Two new persisted preferences in DISPLAY settings, paired to give operators the 'launch TeamsISO, never see Teams' experience the user asked for:
- LaunchTeamsOnStartup: TeamsISO auto-starts Teams in the background each launch (fire-and-forget background task in App.OnStartup, after the main window has materialized so a slow Teams launch doesn't delay the UI).
- AutoHideTeamsWindows: as soon as Teams' windows materialize after launch, hide them. New TeamsLauncher.AutoHideAfterLaunchAsync runs a polling loop (250ms / up to 15s) that catches the splash, main window, and any follow-up panels Teams opens. Teams takes 2-5s to render its main window and the splash arrives separately, so a one-shot hide right after launch wouldn't be enough.
When TeamsISO starts and Teams is already running (from a prior session), the auto-hide path still fires so the 'I only see TeamsISO' rule applies even when Teams was launched externally.
Operator drives everything through the IN-CALL bar (mute / camera / share / leave / marker) + participants DataGrid (ISO routing). Eye-toggle in the rail still restores Teams windows on demand.
Both toggles default to off — opt-in. Persisted via UIPreferences so they survive process restart.
Two user-reported bugs:
1) CheckBox content was clipping in the 380px settings panel ('Control surface (Stream Deck / Companion / w...' / 'LAN-reachable (allow other machines on yo...'). The Wd.CheckBox template used a horizontal StackPanel which doesn't bound child width, so long Content strings ran off the column without wrapping. Replaced StackPanel with a Grid (Auto + *) and injected a TextBlock style with TextWrapping=Wrap into the ContentPresenter resources — when WPF auto-wraps a string Content in a TextBlock, the resource lookup gives it Wrap.
2) The rail Launch Teams button ambushed operators: clicking with Teams already running (which is common when the eye-toggle has hidden Teams' windows) opened a 'Close all Teams windows now?' dialog. Operators expect Launch to mean 'show me Teams', not 'stop Teams'. Split the actions:
- Left-click: Teams not running → launch; Teams hidden → restore + foreground; Teams visible → bring to front. Always idempotent-progressive.
- Right-click: ask to stop Teams (preserves the kill path for those who want it).
TeamsLauncher.TryLaunch now collects per-attempt errors instead of swallowing them — a real failure surfaces 'ms-teams: URI → <reason>' / 'AppsFolder shell → <reason>' / 'classic Update.exe → not found at <path>' so 'No Teams found' isn't a black box.
Also added a 2nd path: explorer.exe shell:appsFolder\\\\MSTeams_8wekyb3d8bbwe!MSTeams (AppX activation via the OS's own Start-menu verb) as a fallback if the URI handler is misconfigured. Removed the broken bare-stub call to %LOCALAPPDATA%\\\\Microsoft\\\\WindowsApps\\\\ms-teams.exe — that's a 0-byte AppX placeholder that never worked outside an AppX context.
F1 help dialog's EXTERNAL CONTROL section now points operators at the LAN-reachable toggle for headless-host scenarios. Operators learn about the closed-network requirement + the netsh urlacl one-shot before they go looking for it.
When LAN-reachable mode is on, the footer's control-surface badge now shows the full http://<lan-ip>:<port> instead of just :<port>. Operators setting up a thin client can read the URL straight off the host PC's footer without having to open Settings → DISPLAY → Copy URL.
Reverts to the existing 'REST :9755 + OSC :9000' compact form when bound to localhost only — no point spelling out 127.0.0.1 since by definition only the host can reach it.
Toast was auto-dismiss-only (3s timer). Operators running a live show want to clear visual clutter without waiting — added a small X button to the right of the message that calls ToastViewModel.DismissCommand (stops timer + hides immediately).
Implementation: ToastViewModel gained a DismissCommand RelayCommand and a Hide() helper. MainWindow toast overlay gained a 20x20 button bound to the command, custom inline template (rounded transparent bg, hover lifts to Wd.Button.HoverBg).
The audio capture loop runs at ~50Hz publishing every buffer's peak via overwrite; the UI stats poll reads at 1Hz. With overwrite semantics the UI sees one of every ~50 audio frames per second — loud transients between reads were invisible to the VU meter.
New design: NdiReceiver maintains an atomic high-water mark, max-updated on each audio frame via CompareExchange CAS loop. IsoPipeline.GetStats now calls ConsumeAudioPeak() which atomically reads + resets to 0, so the next UI tick reflects the loudest sample seen in the next 1s window.
Added PeekAudioPeak() for non-consuming reads (e.g. external diagnostics dashboards that poll faster than the UI).
FakeNdiInterop gained a ReceiverAudioPeaks queue + CaptureAudioPeak override so tests can drive the audio path. 4 new tests in NdiReceiverTests cover: empty case, single-frame consume+reset, max-hold across 3 frames, no-frame leaves high-water mark untouched. 104 + 46 + 9 = 159/159 passing.
Primary button (Apply Changes, Save Preset, Confirm-Stop-All) had only an IsMouseOver trigger. Disabled state looked identical to enabled — confusing for the most-frequently-disabled button in the app.
- Disabled: drops to Wd.Accent.CyanMuted at 70% opacity + Text.Disabled foreground.
- Pressed: 85% opacity for the brief tap.
- IsKeyboardFocused: matches the hover treatment so tab-cycling lights it.
Default WPF ToolTip is cream-on-black with a thin 3D border — looks like a Win98 popup on the dark canvas. The app has dozens of tooltips on settings controls, header pills, IN-CALL bar, and per-row toggles — every one was previously rendering as the OS default.
New style: SurfaceElevated background, BorderStrong border, rounded 6px corner, monospace-friendly text wrapping at 320px so a verbose explanation doesn't stretch across the whole monitor.
Implementation note: ContentPresenter doesn't accept TextBlock.TextWrapping as an attached property — used a templated TextBlock bound directly to Content instead (every tooltip in the app passes a plain string).
Right-click on a participant row in the DataGrid surfaces actions (Toggle ISO / Restart / Open preview / Record / Copy NDI source name) but rendered with WPF's default white-on-grey ContextMenu — looked like a Notepad popup on the dark canvas.
New ContextMenu + MenuItem styles match the rest of the theme: SurfaceElevated background, rounded corners, slim cyan-tinted hover (matches the button hover treatment from the previous pass), monospaced gesture text in Wd.Text.Tertiary, no chevrons on items without submenus.
All Wd.Button.* styles set FocusVisualStyle=x:Null which suppresses WPF's default dotted focus rectangle. With no replacement IsKeyboardFocused trigger, tab-navigating gave NO visual cue — accessibility regression.
Match the keyboard-focus visual to the hover visual so mouse and keyboard land on the same affordance:
- Ghost: cyan border on focus (same as hover, minus the bg fill change so focus + hover compound visibly)
- Caption: lifts to Wd.Button.HoverBg
- RailIcon: lifts to Wd.Button.HoverBg + cyan icon
- IsoToggle: 2px cyan border (same as hover; status bg preserved)
Verified building cleanly. Tab-cycle through the IN-CALL bar / header pills / settings tabs now lights the focused control.
Default WPF ScrollBar template renders chunky Win9x line-up/line-down arrow buttons + a 3D thumb that look out of place on the dark canvas. Replace with a slim transparent track + tinted thumb pattern (Edge / VS Code / GitHub style):
- Thumb: BorderStrong by default, lifts to Text.Tertiary on hover, switches to cyan accent while dragging.
- No arrow buttons — track-clicks above/below the thumb still page-scroll via invisible RepeatButtons (preserves PageUp/PageDown behavior).
- Horizontal orientation has its own template via Style.Triggers so it picks the right Track command set (PageLeft/PageRight).
Affects every ScrollViewer in the app — settings panel, presets dialog list, Notes window, About dialog, Help cheat sheet.
The DataGrid's per-row audio level bar (in the Live column) was inert because IsoHealthStats.PeakAudioLevel always returned 0.0. Engine work needed: capture NDI audio frames, compute peak amplitude, publish through the existing stats path.
Engine:
- AudioPeakComputer (new): max-abs computation across NDI's FLTP / FLT / PCM s16 sample formats. Pure managed code, fully unit-tested (14 cases — clamping behaviour, edge cases like short.MinValue overflow, totalSamples-vs-buffer mismatch defenses).
- INdiInterop.CaptureAudioPeak (new, default-implemented): polls one audio frame, returns peak in [0,1] or null on timeout. FakeNdiInterop inherits the no-op default; production NdiInteropPInvoke overrides with real FLTP decode through a sibling RecvCaptureV3Audio import + RecvFreeAudioV3.
- NdiNative: AudioFrameV3 struct + audio-only RecvCaptureV3 binding + FreeAudioV3.
- NdiReceiver: spins up a sibling audio-capture loop alongside the existing video loop on the same lifetime. Audio failures are caught + logged but never re-thrown (a misbehaving audio path must never tear down the live video pipeline). Latest peak published via Volatile<long> (BitConverter int64 bits) so UI reads are torn-free across threads.
- IsoPipeline.GetStats: surfaces NdiReceiver.LatestAudioPeak as IsoHealthStats.PeakAudioLevel.
UI:
- ParticipantViewModel.OnStatsTick already had the decay logic (max-of-new-or-decayed-old, 0.7 multiplier) waiting for real values. No UI changes needed.
Tests: 14 new + 141 existing = 155/155 passing. 0 warnings, 0 errors.
Hover state on Wd.Button.Ghost / Caption / RailIcon / IsoToggle was barely visible — the Wd.SurfaceHover fill (#242424) was only 14 luminance points off the canvas (#0A0A0A) and 16 off the cards (#141414), so the hover state read as 'no change' to the user.
Changes:
1. Bumped Wd.SurfaceHover #242424 -> #2A2A2A and Wd.SurfaceActive #2D2D2D -> #363636 across the theme (carries to DataGrid rows etc. with the same too-subtle problem).
2. Added Wd.Button.HoverBg (#33333A) + Wd.Button.PressBg (#3F3F47) — dedicated stronger hover/press fills for buttons specifically. The slight blue tint adds chroma so hover reads obviously distinct from a flat darken.
3. Wd.Button.Ghost hover now also sets BorderBrush=Wd.Accent.Cyan — dual cue (lighter fill + bright cyan border) makes the affordance unmistakable on any surface.
4. Wd.Button.RailIcon hover sets foreground=Wd.Accent.Cyan in addition to the brighter fill (rail buttons have icons, not text — color shift on the icon is the strongest signal).
5. Wd.Button.IsoToggle keeps its status-coded background on hover (LIVE cyan / ERROR coral / NO SIGNAL amber) so at-a-glance state isn't lost; instead bumps to a 2px cyan border for the affordance.
Verified: cleanly built, app launches, hover over Refresh button now has obvious brighter fill + tooltip vs. neighbors at rest.
On a Windows host with both Ethernet (10.0.0.123) and Tailscale (169.254.83.107 link-local), the original first-hit-wins picker returned the Tailscale address — useless for the headless-host + thin-client scenario the LAN-reachable mode is designed for.
New picker prefers physical NICs (Ethernet/GigabitEthernet/Wireless80211), skips Tunnel-typed virtuals, and ranks: physical-routable > virtual-routable > APIPA. Verified against this host: now returns 10.0.0.123 instead of 169.254.83.107.
When the new ControlSurfaceLanReachable preference is on, both the REST/WebSocket control surface and the OSC bridge bind to all interfaces (http://+:port/ via HttpListener wildcard, IPAddress.Any for OSC) instead of loopback. The settings VM persists the toggle, restarts both surfaces when flipped, and surfaces a ControlSurfaceUrl computed from the first non-loopback IPv4 + a Copy button so operators can paste the URL onto a control PC.
Use case: a headless host PC runs Teams + TeamsISO; a thin client on the same LAN drives it via /ui or a Stream Deck. Closed-network deployment, no auth — documented as a trusted-LAN-only mode in docs/CONTROL-SURFACE.md, including the one-time 'netsh http add urlacl url=http://+:9755/ user=Everyone' requirement and the firewall rule.
Two real concerns from the code review on ab07297..b266623:
1. ActiveSpeaker removal poisoned the rename-window heuristic. ParticipantTracker.HandleRemoved appends to _recentlyRemoved keyed by MachineName alone; the next Participant Add on the same machine consulted that list with no kind discrimination, so an active-speaker disappearance immediately followed by a participant joining (very common: Teams renames its outputs as participants enter/leave) would cause the new participant to inherit the auto-mix's deterministic v5 GUID. New HandleAutoMixRemoved deliberately skips _recentlyRemoved — the auto-mix row's identity is already stable via the deterministic Id, so re-add restores it without the rename window.
2. IsoPipeline.State writes were not synchronized. Supervisor loop sets State on its own thread; UI thread reads from GetStats. Without volatility, the JIT could cache the field in a register and the UI would stay stuck on Receiving even after Error. Backing field is now an int read/written via Volatile.Read/Volatile.Write, matching the pattern already used for _liveReceiver / _liveSender / _liveProcessor.
Tests: 79/79 (was 78) — added ParticipantTrackerTests.ActiveSpeakerRemove_DoesNotPoisonRenameWindowForLaterParticipant which would have caught (1).
Adds a small auto-dismissing pill notification at the bottom-center of the participants area: 'Settings saved' on Apply Changes, 'Transcoder topology applied — restart Teams to take effect' after the one-click NDI groups setup. ToastViewModel owns its own DispatcherTimer and resets the dismissal countdown on successive calls, so the most recent message is always the one visible. Hooked into MainViewModel and threaded into GlobalSettingsViewModel via constructor injection.
_NEXT.md rewritten to reflect the May 2026 hardening pass: separates engine / UI / networking / Phase E.1 / diagnostics / CI / tests sections, lists every shipped item, and re-prioritizes the remaining work (Phase E.2-E.3 embedded Teams, code-signing the MSI, refresh-discovery affordance, output thumbnail previews, settings panel UX, auto-disable on departure, operator presets).
ParticipantTracker now accepts NdiSourceKind.ActiveSpeaker (Teams' auto-mix output — legacy 'MACHINE (Teams)' or current 'MACHINE (MS Teams - Active Speaker)') and surfaces it as a synthetic row in the participant list with the display name 'Active Speaker'. The operator can route it to its own normalized ISO via the same toggle every other participant uses, so vMix / OBS / Ross can subscribe to a single clean active-speaker feed.
Stable Id: derived from SHA1 of 'auto-mix:<machine>' formatted as a v5 GUID, so a discovery cycle that re-adds the source doesn't duplicate the row and the operator's ISO assignment stays bound across the rename window.
Tests: 78/78 unit (was 76) — added ParticipantTrackerTests.ActiveSpeaker_AppearsAsSyntheticAutoMixParticipant + ActiveSpeaker_ReAddOnSameMachine_PreservesId. Existing NonParticipantSources_AreIgnored still passes (only ActiveSpeaker is opted in; ScreenShare and Audio are still ignored).
Four polish improvements aimed at production-floor usability.
1. Empty-state placeholder for the participants card. When Participants.Count == 0, the DataGrid is hidden in favor of a friendly 'Waiting for Teams' panel: faded dragon mark, headline, explainer, and a four-item checklist (Teams running? NDI broadcast on? Discovery group correct? Firewall clear?). New CountToVisibilityConverter (with optional 'empty' parameter to invert) drives both the placeholder and the DataGrid visibility from the same Participants.Count source.
2. Per-pipeline error / no-signal surfacing. IsoHealthStats grows an init-only State property populated from IsoPipeline.State. ParticipantViewModel.UpdateStats maps that to a StateLabel ('LIVE' / 'NO SIGNAL' / 'ERROR' / 'STARTING' / '—'). The ISO toggle button gains DataTriggers on StateLabel — coral-tinted '● ERROR' when the supervisor gives up, amber-tinted '● NO SIGNAL' when the slate threshold trips. Operators can see at a glance which pipelines are broken.
3. JetBrains Mono Variable v2.304 (OFL) bundled at Assets/Fonts/JetBrainsMono.ttf. Wd.Font.Mono now points at the embedded font so machine names, timecodes, and stat counters render in JetBrains Mono regardless of system fonts. Falls back to Cascadia Mono / Consolas if the resource is missing.
4. Tooltip pass over every interactive control in the settings panel (framerate / resolution / aspect / audio / discovery group / output group / hide-local checkbox / Apply button / per-row Output Name textbox / per-row ISO toggle). Operators learn affordances on hover instead of by trial and error.
Tests: 76/76 unit + 9/9 NDI integration green.
Four polish items + a test pass.
1. Inter Variable (rsms/inter v3.19, OFL) is bundled at Assets/Fonts/Inter.ttf (~800 KB) and registered as a WPF Resource. WildDragonTheme.xaml's Wd.Font.Sans now points at pack://application:,,,/Assets/Fonts/#Inter so the typography matches wilddragon.net regardless of whether the user has Inter installed system-wide. Falls back to Segoe UI Variable Display if the resource is missing.
2. 'Stop all ISOs' button at the right of the participants header. Bound to a new MainViewModel.StopAllIsosCommand that snapshots the enabled list, awaits DisableIsoAsync sequentially, and silently swallows per-pipeline failures (best-effort emergency stop). CanExecute gates on whether any ISO is currently enabled.
3. WindowStateStore service persists the main window's Left/Top/Width/Height/State to %LOCALAPPDATA%\\TeamsISO\\window.json on close and restores it on SourceInitialized. Multi-monitor friendly: a saved position with no corner inside any virtual screen is rejected so a disconnected monitor doesn't strand the window off-screen.
4. Two new unit tests cover FrameProcessor's drops + duplicates accounting. 76/76 unit tests pass (was 74).
WindowChrome.CaptionHeight=44 makes the top 44px of the window a drag region; the dragon-mark rail button lives within it. Without shell:WindowChrome.IsHitTestVisibleInChrome=True, clicks were being eaten by the drag handler instead of firing the button.
Six related polish items, all building on tonight's groundwork.
1. App icon: teamsiso.ico generated from dragon-mark.png at 7 sizes (16-256), wired as ApplicationIcon in the WPF csproj, MainWindow.Icon, AboutWindow.Icon, and ARPPRODUCTICON in the WiX MSI. Taskbar / window / Add-Remove-Programs all show the dragon mark now.
2. Running incoming FPS: ring buffer of last 30 frame timestamps in IsoPipeline; ComputeFps() returns moving-average rate. Surfaced on IsoHealthStats.IncomingFps and shown in the Source column of the participants DataGrid as 'WxH · 59.94 fps'. Resets cleanly on every supervisor restart.
3. Drops counter: FrameProcessor.Stats already aggregated FramesDropped (closest-frame strategy when the receiver outpaces the processor) and FramesDuplicated; just plumbed _liveProcessor through IsoPipeline so GetStats() can read them. Exposed in the Live column under the in/out counters as a coral-tinted 'drop N'.
4. Console --version flag: prints engine version (with embedded git SHA), .NET version, OS, NDI runtime banner, expected prefix, exit-code legend, plus a wilddragon.net link. Useful for support tickets.
5. About dialog: chromeless modal with the dragon mark + version / .NET / OS / NDI runtime fields and a link to wilddragon.net. Triggered by clicking the rail logo.
6. Teams launcher Stop toggle: TeamsLauncher gains IsRunning() and StopAll(). The rail's Teams button now toggles — if Teams is up, ask to close all Teams windows via WM_CLOSE; otherwise launch as before. Confirms before stopping so we don't kill the user's call mid-transition.
Tests: 74/74 unit + 9/9 NDI integration green throughout. MSI builds clean and now embeds the dragon icon for ARP.
Two related deliverables addressing the user's morning asks.
1. Branding: Dragon WHITE.png and Wild Dragon Logo WHITE.png from the brand kit are copied into src/TeamsISO.App/Assets/ and registered as <Resource> items in the .csproj. The rail's placeholder 'W' glyph is replaced by the real dragon mark (40x40, HighQuality bitmap scaling) with a 'Wild Dragon' caption underneath.
2. NDI Access Manager automation: NdiAccessManagerConfig service reads/writes %APPDATA%\\NDI\\ndi-config.v1.json, working in JsonNode trees so we don't clobber unrelated keys. ApplyTranscoderTopology() sets groups.send=[teamsiso-input] and groups.recv=[public, teamsiso-input] so all local senders (Teams + anything else) broadcast on the private group while local receivers can still see public sources too. Engine-side, the user's per-pipeline OutputGroups override pushes TeamsISO outputs back onto Public so downstream switchers see clean ISOs.
Atomic write: temp + replace, with timestamped backup of the prior config. ReadCurrentGroups() can be used by future UI to show what's currently configured. RestoreDefaults() reverts.
Settings panel grows an 'Apply transcoder topology' button under the NDI Network section. Click writes the system config, sets the engine's discovery=teamsiso-input / output=public, refreshes the bound text boxes, and pops a dialog with a 'restart Teams' reminder + the backup path.
First step of Phase E.1 from the new spec at docs/superpowers/specs/2026-05-08-embedded-teams-orchestration.md: a third icon in the left rail launches the Microsoft Teams desktop client as a subprocess of TeamsISO so the operator doesn't have to leave the app to start a meeting.
Services/TeamsLauncher tries the ms-teams: URI first, falls back to %LOCALAPPDATA%\\Microsoft\\WindowsApps\\ms-teams.exe (new Teams), then the classic Update.exe handoff. On failure surfaces a friendly MessageBox with the install link.
The spec doc lays out the full three-phase roadmap (launcher -> window orchestration -> in-app meeting controls via Graph API or UIAutomation) and explicitly calls out what's out of scope (replacing Teams' media stack).
_NEXT.md updated to mark Phase D done and queue Phase E + remaining polish items (code-signing, Inter/JetBrains Mono font bundling, real Wild Dragon dragon-mark, drops counter, running-fps display).
Code review on d14a33a..bab29b0 turned up three real issues, fixed here.
1. EngineLogging.CreateDefault no longer mutates Serilog.Log.Logger. The static set was a belt-and-suspenders attempt to catch any code path that reaches for the singleton, but it doesn't matter (engine code uses ILogger<T>, never Serilog.Log.*) and it raced under xUnit's parallel test execution.
2. IsoPipeline stops holding a RawFrame reference for stats. The receiver-side TappedChannelWriter callback now snapshots only Width/Height into volatile ints — frame's pixel buffer is allowed to GC on its normal schedule and a late stats poll can never resurrect a dropped frame. (Today the buffer is fully managed so a use-after-free wasn't actually possible, but the snapshot pattern is the right ownership shape.)
3. App.xaml.cs's ComponentDispatcher.ThreadFilterMessage subscription now lives in a field and is unsubscribed in OnExit. Mutex release is gated on a new _ownsSingleInstanceMutex flag so the 'lost the race; shut down silently' path doesn't accidentally try to release a handle it never owned.
Plus a load-bearing comment in NdiInteropPInvoke.CreateFinder explaining why we free the UTF-8 group buffers right after the native call returns — same lifetime contract Phase B-2's CreateReceiver / CreateSender have always relied on; if it's wrong, those would fail too. The loopback discovery integration test would catch a regression.
Tests: 74/74 unit + 9/9 NDI integration green.
MainWindow drops the standard Windows title bar (WindowStyle=None + WindowChrome with CaptionHeight=44, ResizeBorderThickness=6, UseAeroCaptionButtons=False) and draws its own minimize / maximize-restore / close buttons inline in the existing header strip. The custom buttons opt into shell:WindowChrome.IsHitTestVisibleInChrome=True so clicks fire on them rather than starting a window drag.
Result: the entire top of the window is now ours, matching the Microsoft Teams desktop client's flush header look. The 'TeamsISO + by Wild Dragon' branding sits at the same baseline as the engine-status pill and the caption controls, and dragging anywhere not occupied by an interactive widget moves the window.
Caption-button styles in the theme: 46x32 hover-tinted, with the close button turning the Windows 11 #C42B1C red on hover. Maximize-button glyph swaps between the single-rectangle and overlapping-rectangles variants on StateChanged.
Drive-by: ParticipantViewModel.{FramesIn,FramesOut,IncomingResolution} setters dropped from private to public so {Run Text=...} bindings (which default to TwoWay on Run) can attach without WPF throwing 'cannot work on read-only property'.
IsoPipeline now publishes refs to its currently-live NdiReceiver and NdiSender (set by RunInnerPipelineAsync, cleared on exit) so a stats poll from any thread can read FramesCaptured / FramesSent without entangling the pipeline's lifetime with its observer. The receiver's raw-frame channel is wrapped with a TappedChannelWriter so the most recent RawFrame is captured for source-resolution display, again without changing the receiver's contract.
IsoController.GetStats() drops the stub return-Empty and instead reads the live pipeline.GetStats() outside the gate so a slow stats read can't serialize the controller's other operations.
WPF: MainViewModel runs a 1 Hz DispatcherTimer that pulls stats for every participant view-model and pushes them via UpdateStats(). ParticipantViewModel grows three displayable properties — FramesIn, FramesOut, IncomingResolution — bound into the participants DataGrid as a new 'Live' column showing the down/up frame counts and the source resolution underneath the machine name.
Tests: 74/74 unit + 9/9 NDI integration green; the existing round-trip integration test exercises the new wiring at runtime (live receiver/sender refs are set, frames flow, channels close cleanly).
