ellipsis;
+    display: -webkit-box;
+    -webkit-line-clamp: 2;
+    line-clamp: 2;
+    -webkit-box-orient: vertical;
+    word-break: break-word;
+}
+.foreground-card .fg-window-title:empty { display: none; }
+
+.foreground-card .fg-chips {
+    display: flex;
+    flex-wrap: wrap;
+    gap: 8px;
+    margin-bottom: 28px;
+}
+.foreground-card .fg-chips:empty { display: none; }
+
+.fg-chip {
+    display: inline-flex;
+    align-items: center;
+    padding: 5px 11px;
+    font-family: var(--mono);
+    font-size: 10px;
+    letter-spacing: 0.18em;
+    text-transform: uppercase;
+    color: var(--ink-soft);
+    background: transparent;
+    border: 1px solid var(--rule-strong);
+    border-radius: 999px;
+    line-height: 1.2;
+    white-space: nowrap;
+}
+.fg-chip.fg-chip-accent {
+    color: var(--copper);
+    border-color: var(--copper);
+    background: rgba(var(--copper-rgb), 0.07);
+}
+.fg-chip.fg-chip-mute {
+    color: var(--ink-mute);
+    border-color: var(--rule);
+}
+
+.foreground-card .fg-details {
+    display: block;
+    margin: 0;
+    border-top: 1px solid var(--rule);
+}
+.foreground-card .fg-row {
+    display: grid;
+    grid-template-columns: minmax(160px, 220px) 1fr;
+    gap: 24px;
+    padding: 14px 0;
+    border-bottom: 1px solid var(--rule);
+    align-items: baseline;
+    min-width: 0;
+}
+.foreground-card .fg-row dt {
+    font-family: var(--mono);
+    font-size: 9px;
+    letter-spacing: 0.22em;
+    text-transform: uppercase;
+    color: var(--copper);
+    margin: 0;
+}
+.foreground-card .fg-row dd {
+    font-family: var(--serif);
+    font-style: italic;
+    font-size: 18px;
+    color: var(--ink);
+    font-variation-settings: 'opsz' 30;
+    margin: 0;
+    min-width: 0;
+    overflow: hidden;
+    text-overflow: ellipsis;
+}
+.foreground-card .fg-mono {
+    font-family: var(--mono);
+    font-style: normal;
+    font-size: 13px;
+    letter-spacing: 0.02em;
+    color: var(--ink-soft);
+    font-variant-numeric: tabular-nums;
+    word-break: break-all;
+}
+
+.foreground-empty {
+    padding: 60px 24px;
+    text-align: center;
+    color: var(--ink-mute);
+}
+.foreground-empty svg {
+    width: 64px;
+    height: 64px;
+    margin-bottom: 14px;
+    opacity: 0.55;
+    color: var(--ink-faint);
+}
+.foreground-empty p {
+    font-family: var(--serif);
+    font-style: italic;
+    font-size: 18px;
+    color: var(--ink-soft);
+    margin: 0;
+}
+.foreground-empty .foreground-empty-error {
+    margin-top: 10px;
+    font-family: var(--mono);
+    font-style: normal;
+    font-size: 11px;
+    letter-spacing: 0.06em;
+    color: var(--ink-mute);
+    word-break: break-word;
+}
+
+/* ─── Header status badge ──────────────────────────────────── */
+.foreground-status-badge {
+    display: inline-flex;
+    align-items: center;
+    gap: 8px;
+    height: 32px;
+    padding: 0 12px 0 10px;
+    margin-right: 4px;
+    background: transparent;
+    border: 1px solid var(--rule-strong);
+    border-radius: 999px;
+    color: var(--ink-soft);
+    font-family: var(--mono);
+    font-size: 11px;
+    letter-spacing: 0.04em;
+    cursor: pointer;
+    max-width: 240px;
+    transition: color 180ms ease, border-color 180ms ease, background 180ms ease;
+}
+.foreground-status-badge:hover {
+    color: var(--ink);
+    border-color: var(--copper);
+    background: rgba(var(--copper-rgb), 0.06);
+}
+.foreground-status-badge.hidden { display: none !important; }
+
+.foreground-status-badge .fg-badge-mark {
+    width: 6px;
+    height: 6px;
+    border-radius: 50%;
+    background: var(--ink-mute);
+    flex-shrink: 0;
+}
+.foreground-status-badge.is-media .fg-badge-mark,
+.foreground-status-badge.is-fullscreen .fg-badge-mark {
+    background: var(--copper);
+    box-shadow: 0 0 8px var(--copper-glow);
+}
+.foreground-status-badge.is-fullscreen {
+    border-color: var(--copper);
+    color: var(--ink);
+}
+
+.foreground-status-badge .fg-badge-name {
+    overflow: hidden;
+    text-overflow: ellipsis;
+    white-space: nowrap;
+    max-width: 140px;
+}
+.foreground-status-badge .fg-badge-tag {
+    color: var(--copper);
+    text-transform: uppercase;
+    letter-spacing: 0.14em;
+    font-size: 10px;
+    flex-shrink: 0;
+}
+.foreground-status-badge .fg-badge-tag.hidden { display: none; }
+
+/* ─── Light theme overrides ──────────────────────────────── */
+:root[data-theme="light"] .foreground-card {
+    background:
+        radial-gradient(120% 80% at 0% 0%, rgba(var(--copper-rgb), 0.05), transparent 60%),
+        var(--bg-paper);
+    box-shadow:
+        0 1px 0 var(--bg-paper),
+        0 22px 50px -24px rgba(26, 23, 21, 0.20),
+        0 6px 16px -8px rgba(26, 23, 21, 0.12);
+}
+:root[data-theme="light"] .foreground-card[data-fullscreen="1"] {
+    box-shadow:
+        0 1px 0 var(--bg-paper),
+        0 22px 50px -24px rgba(26, 23, 21, 0.28),
+        0 0 0 1px rgba(var(--copper-rgb), 0.20),
+        0 0 50px -12px var(--copper-glow);
+}
+:root[data-theme="light"] .foreground-status-badge {
+    border-color: rgba(26, 23, 21, 0.18);
+}
+:root[data-theme="light"] .foreground-status-badge:hover {
+    background: rgba(var(--copper-rgb), 0.08);
+}
+
+/* ─── Mobile breakpoint ──────────────────────────────────── */
+@media (max-width: 720px) {
+    .foreground-card {
+        padding: 22px 18px 20px;
+    }
+    .foreground-card .fg-process {
+        font-size: 30px;
+    }
+    .foreground-card .fg-window-title {
+        font-size: 16px;
+    }
+    .foreground-card .fg-row {
+        grid-template-columns: 1fr;
+        gap: 4px;
+        padding: 12px 0;
+    }
+    .foreground-card .fg-row dd {
+        font-size: 16px;
+    }
+    .foreground-status-badge {
+        max-width: 160px;
+    }
+    .foreground-status-badge .fg-badge-name {
+        max-width: 80px;
+    }
+    .foreground-status-badge .fg-badge-tag {
+        display: none;
+    }
+}

media_server/static/icons/icon.svg

@@ -1,10 +1,33 @@
-<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 100 100">
+<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 256 256" role="img" aria-label="Media Server">
   <defs>
-    <linearGradient id="grad" x1="0%" y1="0%" x2="100%" y2="100%">
+    <linearGradient id="bg" x1="0%" y1="0%" x2="100%" y2="100%">
-      <stop offset="0%" style="stop-color:#1db954;stop-opacity:1" />
+      <stop offset="0%"  stop-color="#0B3D3B"/>
-      <stop offset="100%" style="stop-color:#1ed760;stop-opacity:1" />
+      <stop offset="100%" stop-color="#1A6B5E"/>
     </linearGradient>
+    <linearGradient id="sheen" x1="50%" y1="0%" x2="50%" y2="100%">
+      <stop offset="0%"  stop-color="#FFFFFF" stop-opacity="0.18"/>
+      <stop offset="55%" stop-color="#FFFFFF" stop-opacity="0"/>
+    </linearGradient>
+    <filter id="triShadow" x="-20%" y="-20%" width="140%" height="140%">
+      <feGaussianBlur stdDeviation="3"/>
+      <feOffset dx="3" dy="5"/>
+      <feComponentTransfer><feFuncA type="linear" slope="0.45"/></feComponentTransfer>
+      <feMerge><feMergeNode/><feMergeNode in="SourceGraphic"/></feMerge>
+    </filter>
+    <clipPath id="clip"><rect x="0" y="0" width="256" height="256" rx="58" ry="58"/></clipPath>
   </defs>
-  <circle cx="50" cy="50" r="45" fill="url(#grad)"/>
+  <g clip-path="url(#clip)">
-  <path fill="white" d="M35 25 L35 75 L75 50 Z"/>
+    <rect width="256" height="256" fill="url(#bg)"/>
+    <rect width="256" height="256" fill="url(#sheen)"/>
+    <g stroke="#F5F1E8" stroke-width="4.6" stroke-linecap="round" stroke-linejoin="round" fill="none" opacity="0.18">
+      <polyline points="66,72 105,128 66,184"/>
+      <polyline points="85,82 124,128 85,174" opacity="0.6"/>
+    </g>
+    <path d="M88.3 55 L88.3 201 L193.3 128 Z"
+          fill="#F5F1E8"
+          stroke="#F5F1E8" stroke-width="9" stroke-linejoin="round"
+          filter="url(#triShadow)"/>
+  </g>
+  <rect x="0.5" y="0.5" width="255" height="255" rx="58" ry="58"
+        fill="none" stroke="#000000" stroke-opacity="0.18" stroke-width="1"/>
 </svg>

media_server/static/index.html

@@ -26,16 +26,16 @@
             </div>
         </div>
         <div class="mini-controls">
-            <button class="mini-control-btn mini-nav-btn" data-onclick="previousTrack()" data-i18n-title="player.previous" title="Previous">
+            <button class="mini-control-btn mini-nav-btn" data-onclick="previousTrack()" data-i18n-title="player.previous" data-i18n-aria-label="player.previous" title="Previous" aria-label="Previous">
-                <svg viewBox="0 0 24 24"><path d="M6 6h2v12H6zm3.5 6l8.5 6V6z"/></svg>
+                <svg viewBox="0 0 24 24" aria-hidden="true" focusable="false"><path d="M6 6h2v12H6zm3.5 6l8.5 6V6z"/></svg>
             </button>
-            <button class="mini-control-btn" data-onclick="togglePlayPause()" id="mini-btn-play-pause" title="Play/Pause">
+            <button class="mini-control-btn" data-onclick="togglePlayPause()" id="mini-btn-play-pause" title="Play/Pause" aria-label="Play/Pause">
-                <svg viewBox="0 0 24 24" id="mini-play-pause-icon">
+                <svg viewBox="0 0 24 24" id="mini-play-pause-icon" aria-hidden="true" focusable="false">
                     <path d="M8 5v14l11-7z"/>
                 </svg>
             </button>
-            <button class="mini-control-btn mini-nav-btn" data-onclick="nextTrack()" data-i18n-title="player.next" title="Next">
+            <button class="mini-control-btn mini-nav-btn" data-onclick="nextTrack()" data-i18n-title="player.next" data-i18n-aria-label="player.next" title="Next" aria-label="Next">
-                <svg viewBox="0 0 24 24"><path d="M6 18l8.5-6L6 6v12zM16 6v12h2V6h-2z"/></svg>
+                <svg viewBox="0 0 24 24" aria-hidden="true" focusable="false"><path d="M6 18l8.5-6L6 6v12zM16 6v12h2V6h-2z"/></svg>
             </button>
         </div>
         <div class="mini-progress-container">
@@ -48,8 +48,8 @@
             </div>
         </div>
         <div class="mini-volume-container">
-            <button class="mini-control-btn" data-onclick="toggleMute()" id="mini-btn-mute" title="Mute">
+            <button class="mini-control-btn" data-onclick="toggleMute()" id="mini-btn-mute" title="Mute" aria-label="Mute" aria-pressed="false">
-                <svg viewBox="0 0 24 24" id="mini-mute-icon">
+                <svg viewBox="0 0 24 24" id="mini-mute-icon" aria-hidden="true" focusable="false">
                     <path d="M3 9v6h4l5 5V4L7 9H3zm13.5 3c0-1.77-1.02-3.29-2.5-4.03v8.05c1.48-.73 2.5-2.25 2.5-4.02z"/>
                 </svg>
             </button>
@@ -88,7 +88,7 @@
             </div>
             <div class="header-toolbar">
                 <div id="headerLinks" class="header-links"></div>
-                <a class="header-btn" href="/docs" target="_blank" title="API Documentation" aria-label="API Documentation">
+                <a class="header-btn" href="/docs" target="_blank" rel="noopener noreferrer" referrerpolicy="no-referrer" title="API Documentation" aria-label="API Documentation">
                     <svg viewBox="0 0 24 24"><path fill="currentColor" d="M14 2H6c-1.1 0-2 .9-2 2v16c0 1.1.9 2 2 2h12c1.1 0 2-.9 2-2V8l-6-6zm-1 2l5 5h-5V4zM6 20V4h5v7h7v9H6zm2-4h8v2H8v-2zm0-3h8v2H8v-2z"/></svg>
                 </a>
                 <button class="header-btn" data-onclick="showAboutDialog()" data-i18n-title="about.button_title" title="About" aria-label="About">
@@ -535,7 +535,7 @@
             </details>
         </div>
 
-        <!-- Display Control Section -->
+        <!-- Display Control Section (monitors first, foreground overview below) -->
         <div class="display-container" data-tab-content="display" role="tabpanel" id="panel-display">
             <div class="display-monitors" id="displayMonitors">
                 <div class="empty-state-illustration">
@@ -543,6 +543,12 @@
                     <p data-i18n="display.loading">Loading monitors...</p>
                 </div>
             </div>
+            <div class="foreground-stage" id="foregroundStage">
+                <div class="empty-state-illustration">
+                    <svg viewBox="0 0 64 64"><rect x="6" y="12" width="52" height="36" rx="3" fill="none" stroke="currentColor" stroke-width="2"/><line x1="6" y1="20" x2="58" y2="20" stroke="currentColor" stroke-width="2"/><circle cx="11" cy="16" r="1.4" fill="currentColor"/><circle cx="15" cy="16" r="1.4" fill="currentColor"/><circle cx="19" cy="16" r="1.4" fill="currentColor"/></svg>
+                    <p data-i18n="foreground.loading">Waiting for foreground signal…</p>
+                </div>
+            </div>
         </div>
     </div>
 

media_server/static/js/app.js

@@ -74,6 +74,10 @@ import {
     toggleDynamicBackground, applyDynamicBackground, updateBackgroundColors,
 } from './background.js';
 
+import {
+    updateForegroundUI, loadForegroundProcess,
+} from './foreground.js';
+
 // ============================================================
 // Register late-bound callbacks for core's updateAllText()
 // ============================================================
@@ -136,6 +140,8 @@ Object.assign(window, {
     onAudioDeviceChanged,
     // About
     showAboutDialog, closeAboutDialog,
+    // Foreground
+    loadForegroundProcess,
 });
 
 // ============================================================

media_server/static/js/foreground.js

@@ -0,0 +1,188 @@
+// ============================================================
+// Foreground: Currently-focused desktop process card (rendered at
+// the top of the Display tab)
+// ============================================================
+
+import { t } from './core.js';
+
+let latestForeground = null;
+let agoTickTimer = null;
+
+function escapeHtml(s) {
+    if (s === null || s === undefined) return '';
+    return String(s)
+        .replace(/&/g, '&')
+        .replace(/</g, '&lt;')
+        .replace(/>/g, '&gt;')
+        .replace(/"/g, '&quot;')
+        .replace(/'/g, '&#39;');
+}
+
+function formatAgo(epoch) {
+    if (!epoch) return '';
+    const now = Date.now() / 1000;
+    const diff = Math.max(0, now - epoch);
+    if (diff < 60) {
+        return t('foreground.ago.seconds', { n: Math.floor(diff) });
+    }
+    if (diff < 3600) {
+        return t('foreground.ago.minutes', { n: Math.floor(diff / 60) });
+    }
+    if (diff < 86400) {
+        const h = Math.floor(diff / 3600);
+        const m = Math.floor((diff % 3600) / 60);
+        return t('foreground.ago.hours', { n: h, m: m });
+    }
+    return t('foreground.ago.days', { n: Math.floor(diff / 86400) });
+}
+
+function formatGeometry(g) {
+    if (!g) return '—';
+    const w = g.width ?? (g.right - g.left);
+    const h = g.height ?? (g.bottom - g.top);
+    return `${w}×${h} @ (${g.left}, ${g.top})`;
+}
+
+function truncatePath(p, max = 64) {
+    if (!p) return '';
+    if (p.length <= max) return p;
+    // Keep the tail (filename) visible — that's the part the user cares about.
+    return '…' + p.slice(-(max - 1));
+}
+
+function renderEmpty(message, errorMsg) {
+    const stage = document.getElementById('foregroundStage');
+    if (!stage) return;
+    stage.innerHTML = `
+        <div class="empty-state-illustration foreground-empty">
+            <svg viewBox="0 0 64 64"><rect x="6" y="12" width="52" height="36" rx="3" fill="none" stroke="currentColor" stroke-width="2"/><line x1="6" y1="20" x2="58" y2="20" stroke="currentColor" stroke-width="2"/><circle cx="11" cy="16" r="1.4" fill="currentColor"/><circle cx="15" cy="16" r="1.4" fill="currentColor"/><circle cx="19" cy="16" r="1.4" fill="currentColor"/></svg>
+            <p>${escapeHtml(message)}</p>
+            ${errorMsg ? `<p class="foreground-empty-error">${escapeHtml(errorMsg)}</p>` : ''}
+        </div>
+    `;
+}
+
+function renderTile(data) {
+    const stage = document.getElementById('foregroundStage');
+    if (!stage) return;
+
+    const procName = data.process_name || '—';
+    const winTitle = data.window_title || '';
+    const execPath = data.executable_path || '';
+    const pid = data.pid ?? '—';
+    const startedEpoch = data.started_at;
+    const startedAgo = startedEpoch ? formatAgo(startedEpoch) : '—';
+    const startedAbs = startedEpoch
+        ? new Date(startedEpoch * 1000).toLocaleString()
+        : '';
+    const geom = formatGeometry(data.window_geometry);
+    const platform = data.platform || '—';
+    const monitorId = data.monitor_id;
+
+    // Chips: only render ones that apply
+    const chips = [];
+    if (data.is_fullscreen) {
+        chips.push(`<span class="fg-chip fg-chip-accent">${escapeHtml(t('foreground.fullscreen'))}</span>`);
+    } else if (!data.is_minimized) {
+        chips.push(`<span class="fg-chip">${escapeHtml(t('foreground.windowed'))}</span>`);
+    }
+    if (data.is_minimized) {
+        chips.push(`<span class="fg-chip fg-chip-mute">${escapeHtml(t('foreground.minimized'))}</span>`);
+    }
+    if (monitorId !== null && monitorId !== undefined) {
+        chips.push(`<span class="fg-chip">${escapeHtml(t('foreground.monitor', { n: monitorId + 1 }))}</span>`);
+    }
+    if (data.is_browser) {
+        chips.push(`<span class="fg-chip fg-chip-mute">${escapeHtml(t('foreground.browser'))}</span>`);
+    }
+
+    // Optional browser-only detail rows (page title + URL when available)
+    const browserRows = [];
+    if (data.is_browser) {
+        if (data.browser_page_title) {
+            browserRows.push(`
+                <div class="fg-row">
+                    <dt>${escapeHtml(t('foreground.page_title'))}</dt>
+                    <dd title="${escapeHtml(data.browser_page_title)}">${escapeHtml(data.browser_page_title)}</dd>
+                </div>
+            `);
+        }
+        if (data.browser_url) {
+            browserRows.push(`
+                <div class="fg-row">
+                    <dt>${escapeHtml(t('foreground.url'))}</dt>
+                    <dd title="${escapeHtml(data.browser_url)}"><span class="fg-mono">${escapeHtml(truncatePath(data.browser_url, 80))}</span></dd>
+                </div>
+            `);
+        }
+    }
+
+    stage.innerHTML = `
+        <article class="foreground-card" data-fullscreen="${data.is_fullscreen ? '1' : '0'}">
+            <div class="fg-kicker">
+                <span data-i18n="foreground.kicker">Foreground</span>
+            </div>
+            <h1 class="fg-process" title="${escapeHtml(procName)}">${escapeHtml(procName)}</h1>
+            <div class="fg-window-title" title="${escapeHtml(winTitle)}">${escapeHtml(winTitle)}</div>
+
+            <div class="fg-chips">${chips.join('')}</div>
+
+            <dl class="fg-details">
+                ${browserRows.join('')}
+                <div class="fg-row">
+                    <dt>${escapeHtml(t('foreground.executable'))}</dt>
+                    <dd title="${escapeHtml(execPath)}"><span class="fg-mono">${escapeHtml(truncatePath(execPath))}</span></dd>
+                </div>
+                <div class="fg-row">
+                    <dt>${escapeHtml(t('foreground.pid'))}</dt>
+                    <dd><span class="fg-mono">${escapeHtml(String(pid))}</span></dd>
+                </div>
+                <div class="fg-row">
+                    <dt>${escapeHtml(t('foreground.started'))}</dt>
+                    <dd title="${escapeHtml(startedAbs)}"><span class="fg-ago" data-started="${startedEpoch ?? ''}">${escapeHtml(startedAgo)}</span></dd>
+                </div>
+                <div class="fg-row">
+                    <dt>${escapeHtml(t('foreground.geometry'))}</dt>
+                    <dd><span class="fg-mono">${escapeHtml(geom)}</span></dd>
+                </div>
+                <div class="fg-row">
+                    <dt>${escapeHtml(t('foreground.platform'))}</dt>
+                    <dd>${escapeHtml(platform)}</dd>
+                </div>
+            </dl>
+        </article>
+    `;
+}
+
+function startAgoTicker() {
+    if (agoTickTimer) return;
+    agoTickTimer = setInterval(() => {
+        const el = document.querySelector('.fg-ago[data-started]');
+        if (!el) return;
+        const epoch = parseFloat(el.getAttribute('data-started'));
+        if (!epoch) return;
+        el.textContent = formatAgo(epoch);
+    }, 15000);
+}
+
+export function updateForegroundUI(data) {
+    latestForeground = data;
+
+    if (!data || data.available === false) {
+        const errMsg = data && data.error ? data.error : '';
+        renderEmpty(t('foreground.unavailable'), errMsg);
+    } else if (!data.process_name && !data.pid) {
+        renderEmpty(t('foreground.no_process'), '');
+    } else {
+        renderTile(data);
+        startAgoTicker();
+    }
+}
+
+export function loadForegroundProcess() {
+    // Push-only — just render the cached state. If nothing has arrived
+    // yet, leave the loading placeholder visible.
+    if (latestForeground !== null) {
+        updateForegroundUI(latestForeground);
+    }
+}

media_server/static/js/links.js

@@ -182,8 +182,7 @@ export async function loadDisplayMonitors() {
                     <span class="display-slider-label" data-i18n="display.contrast">${t('display.contrast')}</span>
                     <input type="range" class="display-slider display-contrast-slider"
                            min="0" max="100" value="${contrastValue}"
-                           oninput="onDisplayContrastInput(${monitor.id}, this.value)"
+                           data-display-slider="contrast" data-monitor-id="${monitor.id}">
-                           onchange="onDisplayContrastChange(${monitor.id}, this.value)">
                     <span class="display-slider-value" id="contrast-val-${monitor.id}">${contrastValue}%</span>
                 </div>`;
             }
@@ -296,8 +295,7 @@ export async function loadDisplayMonitors() {
                     </svg>
                     <span class="display-slider-label" data-i18n="display.brightness">${t('display.brightness')}</span>
                     <input type="range" class="display-slider display-brightness-slider" min="0" max="100" value="${brightnessValue}" ${brightnessDisabled}
-                           oninput="onDisplayBrightnessInput(${monitor.id}, this.value)"
+                           data-display-slider="brightness" data-monitor-id="${monitor.id}">
-                           onchange="onDisplayBrightnessChange(${monitor.id}, this.value)">
                     <span class="display-slider-value display-brightness-value" id="brightness-val-${monitor.id}">${brightnessValue}%</span>
                 </div>
                 ${contrastRow}
@@ -306,10 +304,15 @@ export async function loadDisplayMonitors() {
             container.appendChild(card);
         });
 
-        // Bind a single delegated click handler for the power buttons.
+        // Bind a single delegated click handler for the power buttons,
-        // Avoids inline onclick="..." with interpolated monitor data.
+        // plus input/change handlers for the brightness & contrast sliders.
+        // Avoids inline on* attributes (blocked by script-src 'self' CSP).
         container.removeEventListener('click', _onPowerButtonClick);
         container.addEventListener('click', _onPowerButtonClick);
+        container.removeEventListener('input', _onDisplaySliderInput);
+        container.addEventListener('input', _onDisplaySliderInput);
+        container.removeEventListener('change', _onDisplaySliderChange);
+        container.addEventListener('change', _onDisplaySliderChange);
 
         // Enhance every tuning <select> with an IconSelect now that the
         // cards are in the DOM (IconSelect needs offsetParent + sibling).
@@ -456,6 +459,30 @@ function _onPowerButtonClick(event) {
     if (Number.isFinite(id)) toggleDisplayPower(id);
 }
 
+function _onDisplaySliderInput(event) {
+    const el = event.target.closest('input[data-display-slider]');
+    if (!el) return;
+    const id = Number(el.dataset.monitorId);
+    if (!Number.isFinite(id)) return;
+    if (el.dataset.displaySlider === 'brightness') {
+        onDisplayBrightnessInput(id, el.value);
+    } else if (el.dataset.displaySlider === 'contrast') {
+        onDisplayContrastInput(id, el.value);
+    }
+}
+
+function _onDisplaySliderChange(event) {
+    const el = event.target.closest('input[data-display-slider]');
+    if (!el) return;
+    const id = Number(el.dataset.monitorId);
+    if (!Number.isFinite(id)) return;
+    if (el.dataset.displaySlider === 'brightness') {
+        onDisplayBrightnessChange(id, el.value);
+    } else if (el.dataset.displaySlider === 'contrast') {
+        onDisplayContrastChange(id, el.value);
+    }
+}
+
 export async function toggleDisplayPower(monitorId) {
     const btn = document.getElementById(`power-btn-${monitorId}`);
     const isOn = btn && btn.classList.contains('on');
@@ -509,6 +536,8 @@ export async function loadHeaderLinks() {
             a.href = link.url;
             a.target = '_blank';
             a.rel = 'noopener noreferrer';
+            // Prevent leaking the WebUI URL (with ?token=) via Referer.
+            a.referrerPolicy = 'no-referrer';
             a.className = 'header-link';
             a.title = link.label || link.url;
 

media_server/static/js/player.js

@@ -10,9 +10,11 @@ import {
     lastStatus, setLastStatus, currentPlayState, setCurrentPlayState,
     POSITION_INTERPOLATION_MS, seek, notifyRemoteVolume,
     getAuthHeaders, hasCredentials,
+    togglePlayPause, nextTrack, previousTrack,
 } from './core.js';
 import { updateBackgroundColors } from './background.js';
 import { loadDisplayMonitors } from './links.js';
+import { loadForegroundProcess } from './foreground.js';
 import { IconSelect } from './icon-select.js';
 
 // Tab management
@@ -75,6 +77,7 @@ export function switchTab(tabName) {
 
     if (tabName === 'display') {
         loadDisplayMonitors();
+        loadForegroundProcess();
     }
 
     localStorage.setItem('activeTab', tabName);
@@ -205,20 +208,39 @@ export function renderAccentSwatches() {
     const swatches = accentPresets.map(p =>
         `<div class="accent-swatch ${p.color === current ? 'active' : ''}"
               style="background: ${p.color}"
-              onclick="selectAccentColor('${p.color}', '${p.hover}')"
+              data-accent-color="${p.color}" data-accent-hover="${p.hover}"
               title="${p.name}"></div>`
     ).join('');
 
     const customRow = `
-        <div class="accent-custom-row ${isCustom ? 'active' : ''}" onclick="document.getElementById('accentCustomInput').click()">
+        <div class="accent-custom-row ${isCustom ? 'active' : ''}" data-accent-custom-row>
             <span class="accent-custom-swatch" style="background: ${isCustom ? current : '#888'}"></span>
             <span class="accent-custom-label">${t('accent.custom')}</span>
-            <input type="color" id="accentCustomInput" value="${current}"
+            <input type="color" id="accentCustomInput" value="${current}">
-                   onclick="event.stopPropagation()"
-                   onchange="selectAccentColor(this.value, lightenColor(this.value, 15))">
         </div>`;
 
     dropdown.innerHTML = swatches + customRow;
+
+    // Wire CSP-safe handlers (script-src 'self' blocks inline on* attributes).
+    dropdown.querySelectorAll('.accent-swatch[data-accent-color]').forEach(el => {
+        el.addEventListener('click', () => {
+            selectAccentColor(el.dataset.accentColor, el.dataset.accentHover);
+        });
+    });
+    const customRowEl = dropdown.querySelector('[data-accent-custom-row]');
+    const customInput = dropdown.querySelector('#accentCustomInput');
+    if (customRowEl && customInput) {
+        customRowEl.addEventListener('click', (e) => {
+            // The native color popup only opens from a user-initiated click on
+            // the <input>. Forward clicks on the row to the input — except when
+            // the input itself was the source (avoids re-entry).
+            if (e.target !== customInput) customInput.click();
+        });
+        customInput.addEventListener('click', (e) => e.stopPropagation());
+        customInput.addEventListener('change', () => {
+            selectAccentColor(customInput.value, lightenColor(customInput.value, 15));
+        });
+    }
 }
 
 export function selectAccentColor(color, hover) {
@@ -360,11 +382,85 @@ function buildVisualizerGradient() {
 
 function startVisualizerRender() {
     if (visualizerAnimFrame) return;
+    // Don't even queue a frame while the tab is hidden — rAF still fires on
+    // hidden tabs (throttled but not paused) and would burn CPU + battery
+    // smoothing into bars no one can see. We resume on `visibilitychange`.
+    if (typeof document !== 'undefined' && document.hidden) return;
     // Cache editorial spectrum bar refs once per start.
     cacheEditorialSpectrumBars();
     renderVisualizerFrame();
 }
 
+// ─── OS Media Session integration ─────────────────────────────
+// Hooks the page into the system's media session so headset / lockscreen /
+// Bluetooth play-pause-skip buttons drive the active track. Action handlers
+// are set once and never re-registered; only the metadata + playback state
+// flip when a track changes.
+let _mediaSessionInitialised = false;
+let _lastMediaSessionKey = '';
+function syncMediaSession(status) {
+    if (typeof navigator === 'undefined' || !('mediaSession' in navigator)) return;
+    const session = navigator.mediaSession;
+
+    if (!_mediaSessionInitialised) {
+        const setHandler = (name, fn) => {
+            try { session.setActionHandler(name, fn); } catch { /* unsupported action */ }
+        };
+        setHandler('play', () => togglePlayPause());
+        setHandler('pause', () => togglePlayPause());
+        setHandler('nexttrack', () => nextTrack());
+        setHandler('previoustrack', () => previousTrack());
+        setHandler('seekto', (ev) => { if (ev && typeof ev.seekTime === 'number') seek(ev.seekTime); });
+        _mediaSessionInitialised = true;
+    }
+
+    // Track-identity key — re-build metadata only when title/artist/album change.
+    const artworkSrc = status && status.album_art_url ? '/api/media/artwork' : '';
+    const key = `${status.title || ''}|${status.artist || ''}|${status.album || ''}|${artworkSrc}`;
+    if (key !== _lastMediaSessionKey) {
+        _lastMediaSessionKey = key;
+        try {
+            session.metadata = new MediaMetadata({
+                title: status.title || '',
+                artist: status.artist || '',
+                album: status.album || '',
+                artwork: artworkSrc ? [{ src: artworkSrc, sizes: '512x512', type: 'image/*' }] : [],
+            });
+        } catch { /* MediaMetadata unsupported on very old browsers */ }
+    }
+
+    session.playbackState =
+        status.state === 'playing' ? 'playing'
+        : status.state === 'paused' ? 'paused'
+        : 'none';
+
+    if (typeof session.setPositionState === 'function'
+        && status.duration && status.duration > 0
+        && typeof status.position === 'number') {
+        try {
+            session.setPositionState({
+                duration: status.duration,
+                position: Math.min(status.position, status.duration),
+                playbackRate: 1.0,
+            });
+        } catch { /* invalid range — ignore */ }
+    }
+}
+
+// Pause / resume the visualizer with tab visibility. Idempotent: called once
+// at module init below, no-op if no listener support.
+if (typeof document !== 'undefined' && document.addEventListener) {
+    document.addEventListener('visibilitychange', () => {
+        if (document.hidden) {
+            stopVisualizerRender();
+        } else if (frequencyData) {
+            // Only restart if a payload is live (otherwise startVisualizerRender
+            // would queue a no-op rAF chain forever waiting for one).
+            startVisualizerRender();
+        }
+    });
+}
+
 export function stopVisualizerRender() {
     if (visualizerAnimFrame) {
         cancelAnimationFrame(visualizerAnimFrame);
@@ -882,6 +978,11 @@ export function updateUI(status) {
 
     updateMuteIcon(status.muted);
 
+    // Wire the OS Media Session so headset buttons, lockscreen controls, and
+    // Bluetooth remotes drive the active media (not the WebUI tab). Cheap and
+    // idempotent — re-running setActionHandler with the same fn is a no-op.
+    syncMediaSession(status);
+
     const src = resolveMediaSource(status.source);
     dom.source.textContent = src ? src.name : t('player.unknown_source');
     dom.sourceIcon.innerHTML = src?.icon || '';

media_server/static/js/scripts.js

@@ -80,6 +80,9 @@ export async function displayQuickAccess() {
                     card.href = link.url;
                     card.target = '_blank';
                     card.rel = 'noopener noreferrer';
+                    // Prevent the WebUI's URL (which may carry ?token=...) from
+                    // leaking to third-party sites via Referer.
+                    card.referrerPolicy = 'no-referrer';
 
                     if (link.icon) {
                         const iconEl = document.createElement('div');

media_server/static/js/websocket.js

@@ -12,6 +12,7 @@ import { updateUI, setFrequencyData, stopPositionInterpolation, loadAudioDevices
 import { loadScripts, loadScriptsTable, displayQuickAccess } from './scripts.js';
 import { loadCallbacksTable } from './callbacks.js';
 import { loadHeaderLinks, loadLinksTable } from './links.js';
+import { updateForegroundUI } from './foreground.js';
 
 let reconnectTimeout = null;
 let wsReconnectAttempts = 0;
@@ -87,9 +88,22 @@ export function connectWebSocket(token) {
 
     const protocol = window.location.protocol === 'https:' ? 'wss:' : 'ws:';
     const wsBase = `${protocol}//${window.location.host}/api/media/ws`;
-    const wsUrl = token ? `${wsBase}?token=${encodeURIComponent(token)}` : wsBase;
 
-    const newWs = new WebSocket(wsUrl);
+    // Prefer Sec-WebSocket-Protocol-based auth so the token never appears in
+    // the URL (which would otherwise land in browser history, server access
+    // logs, and Referer headers). Keep the ?token=... fallback for clients
+    // that pre-date this change and don't speak the subprotocol.
+    let newWs;
+    if (token) {
+        try {
+            newWs = new WebSocket(wsBase, [`media-server.token.${token}`]);
+        } catch (e) {
+            console.warn('Subprotocol WS handshake failed, falling back to ?token=', e);
+            newWs = new WebSocket(`${wsBase}?token=${encodeURIComponent(token)}`);
+        }
+    } else {
+        newWs = new WebSocket(wsBase);
+    }
     activeSocket = newWs;
     setWs(newWs);
 
@@ -118,6 +132,8 @@ export function connectWebSocket(token) {
 
         if (msg.type === 'status' || msg.type === 'status_update') {
             updateUI(msg.data);
+        } else if (msg.type === 'foreground' || msg.type === 'foreground_update') {
+            updateForegroundUI(msg.data);
         } else if (msg.type === 'scripts_changed') {
             console.log('Scripts changed, reloading...');
             loadScripts();

media_server/static/locales/en.json

@@ -292,5 +292,29 @@
   "about.source_code": "Source Code",
   "dialog.close": "Close",
   "update.available": "Update available: v{version}",
-  "update.view_release": "View Release"
+  "update.view_release": "View Release",
+  "tab.foreground": "Foreground",
+  "foreground.kicker": "Foreground",
+  "foreground.loading": "Waiting for foreground signal…",
+  "foreground.no_process": "No foreground process",
+  "foreground.unavailable": "Foreground tracking unavailable on this platform",
+  "foreground.process": "Process",
+  "foreground.window_title": "Window title",
+  "foreground.executable": "Executable",
+  "foreground.pid": "PID",
+  "foreground.monitor": "Monitor {n}",
+  "foreground.started": "Started",
+  "foreground.geometry": "Geometry",
+  "foreground.platform": "Platform",
+  "foreground.fullscreen": "Fullscreen",
+  "foreground.minimized": "Minimized",
+  "foreground.windowed": "Windowed",
+  "foreground.browser": "Browser",
+  "foreground.page_title": "Page title",
+  "foreground.url": "URL",
+  "foreground.badge.title": "View foreground process",
+  "foreground.ago.seconds": "{n}s ago",
+  "foreground.ago.minutes": "{n}m ago",
+  "foreground.ago.hours": "{n}h {m}m ago",
+  "foreground.ago.days": "{n}d ago"
 }

media_server/static/locales/ru.json

@@ -292,5 +292,29 @@
   "about.source_code": "Исходный код",
   "dialog.close": "Закрыть",
   "update.available": "Доступно обновление: v{version}",
-  "update.view_release": "Перейти к релизу"
+  "update.view_release": "Перейти к релизу",
+  "tab.foreground": "Активное окно",
+  "foreground.kicker": "Активное окно",
+  "foreground.loading": "Ожидание сигнала об активном окне…",
+  "foreground.no_process": "Активное окно не определено",
+  "foreground.unavailable": "Отслеживание активного окна недоступно",
+  "foreground.process": "Процесс",
+  "foreground.window_title": "Заголовок окна",
+  "foreground.executable": "Путь к программе",
+  "foreground.pid": "PID",
+  "foreground.monitor": "Монитор {n}",
+  "foreground.started": "Запущено",
+  "foreground.geometry": "Геометрия",
+  "foreground.platform": "Платформа",
+  "foreground.fullscreen": "Полноэкранный",
+  "foreground.minimized": "Свёрнут",
+  "foreground.windowed": "Оконный",
+  "foreground.browser": "Браузер",
+  "foreground.page_title": "Заголовок страницы",
+  "foreground.url": "URL",
+  "foreground.badge.title": "Открыть активное окно",
+  "foreground.ago.seconds": "{n} с назад",
+  "foreground.ago.minutes": "{n} мин назад",
+  "foreground.ago.hours": "{n} ч {m} мин назад",
+  "foreground.ago.days": "{n} дн назад"
 }

media_server/static/manifest.json

@@ -1,12 +1,14 @@
 {
+  "id": "/",
   "name": "Media Server",
   "short_name": "Media",
   "description": "Remote media player control and file browser",
   "start_url": "/",
+  "scope": "/",
   "display": "standalone",
   "orientation": "any",
-  "background_color": "#121212",
+  "background_color": "#0E0D0B",
-  "theme_color": "#121212",
+  "theme_color": "#0E0D0B",
   "icons": [
     {
       "src": "/static/icons/icon.svg",

media_server/tray.py

@@ -3,6 +3,7 @@
 import ctypes
 import io
 import logging
+import sys
 import webbrowser
 from pathlib import Path
 from typing import Callable
@@ -30,62 +31,136 @@ _IDYES = 6
 
 
 def _confirm(title: str, message: str) -> bool:
-    """Show a Yes/No dialog using native Windows MessageBox."""
+    """Show a Yes/No dialog before a destructive tray action.
-    result = ctypes.windll.user32.MessageBoxW(
+
-        0, message, title, _MB_YESNO | _MB_ICONQUESTION | _MB_TOPMOST | _MB_SETFOREGROUND
+    Uses the native Windows MessageBox on win32; on Linux/macOS we don't
-    )
+    have a no-dependency GUI dialog available (Tk pulls in tkinter, gtk
-    return result == _IDYES
+    pulls in PyGObject), so we log + auto-confirm — the tray menu items
+    themselves require a deliberate click already.
+    """
+    if sys.platform == "win32":
+        result = ctypes.windll.user32.MessageBoxW(
+            0,
+            message,
+            title,
+            _MB_YESNO | _MB_ICONQUESTION | _MB_TOPMOST | _MB_SETFOREGROUND,
+        )
+        return result == _IDYES
+
+    logger.info("Tray confirm (auto-yes on non-Windows): %s — %s", title, message)
+    return True
 
 
-def _create_icon_image(size: int = 64) -> Image.Image:
+# Frame size we ask the multi-resolution ICO for. Most Windows tray surfaces
-    """Create a tray icon: green circle with white play triangle."""
+# show 16x16 in the notification area and 32x32 in jump lists / Alt+Tab; 64
+# gives pystray enough headroom for both without forcing it to upscale.
+_TRAY_ICON_SIZE = 64
+
+# Palette mirrors media_server/static/icons/icon.svg ("Beacon" design).
+_BG_DARK = (11, 61, 59, 255)        # #0B3D3B
+_BG_LIGHT = (26, 107, 94, 255)      # #1A6B5E
+_FG_PARCHMENT = (245, 241, 232, 255)  # #F5F1E8
+
+
+def _create_icon_image(size: int = _TRAY_ICON_SIZE) -> Image.Image:
+    """Procedural fallback when no icon file is available.
+
+    Matches the "Beacon" palette (deep teal squircle + warm parchment play
+    triangle) so a missing icon.ico does not regress us back to the old
+    Spotify-green circle.
+    """
     img = Image.new("RGBA", (size, size), (0, 0, 0, 0))
     draw = ImageDraw.Draw(img)
 
-    # Green circle background
+    # Squircle background. Vertical gradient approximates the diagonal one
-    padding = 2
+    # in the real SVG well enough for a 64px fallback.
-    draw.ellipse(
+    radius = int(size * 0.225)
-        [padding, padding, size - padding, size - padding],
+    for y in range(size):
-        fill=(29, 185, 84, 255),
+        t = y / max(1, size - 1)
+        color = tuple(
+            round(_BG_DARK[i] + (_BG_LIGHT[i] - _BG_DARK[i]) * t) for i in range(3)
+        ) + (255,)
+        draw.line([(0, y), (size - 1, y)], fill=color)
+    mask = Image.new("L", (size, size), 0)
+    ImageDraw.Draw(mask).rounded_rectangle((0, 0, size - 1, size - 1), radius=radius, fill=255)
+    bg = img.copy()
+    img = Image.new("RGBA", (size, size), (0, 0, 0, 0))
+    img.paste(bg, (0, 0), mask=mask)
+
+    # Play triangle, positioned to match icon.svg's geometry.
+    draw = ImageDraw.Draw(img)
+    draw.polygon(
+        [
+            (size * 0.345, size * 0.215),
+            (size * 0.345, size * 0.785),
+            (size * 0.755, size * 0.500),
+        ],
+        fill=_FG_PARCHMENT,
     )
-
-    # White play triangle
-    cx, cy = size // 2, size // 2
-    r = size * 0.28
-    triangle = [
-        (cx - r * 0.6, cy - r),
-        (cx - r * 0.6, cy + r),
-        (cx + r * 0.9, cy),
-    ]
-    draw.polygon(triangle, fill=(255, 255, 255, 255))
-
     return img
 
 
-def _load_icon_image() -> Image.Image:
+def _select_frame(image: Image.Image, target: int) -> Image.Image:
-    """Load the ICO/SVG app icon, falling back to a generated image."""
+    """Pick the best frame from a multi-resolution ICO.
+
+    Pillow's ICO loader exposes the embedded sizes via ``image.info['sizes']``.
+    We pick the smallest frame at least as large as the target (so we never
+    upscale) and resize down to ``target x target`` with LANCZOS.
+    """
+    sizes = sorted(image.info.get("sizes", []) or [], key=lambda wh: wh[0])
+    chosen = next((wh for wh in sizes if wh[0] >= target), sizes[-1] if sizes else None)
+    if chosen is not None:
+        image.size = chosen
+    frame = image.copy().convert("RGBA")
+    if frame.size != (target, target):
+        frame = frame.resize((target, target), Image.LANCZOS)
+    return frame
+
+
+def _load_icon_image(size: int = _TRAY_ICON_SIZE) -> Image.Image:
+    """Load the app icon for the tray.
+
+    Order:
+      1. ``icon.ico`` — the multi-resolution Windows icon ships with every
+         build; pick the frame closest to ``size`` and downscale if needed.
+      2. ``icon.svg`` via resvg-py (preferred) or cairosvg (legacy).
+      3. Procedural ``_create_icon_image`` fallback.
+    """
     icons_dir = Path(__file__).parent / "static" / "icons"
 
-    # Try .ico first (best for Windows tray)
     ico_path = icons_dir / "icon.ico"
     if ico_path.exists():
         try:
-            return Image.open(ico_path)
+            with Image.open(ico_path) as ico:
+                return _select_frame(ico, size)
+        except Exception as exc:
+            logger.warning("Failed to load tray icon from %s: %s", ico_path, exc)
+
+    svg_path = icons_dir / "icon.svg"
+    if svg_path.exists():
+        try:
+            import resvg_py
+
+            png_data = resvg_py.svg_to_bytes(
+                svg_string=svg_path.read_text(encoding="utf-8"),
+                width=size,
+                height=size,
+            )
+            return Image.open(io.BytesIO(bytes(png_data))).convert("RGBA")
+        except ImportError:
+            pass
+        except Exception as exc:
+            logger.warning("resvg rasterization of %s failed: %s", svg_path, exc)
+
+        try:
+            import cairosvg
+
+            png_data = cairosvg.svg2png(url=str(svg_path), output_width=size, output_height=size)
+            return Image.open(io.BytesIO(png_data)).convert("RGBA")
         except Exception:
             pass
 
-    # Try SVG via cairosvg
+    return _create_icon_image(size)
-    try:
-        import cairosvg
-
-        svg_path = icons_dir / "icon.svg"
-        if svg_path.exists():
-            png_data = cairosvg.svg2png(url=str(svg_path), output_width=64, output_height=64)
-            return Image.open(io.BytesIO(png_data))
-    except Exception:
-        pass
-
-    return _create_icon_image()
 
 
 class TrayManager:
@@ -101,6 +176,9 @@ class TrayManager:
 
         self._port = port
         self._on_exit = on_exit
+        # Initialize so the property and any cross-thread reader cannot ever
+        # observe an uninitialized attribute. Set before _on_exit() fires.
+        self._restart_requested = False
 
         menu = pystray.Menu(
             pystray.MenuItem("Show UI", self._show_ui, default=True),
@@ -123,13 +201,16 @@ class TrayManager:
         if not _confirm("Media Server", "Restart the server?"):
             return
         logger.info("Restart requested from tray")
+        # Set the flag BEFORE signalling exit so the main thread observes it
+        # when it wakes from server_thread.join() — order matters across the
+        # tray/uvicorn handoff.
         self._restart_requested = True
         self._on_exit()
         self._icon.stop()
 
     @property
     def restart_requested(self) -> bool:
-        return getattr(self, "_restart_requested", False)
+        return self._restart_requested
 
     def _shutdown(self, icon: "pystray.Icon", item: "pystray.MenuItem") -> None:
         if not _confirm("Media Server", "Shut down the server?"):

package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "media-server-frontend",
-  "version": "0.2.5",
+  "version": "0.4.0",
   "lockfileVersion": 2,
   "requires": true,
   "packages": {
     "": {
       "name": "media-server-frontend",
-      "version": "0.2.5",
+      "version": "0.4.0",
       "devDependencies": {
         "esbuild": "^0.27.4"
       }

package.json

@@ -1,6 +1,6 @@
 {
   "name": "media-server-frontend",
-  "version": "0.2.5",
+  "version": "0.4.0",
   "private": true,
   "description": "Frontend build tooling for media server WebUI",
   "scripts": {

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "media-server"
-version = "0.2.5"
+version = "0.4.0"
 description = "REST API server for controlling system-wide media playback"
 readme = "README.md"
 license = { text = "MIT" }
@@ -34,24 +34,44 @@ dependencies = [
     "pillow>=10.0.0",
     "soundcard>=0.4.0",
     "numpy>=1.24.0,<2.0",
+    # screen-brightness-control + monitorcontrol are cross-platform
+    # (Windows / Linux i2c-dev). Kept in base deps so a plain
+    # `pip install media-server` yields a working /api/display on any
+    # platform that has the underlying hardware support.
+    "screen-brightness-control>=0.20.0",
+    "monitorcontrol>=3.0.0",
 ]
 
 [project.optional-dependencies]
 windows = [
     "winsdk>=1.0.0b10",
-    "pywin32>=306",
+    "pywin32>=306; sys_platform == 'win32'",
-    "comtypes>=1.2.0",
+    "comtypes>=1.2.0; sys_platform == 'win32'",
-    "pycaw>=20230407",
+    "pycaw>=20230407; sys_platform == 'win32'",
-    "screen-brightness-control>=0.20.0",
+    "wmi>=1.5.1; sys_platform == 'win32'",
-    "wmi>=1.5.1",
-    "monitorcontrol>=3.0.0",
     "pystray>=0.19.0",
 ]
+linux = [
+    # MPRIS / D-Bus media control. Requires libdbus-1-dev, libglib2.0-dev,
+    # and pkg-config installed on the host before pip install.
+    "dbus-python>=1.3.2; sys_platform == 'linux'",
+    "PyGObject>=3.46; sys_platform == 'linux'",
+    # X11 foreground-window probe (Wayland sessions degrade gracefully).
+    "python-xlib>=0.33; sys_platform == 'linux'",
+]
+macos = [
+    # PyObjC for the foreground-window + active-app probes.
+    "pyobjc-framework-Cocoa>=10.0; sys_platform == 'darwin'",
+    "pyobjc-framework-Quartz>=10.0; sys_platform == 'darwin'",
+]
 dev = [
     "pytest>=7.0",
     "pytest-asyncio>=0.21",
     "httpx>=0.24",
     "ruff>=0.4.0",
+    # SVG -> PNG rasterizer used by scripts/generate-icon.py to (re)build
+    # media_server/static/icons/icon.ico from icon.svg. Build-time only.
+    "resvg-py>=0.3.2",
 ]
 
 [project.urls]

scripts/generate-icon.py

@@ -0,0 +1,113 @@
+"""Generate the Media Server application icon.
+
+The SVG in ``media_server/static/icons/icon.svg`` is the single source of
+truth. This script rasterizes it at every Windows ICO size via ``resvg-py``
+(Rust-backed, identical math to Firefox's SVG renderer) and packs them all
+into a multi-resolution ``icon.ico``.
+
+This replaces the original 16x16-only ICO that Windows was upscaling into
+mush for the installer chrome, Start Menu, desktop shortcuts, and Alt+Tab.
+
+Usage:
+    python scripts/generate-icon.py
+
+Dependencies:
+    pip install resvg-py Pillow
+"""
+
+from __future__ import annotations
+
+import io
+from pathlib import Path
+
+import resvg_py
+from PIL import Image
+
+# Sizes packed into the ICO. Windows picks the closest match per surface;
+# more sizes = sharper rendering everywhere (taskbar, installer header,
+# Alt+Tab, jump lists, Start tile, desktop, file explorer details/tiles).
+ICO_SIZES: tuple[int, ...] = (16, 20, 24, 32, 40, 48, 64, 96, 128, 256)
+
+# The SVG source. Squircle + diagonal teal gradient + warm parchment play
+# triangle with a drop shadow + ghosted echo-chevrons hinting at broadcast.
+SVG_SOURCE = """<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 256 256" role="img" aria-label="Media Server">
+  <defs>
+    <linearGradient id="bg" x1="0%" y1="0%" x2="100%" y2="100%">
+      <stop offset="0%"  stop-color="#0B3D3B"/>
+      <stop offset="100%" stop-color="#1A6B5E"/>
+    </linearGradient>
+    <linearGradient id="sheen" x1="50%" y1="0%" x2="50%" y2="100%">
+      <stop offset="0%"  stop-color="#FFFFFF" stop-opacity="0.18"/>
+      <stop offset="55%" stop-color="#FFFFFF" stop-opacity="0"/>
+    </linearGradient>
+    <filter id="triShadow" x="-20%" y="-20%" width="140%" height="140%">
+      <feGaussianBlur stdDeviation="3"/>
+      <feOffset dx="3" dy="5"/>
+      <feComponentTransfer><feFuncA type="linear" slope="0.45"/></feComponentTransfer>
+      <feMerge><feMergeNode/><feMergeNode in="SourceGraphic"/></feMerge>
+    </filter>
+    <clipPath id="clip"><rect x="0" y="0" width="256" height="256" rx="58" ry="58"/></clipPath>
+  </defs>
+  <g clip-path="url(#clip)">
+    <rect width="256" height="256" fill="url(#bg)"/>
+    <rect width="256" height="256" fill="url(#sheen)"/>
+    <g stroke="#F5F1E8" stroke-width="4.6" stroke-linecap="round" stroke-linejoin="round" fill="none" opacity="0.18">
+      <polyline points="66,72 105,128 66,184"/>
+      <polyline points="85,82 124,128 85,174" opacity="0.6"/>
+    </g>
+    <path d="M88.3 55 L88.3 201 L193.3 128 Z"
+          fill="#F5F1E8"
+          stroke="#F5F1E8" stroke-width="9" stroke-linejoin="round"
+          filter="url(#triShadow)"/>
+  </g>
+  <rect x="0.5" y="0.5" width="255" height="255" rx="58" ry="58"
+        fill="none" stroke="#000000" stroke-opacity="0.18" stroke-width="1"/>
+</svg>
+"""
+
+
+def _render_png(size: int) -> Image.Image:
+    """Rasterize the SVG to a PNG at ``size x size`` via resvg."""
+    data = resvg_py.svg_to_bytes(
+        svg_string=SVG_SOURCE,
+        width=size,
+        height=size,
+        shape_rendering="geometric_precision",
+    )
+    return Image.open(io.BytesIO(bytes(data))).convert("RGBA")
+
+
+def main() -> None:
+    root = Path(__file__).resolve().parent.parent
+    out_dir = root / "media_server" / "static" / "icons"
+    out_dir.mkdir(parents=True, exist_ok=True)
+
+    # SVG — canonical source, also used as the Web UI favicon.
+    svg_path = out_dir / "icon.svg"
+    svg_path.write_text(SVG_SOURCE, encoding="utf-8")
+    print(f"wrote {svg_path}")
+
+    # Rasterize every ICO size via resvg.
+    frames = [_render_png(size) for size in ICO_SIZES]
+
+    # Pack into a multi-resolution ICO. The "primary" image must be the
+    # largest; the rest go via append_images. Pillow's ICO writer then
+    # serializes one frame per size.
+    primary = frames[-1]
+    ico_path = out_dir / "icon.ico"
+    primary.save(
+        ico_path,
+        format="ICO",
+        sizes=[(s, s) for s in ICO_SIZES],
+        append_images=frames[:-1],
+    )
+    print(f"wrote {ico_path}  ({ico_path.stat().st_size:,} bytes, sizes={list(ICO_SIZES)})")
+
+    # Largest PNG for documentation / non-Windows surfaces.
+    png_path = out_dir / "icon-256.png"
+    primary.save(png_path, format="PNG", optimize=True)
+    print(f"wrote {png_path}")
+
+
+if __name__ == "__main__":
+    main()

tests/test_auth_scopes.py

@@ -0,0 +1,58 @@
+"""Tests for token scope hierarchy + back-compat with legacy bare-string tokens."""
+
+from __future__ import annotations
+
+import pytest
+
+from media_server.config import Settings, TokenSpec
+
+
+def test_bare_string_token_promotes_to_admin_scope():
+    """Legacy `label: <token>` form must still work and grant admin."""
+    s = Settings(api_tokens={"legacy": "deadbeef-deadbeef-deadbeef-deadbeef"})
+    spec = s.api_tokens["legacy"]
+    assert isinstance(spec, TokenSpec)
+    assert spec.token == "deadbeef-deadbeef-deadbeef-deadbeef"
+    assert spec.scopes == ["admin"]
+    assert spec.grants("admin")
+    assert spec.grants("control")
+    assert spec.grants("read")
+
+
+def test_dict_token_with_explicit_scopes():
+    s = Settings(api_tokens={
+        "ha": {"token": "aaaaaaaaaaaaaaaa", "scopes": ["read", "control"]},
+    })
+    spec = s.api_tokens["ha"]
+    assert spec.grants("control")
+    assert spec.grants("read")
+    assert not spec.grants("admin")
+
+
+def test_read_only_scope_grants_only_read():
+    spec = TokenSpec(token="xxxxxxxxxxxxxxxx", scopes=["read"])
+    assert spec.grants("read")
+    assert not spec.grants("control")
+    assert not spec.grants("admin")
+
+
+def test_admin_scope_implies_control_and_read():
+    spec = TokenSpec(token="xxxxxxxxxxxxxxxx", scopes=["admin"])
+    assert spec.grants("read")
+    assert spec.grants("control")
+    assert spec.grants("admin")
+
+
+def test_unknown_scope_rejected():
+    with pytest.raises(ValueError, match="unknown scopes"):
+        TokenSpec(token="xxxxxxxxxxxxxxxx", scopes=["root"])
+
+
+def test_empty_scopes_rejected():
+    with pytest.raises(ValueError, match="at least one"):
+        TokenSpec(token="xxxxxxxxxxxxxxxx", scopes=[])
+
+
+def test_short_token_rejected():
+    with pytest.raises(ValueError):
+        TokenSpec(token="short", scopes=["read"])

tests/test_browser_validate_path.py

@@ -0,0 +1,84 @@
+"""Path traversal defence for BrowserService.validate_path.
+
+The browser endpoint is the single most security-critical filesystem entry
+point in the app: it serves file contents and folder listings to the WebUI.
+A bypass here = arbitrary read of any file the server process can see.
+
+The current implementation signals rejection by *raising* (ValueError for
+traversal/NUL/unknown folder, FileNotFoundError for non-existent absolute
+paths). Either rejection mode is acceptable — these tests assert that the
+adversarial input never returns a path *inside* the configured base.
+"""
+
+from __future__ import annotations
+
+import tempfile
+from pathlib import Path
+from unittest.mock import patch
+
+import pytest
+
+from media_server.services.browser_service import BrowserService
+
+
+@pytest.fixture
+def tmp_media_folder():
+    """A real temp dir registered as a media folder for the test duration."""
+    with tempfile.TemporaryDirectory() as tmp:
+        base = Path(tmp).resolve()
+        (base / "ok.mp3").write_bytes(b"id3")
+        (base / "sub").mkdir()
+        (base / "sub" / "nested.mp3").write_bytes(b"id3")
+
+        from media_server.config import MediaFolderConfig
+        folders = {"test": MediaFolderConfig(path=str(base), label="Test", enabled=True)}
+        with patch("media_server.services.browser_service.settings.media_folders", folders):
+            yield base
+
+
+def _is_rejected(folder_id: str, rel: str) -> bool:
+    """Helper: True iff validate_path either raises or returns None."""
+    try:
+        result = BrowserService.validate_path(folder_id, rel)
+    except (ValueError, FileNotFoundError, OSError):
+        return True
+    return result is None
+
+
+def test_validate_path_accepts_a_real_file(tmp_media_folder: Path):
+    p = BrowserService.validate_path("test", "ok.mp3")
+    assert p is not None
+    assert p.is_file()
+    # Defence-in-depth: resolved path must live inside the base.
+    assert tmp_media_folder in p.resolve().parents or p.resolve().parent == tmp_media_folder
+
+
+def test_validate_path_accepts_nested(tmp_media_folder: Path):
+    p = BrowserService.validate_path("test", "sub/nested.mp3")
+    assert p is not None
+
+
+def test_unknown_folder_rejected(tmp_media_folder: Path):
+    assert _is_rejected("ghost", "ok.mp3")
+
+
+def test_dotdot_traversal_rejected(tmp_media_folder: Path):
+    assert _is_rejected("test", "../etc/passwd")
+    assert _is_rejected("test", "..\\..\\Windows\\System32")
+    assert _is_rejected("test", "sub/../../etc/passwd")
+
+
+def test_absolute_path_rejected(tmp_media_folder: Path):
+    assert _is_rejected("test", "/etc/passwd")
+    assert _is_rejected("test", "C:\\Windows\\System32")
+    assert _is_rejected("test", "C:/Windows")
+
+
+def test_unc_path_rejected(tmp_media_folder: Path):
+    assert _is_rejected("test", "\\\\server\\share")
+    assert _is_rejected("test", "//server/share")
+
+
+def test_null_byte_rejected(tmp_media_folder: Path):
+    assert _is_rejected("test", "ok.mp3\x00.png")
+    assert _is_rejected("test", "sub\x00/nested.mp3")

tests/test_config_atomic_write.py

@@ -0,0 +1,46 @@
+"""Atomic config writes + POSIX permission hardening."""
+
+from __future__ import annotations
+
+import os
+import stat
+import sys
+import tempfile
+from pathlib import Path
+
+import pytest
+
+from media_server.config import _restrict_config_perms, _write_yaml_atomic
+
+
+def test_atomic_write_round_trip():
+    with tempfile.TemporaryDirectory() as tmp:
+        path = Path(tmp) / "config.yaml"
+        _write_yaml_atomic(path, {"port": 8765, "host": "127.0.0.1"})
+        assert path.exists()
+        # Tmp file from the rename should be gone.
+        assert not path.with_suffix(path.suffix + ".tmp").exists()
+        # Contents are valid YAML and round-trip.
+        import yaml
+        data = yaml.safe_load(path.read_text())
+        assert data == {"port": 8765, "host": "127.0.0.1"}
+
+
+def test_atomic_write_replaces_existing():
+    with tempfile.TemporaryDirectory() as tmp:
+        path = Path(tmp) / "config.yaml"
+        path.write_text("old: 1\n")
+        _write_yaml_atomic(path, {"new": 2})
+        import yaml
+        assert yaml.safe_load(path.read_text()) == {"new": 2}
+
+
+@pytest.mark.skipif(sys.platform == "win32", reason="POSIX-only permission check")
+def test_restrict_config_perms_posix():
+    with tempfile.TemporaryDirectory() as tmp:
+        path = Path(tmp) / "config.yaml"
+        path.write_text("token: secret\n")
+        _restrict_config_perms(path)
+        mode = stat.S_IMODE(os.stat(path).st_mode)
+        # Owner read+write only.
+        assert mode == 0o600, f"got {oct(mode)}"

tests/test_foreground_service.py

@@ -0,0 +1,83 @@
+"""Smoke tests for the foreground tracker.
+
+The OS-specific probe code is hard to mock end-to-end inside a CI container,
+so these tests focus on the platform-agnostic surface: the dataclass shape,
+TTL caching, and graceful fallback when the platform probe raises. The
+Windows/Linux/macOS probes themselves are exercised through manual runs.
+"""
+
+from __future__ import annotations
+
+from media_server.services import foreground_service as fg
+
+
+def setup_function(_):
+    fg.reset_cache()
+
+
+def test_unavailable_default_shape():
+    info = fg.ForegroundInfo(available=False)
+    d = info.to_dict()
+    assert d["available"] is False
+    assert d["pid"] is None
+    assert d["process_name"] is None
+    assert d["is_fullscreen"] is False
+    assert "platform" in d
+
+
+def test_cache_returns_same_instance(monkeypatch):
+    calls = {"n": 0}
+
+    def fake_probe():
+        calls["n"] += 1
+        return fg.ForegroundInfo(available=True, pid=42, process_name="x.exe")
+
+    monkeypatch.setattr(fg, "_probe", fake_probe)
+
+    a = fg.get_foreground_info()
+    b = fg.get_foreground_info()
+    assert a is b
+    assert calls["n"] == 1
+
+
+def test_cache_force_refresh(monkeypatch):
+    calls = {"n": 0}
+
+    def fake_probe():
+        calls["n"] += 1
+        return fg.ForegroundInfo(available=True, pid=calls["n"])
+
+    monkeypatch.setattr(fg, "_probe", fake_probe)
+
+    fg.get_foreground_info()
+    fg.get_foreground_info(force_refresh=True)
+    assert calls["n"] == 2
+
+
+def test_cache_ttl_expiry(monkeypatch):
+    calls = {"n": 0}
+
+    def fake_probe():
+        calls["n"] += 1
+        return fg.ForegroundInfo(available=True, pid=calls["n"])
+
+    monkeypatch.setattr(fg, "_probe", fake_probe)
+    monkeypatch.setattr(fg, "_CACHE_TTL", 0.0)
+    # Re-bind the cache's TTL by exercising it twice with TTL 0.
+    fg.get_foreground_info()
+    fg.get_foreground_info()
+    assert calls["n"] == 2
+
+
+def test_probe_crash_returns_unavailable(monkeypatch):
+    def boom():
+        raise RuntimeError("kaboom")
+
+    # Force every platform branch to call our crashing probe.
+    monkeypatch.setattr(fg, "_probe_windows", boom)
+    monkeypatch.setattr(fg, "_probe_linux", boom)
+    monkeypatch.setattr(fg, "_probe_macos", boom)
+
+    info = fg._probe()
+    assert info.available is False
+    assert info.error and "kaboom" in info.error

tests/test_gitea_tag_validation.py

@@ -0,0 +1,45 @@
+"""Tag-name validation in the Gitea release provider.
+
+Whitelist regex protects the URL we broadcast to clients from any path
+traversal or character-set abuse in a hostile (or MITM'd) upstream response.
+"""
+
+from __future__ import annotations
+
+from media_server.services.gitea_release_provider import _TAG_RE
+
+
+def test_accepts_plain_semver():
+    assert _TAG_RE.match("1.0.0")
+    assert _TAG_RE.match("v1.0.0")
+    assert _TAG_RE.match("0.3.7")
+
+
+def test_accepts_pre_release_suffix():
+    assert _TAG_RE.match("v1.0.0-alpha.1")
+    assert _TAG_RE.match("v2.3.4-rc.10")
+    assert _TAG_RE.match("v0.2.7+build.42")
+
+
+def test_rejects_path_traversal():
+    assert not _TAG_RE.match("../etc/passwd")
+    assert not _TAG_RE.match("v1.0.0/../../evil")
+    assert not _TAG_RE.match("v1.0.0/secret")
+
+
+def test_rejects_url_injection():
+    assert not _TAG_RE.match("v1.0.0?evil=1")
+    assert not _TAG_RE.match("v1.0.0#frag")
+    assert not _TAG_RE.match("v1.0.0 OR 1=1")
+    assert not _TAG_RE.match("https://evil.example/")
+
+
+def test_rejects_empty_and_garbage():
+    assert not _TAG_RE.match("")
+    assert not _TAG_RE.match("not-a-version")
+    assert not _TAG_RE.match("v")
+
+
+def test_rejects_excessively_long_suffix():
+    long_suffix = "x" * 40
+    assert not _TAG_RE.match(f"v1.0.0-{long_suffix}")

tests/test_rate_limit.py

@@ -0,0 +1,68 @@
+"""Token-bucket rate limiter behaviour."""
+
+from __future__ import annotations
+
+import time
+
+import pytest
+
+from media_server.services import rate_limit
+
+
+@pytest.fixture(autouse=True)
+def _reset_state():
+    rate_limit._state.clear()
+    rate_limit._LAST_CLEANUP = 0.0
+    yield
+    rate_limit._state.clear()
+
+
+def test_allows_up_to_capacity_then_blocks(monkeypatch):
+    """Default execute bucket = 10/min."""
+    peer = "10.0.0.1"
+    for i in range(10):
+        ok, retry = rate_limit.check("execute", peer)
+        assert ok, f"expected allow on attempt {i + 1}, got block (retry={retry})"
+    ok, retry = rate_limit.check("execute", peer)
+    assert ok is False
+    assert retry is not None and retry > 0
+
+
+def test_different_peers_independent():
+    for _ in range(10):
+        assert rate_limit.check("execute", "10.0.0.1")[0]
+    # Different peer should still be allowed.
+    assert rate_limit.check("execute", "10.0.0.2")[0]
+
+
+def test_unknown_bucket_uses_default():
+    peer = "10.0.0.3"
+    # default = 60/min — first call always allowed.
+    allowed, _ = rate_limit.check("nonexistent-bucket", peer)
+    assert allowed
+
+
+def test_auth_bucket_is_strict():
+    """auth bucket = 5/min."""
+    peer = "10.0.0.4"
+    for _ in range(5):
+        assert rate_limit.check("auth", peer)[0]
+    blocked, retry = rate_limit.check("auth", peer)
+    assert not blocked
+    assert retry is not None
+
+
+def test_refill_eventually_unblocks(monkeypatch):
+    """Verify the bucket refills — exhaust then wait one refill period."""
+    peer = "10.0.0.5"
+    # Replace BUCKETS with a fast-refilling one for the test only.
+    monkeypatch.setitem(
+        rate_limit.BUCKETS,
+        "fast-test",
+        rate_limit.BucketConfig(capacity=2, refill_per_sec=10.0),
+    )
+    assert rate_limit.check("fast-test", peer)[0]
+    assert rate_limit.check("fast-test", peer)[0]
+    assert not rate_limit.check("fast-test", peer)[0]
+    time.sleep(0.15)  # 0.15 * 10 = 1.5 tokens
+    assert rate_limit.check("fast-test", peer)[0]

tests/test_script_params.py

@@ -0,0 +1,77 @@
+"""Validation rules for script parameters (type coercion, regex pattern)."""
+
+from __future__ import annotations
+
+import pytest
+from fastapi import HTTPException
+
+from media_server.config import ScriptParameterConfig
+from media_server.routes.scripts import _validate_params
+
+
+def _defs(**kwargs) -> dict[str, ScriptParameterConfig]:
+    return {name: ScriptParameterConfig(**spec) for name, spec in kwargs.items()}
+
+
+def test_unknown_param_rejected():
+    with pytest.raises(HTTPException) as ei:
+        _validate_params({"x": "1"}, _defs())
+    assert ei.value.status_code == 400
+    assert "Unknown" in ei.value.detail
+
+
+def test_missing_required_rejected():
+    defs = _defs(name={"type": "string", "required": True})
+    with pytest.raises(HTTPException, match="missing"):
+        _validate_params({}, defs)
+
+
+def test_integer_coercion_and_bounds():
+    defs = _defs(volume={"type": "integer", "min": 0, "max": 100})
+    out = _validate_params({"volume": "42"}, defs)
+    assert out == {"SCRIPT_PARAM_VOLUME": "42"}
+
+    with pytest.raises(HTTPException, match="<="):
+        _validate_params({"volume": 200}, defs)
+    with pytest.raises(HTTPException, match=">="):
+        _validate_params({"volume": -1}, defs)
+    with pytest.raises(HTTPException, match="integer"):
+        _validate_params({"volume": "not-a-number"}, defs)
+
+
+def test_boolean_coercion():
+    defs = _defs(flag={"type": "boolean"})
+    assert _validate_params({"flag": "true"}, defs) == {"SCRIPT_PARAM_FLAG": "True"}
+    assert _validate_params({"flag": "no"}, defs) == {"SCRIPT_PARAM_FLAG": "False"}
+    with pytest.raises(HTTPException, match="boolean"):
+        _validate_params({"flag": "maybe"}, defs)
+
+
+def test_select_rejects_non_option():
+    defs = _defs(mode={"type": "select", "options": ["a", "b", "c"]})
+    assert _validate_params({"mode": "a"}, defs) == {"SCRIPT_PARAM_MODE": "a"}
+    with pytest.raises(HTTPException, match="must be one of"):
+        _validate_params({"mode": "z"}, defs)
+
+
+def test_pattern_enforced_on_string():
+    """Regex pattern is the defence against shell metachars in shell=true scripts."""
+    defs = _defs(host={"type": "string", "pattern": r"^[a-z0-9.\-]+$"})
+    assert _validate_params({"host": "example.com"}, defs) == {"SCRIPT_PARAM_HOST": "example.com"}
+    with pytest.raises(HTTPException, match="pattern"):
+        _validate_params({"host": "evil & calc.exe"}, defs)
+    with pytest.raises(HTTPException, match="pattern"):
+        _validate_params({"host": "$(rm -rf /)"}, defs)
+
+
+def test_pattern_can_disallow_empty():
+    defs = _defs(host={"type": "string", "pattern": r"^[a-z]+$"})
+    with pytest.raises(HTTPException, match="pattern"):
+        _validate_params({"host": ""}, defs)
+
+
+def test_invalid_pattern_in_config_fails_closed():
+    defs = _defs(host={"type": "string", "pattern": r"["})  # unmatched bracket
+    with pytest.raises(HTTPException) as ei:
+        _validate_params({"host": "x"}, defs)
+    assert ei.value.status_code == 500