Skip to content

Run a virtual iPhone on an Apple Silicon Mac.

Swift

Skip the Audio Mix analysis session on a guest with no capture hardware

Jacky6e3acf74f9da
diff --git a/Research/0_binary_patch_comparison.md b/Research/0_binary_patch_comparison.md
index 9697d0d9..a32a12da 100644
--- a/Research/0_binary_patch_comparison.md
+++ b/Research/0_binary_patch_comparison.md
@@ -524,7 +524,7 @@ do NOT execute these).
| 19 | **Superseded 2026-09-30, as row 18.** The same `libmisfix.dylib` injected into `misagent`, where it interposes `MGCopyAnswer` / `MGCopyAnswerWithError` and answers `UniqueDeviceID` from a config file — **all bases** | `/usr/libexec/misagent` (+ `/usr/lib/libmisfix.plist`) | **Let a provisioning profile written for a device you already own install on the guest.** `0xE8008012` (`misagent: attempt to install invalid profile`) is the ordinary refusal when the profile's `ProvisionedDevices` does not list this device — correct behaviour, and distinct from row 17's `0xE8008026`. It blocks the case the user actually wants: reusing a paid team's already-registered device rather than burning a registration slot on every VM. The UDID being compared comes from `MGCopyAnswer(kMGUniqueDeviceID)`, and three cheaper routes were ruled out by measurement before any hook was written. (1) **AMFI cannot supply it**: `security.codesigning.config` is a 4-byte flags word (read back as `0x000000CC`), not a string, so `amfi_emulate_device_udid` is dead on this board. (2) **No daemon can write it**: `/private/var/root/Library/Lockdown/data_ark.plist` carries no `UniqueDeviceID` — lockdown derives it each boot. (3) **The real source is out of reach**: TXM composes `UniqueDeviceID` from the device tree's `/chosen/chip-id`, `/chosen/unique-chip-id` and `/product/udid-version` *before the kernel runs*, and `chip-id` is fixed at `0x0000FE01` while `unique-chip-id` is the ECID the SHSH blob is bound to — changing it means the VM no longer restores. So `MGCopyAnswer` is necessarily where this is done. Fix: interpose it in `misagent` only, return a `+1` copy (matching `MGCopyAnswer`'s contract) of the configured string for `UniqueDeviceID`, and fall through untouched for every other property. Configuration is `/var/db/vphone/misfix.plist`, falling back to `/usr/lib/libmisfix.plist`, cached against mtime and size so editing the file takes effect on the next query with no restart — that is the "put it in the plist and it applies" entry point. The shipped plist is **empty**: with no `UniqueDeviceID` key the interpose returns NULL and the patch is inert, so it changes nothing until someone deliberately pastes a real device's UDID in. `installMISFixDefaults` writes the fallback plist **only when it is absent**, so re-running `cfw install` never clobbers a configured value. **Accepted inconsistency, agreed with the user rather than hidden:** only `misagent`'s profile matching sees the borrowed UDID. Xcode, lockdown and `devicectl` keep reporting the VM's own (`0000FE01-…`), because those come from a different process that is not hooked. The two therefore disagree, which is exactly what makes the profile match while the device stays identifiable as itself. Declared `system-misagent-cfw-device_identity` in `com.vphone.patchset.guest.system`; no `applicability`, not `bootEssential`, on in `standard` but inert without a configured UDID. **Partially validated (2026-09-30, `06-xcode-27`):** `cfw install` reports `[+] LC_LOAD_WEAK_DYLIB /usr/lib/libmisfix.dylib -> misagent` and `misagent` runs normally on the booted guest. Installing a real paid-team IPA against a borrowed UDID is **not** yet demonstrated, for the same host-side `devicectl` reason as row 18. See `VPhoneGuestComponents/MISFix/MISFixDeviceIdentity.c`, `MISFixConfig.c` and `Research/Guest/xcode_install_signature_gate.md`. | Y | Y | Y |
| 20 | No binary change: the launchd hook inserts `libmisfix.dylib` into `SpringBoard` at spawn, beside SystemHook, and both spawn hooks do the same for `installd`, `misagent`, `lockdownd` and `remoted` (the last two for the UDID override only) — **all bases** | none (the guest environment's `launchdhook-vphone.dylib` and `SystemHook-vphone.dylib`) | **Let an installed developer-signed app launch.** SpringBoard validates the app with MIS again before launch and, on a hacktivated guest, fails row 17's `checkTrustAndAuthorization` with `0xE8008026` ("Unable to Verify App"). The hook's `MISValidateSignatureAndCopyInfoWithProgress` detour passes `RespectUppTrustAndAuthorization = false`, so the check is never called — row 17's effect with no shared-cache write, which is what makes it usable on 27.0. launchd starts SpringBoard itself rather than through xpcproxy, so the libraries it gets are the launchd hook's choice; that hook now asks the same `vpIsMISFixTarget` SystemHook does. `MISFixInstallPolicy` stays out of it. No declaration of its own: it ships with `system-launchdaemons-boot-environment`. A load-command version (`/mf` alias, `LC_SOURCE_VERSION` dropped) was committed and reverted the same day. See "One route for libmisfix, SpringBoard included (2026-09-30)" below. | Y | Y | Y |
| 21 | No binary change: SystemHook `dlopen`s `libhapticsfix.dylib` into `SpringBoard` at startup, where it replaces every `CHHapticEngine` initialiser's IMP so the initialiser returns nil — **all bases** | none (new guest dylib `VPhoneGuestComponents/HapticsFix/libhapticsfix.c`, installed at `/usr/lib/libhapticsfix.dylib`) | **Stop SpringBoard dying the first time UIKit plays a haptic.** The VM has no haptic device — Virtualization.framework exposes none on any base — and a real device without one gets nil from `CHHapticEngine`'s initialisers, so UIKit runs with no feedback, as on every non-Pro iPad. A guest reporting an iPad Pro model instead has an OS that *expects* the haptic stack, and CoreHaptics treats its absence as an internal error: measured on `iPad16,1 26.6.2 (23G90)`, 4h14m after launch, the first feedback event ran `-[_UIFeedbackCoreHapticsIgnoreCaptureHapticsOnlyEngine _internal_createCoreHapticsEngine]` → `-[CHHapticEngine initWithAudioSession:sessionIsShared:options:error:]` → `createHapticPlayerWithOptions:` → `Haptic_RaiseException` → `objc_exception_throw`, and the unwind resumed at an address inside a `---/---` Memory Tag 255 COW region next to the commpage — `EXC_BAD_ACCESS (SIGBUS)`, `KERN_PROTECTION_FAILURE`, a dead SpringBoard. The hook restores the no-hardware answer before UIKit can reach an initialiser: the constructor `dlopen`s CoreHaptics (it is not loaded that early, and the class must be in the runtime before UIKit asks), then `method_setImplementation` on every selector that is `init` or `init` + uppercase part — covering `initWithAudioSession:sessionIsShared:options:error:` today and any renamed sibling on other bases — with one IMP that ignores its arguments and returns nil, the framework's own answer on haptic-less hardware. Scope is SpringBoard only (`vpIsSpringBoard` in SystemHook, loaded before the app/bootstrap gate that would otherwise return first): it is the one UIKit process measured creating the haptics-only engine without asking CoreHaptics first, and apps that follow the documented `capabilitiesForHardware` path already answer no on the guest. Loaded like the camera hooks rather than inserted like libmisfix because interposing must be in place at launch while a method swizzle just has to run before first use; `DISABLE_TWEAKS` and safe mode skip it with everything else SystemHook loads. Ships with `system-launchdaemons-boot-environment`; existing guests receive it through the vphoned environment update (a missing file reports no hash, which is what triggers the push) and need a respring, which is deliberately the caller's call. See `VPhoneGuestComponents/HapticsFix/libhapticsfix.c`. | Y | Y | Y |
-| 22 | No binary change: `libvcamcaptured.dylib` (already loaded into `cameracaptured` by SystemHook) wraps `+[FigCaptureSourceBackingsProvider sharedCaptureSourceBackingsProvider]` from its constructor; when the original returns nil it returns a microphone-only provider instead — **iOS 26.6+ layout, measured on iOS 27.0** | none (`VPhoneGuestComponents/VCamCaptured/Microphone/VCamMicrophoneSource.m`) | **Give an audio-only `AVCaptureSession` its microphone on a guest with no camera.** The built-in sources come from one provider that CMCapture builds only after `BWFigCaptureDeviceVendor` creates the `Default` camera device; a VM has no ISP plugin, so `Cannot create device without create function!`, -12786, `Wiping com.apple.cameracapture.volatile`, and no provider — the microphone, described by the same `AVCaptureSession.plist` call, is lost with the camera, and iOS 27 Voice Memos fails with `CaptureSessionRecorderError` 1 (`failedToCreateAudioDevice`). The wrapper feeds the exported `FigCaptureCreateSourceInfoArrayFromDeviceAndModelSpecificPlist` a NULL device and the shipped product plist (the guest's own `VPHONE600` has none; `D47` on the iPhone17,3 image) reduced to its `"soun"` entries, without persisting, and builds the provider with `-initWithSourceInfoDictionaries:commonSettings:`. A provider the daemon built itself passes through untouched. See `Research/Guest/ios27_capture_microphone_source.md`. | Y | Y | Y |
+| 22 | No binary change: `libvcamcaptured.dylib` (already loaded into `cameracaptured` by SystemHook) wraps `+[FigCaptureSourceBackingsProvider sharedCaptureSourceBackingsProvider]` from its constructor; when the original returns nil it returns a microphone-only provider instead — **iOS 26.6+ layout, measured on iOS 27.0** | none (`VPhoneGuestComponents/VCamCaptured/Microphone/VCamMicrophoneSource.m`) | **Give an audio-only `AVCaptureSession` its microphone on a guest with no camera.** The built-in sources come from one provider that CMCapture builds only after `BWFigCaptureDeviceVendor` creates the `Default` camera device; a VM has no ISP plugin, so `Cannot create device without create function!`, -12786, `Wiping com.apple.cameracapture.volatile`, and no provider — the microphone, described by the same `AVCaptureSession.plist` call, is lost with the camera, and iOS 27 Voice Memos fails with `CaptureSessionRecorderError` 1 (`failedToCreateAudioDevice`). The wrapper feeds the exported `FigCaptureCreateSourceInfoArrayFromDeviceAndModelSpecificPlist` a NULL device and the shipped product plist (the guest's own `VPHONE600` has none; `D47` on the iPhone17,3 image) reduced to its `"soun"` entries, without persisting, and builds the provider with `-initWithSourceInfoDictionaries:commonSettings:`. A provider the daemon built itself passes through untouched. While that provider is in use, the same dylib also makes `-[AudioRemixSessionManager startNewSessionBlocking]` return -16992 (its own session-failed code) without building the SoundAnalysis Audio Mix session, whose neural net faults cameracaptured 2 s into every spatial recording on the guest (`BNNSGraphContextMakeStreaming`, `KERN_INVALID_ADDRESS 0x300`); `BWAudioRemixAnalysisMetadataNode` already passes audio through when the session does not start. `VCamCaptured/Microphone/VCamMicrophoneRemix.m`. See `Research/Guest/ios27_capture_microphone_source.md`. | Y | Y | Y |
### Swift port status — the eight DSC patchers (2026-09-23)
diff --git a/Research/Guest/ios27_capture_microphone_source.md b/Research/Guest/ios27_capture_microphone_source.md
index f02b14eb..2617e1fa 100644
--- a/Research/Guest/ios27_capture_microphone_source.md
+++ b/Research/Guest/ios27_capture_microphone_source.md
@@ -171,20 +171,100 @@ mic source: no AVCaptureSession.plist for model VPHONE600
The second line is only reached when the original returned nil, so the
daemon's own provider is nil at its first source query, as §2 reads. It also
showed that the guest's model has no plist, which the product-folder fallback
-in §5 answers. That second build has not run in a guest yet: its install
-stopped at Launchpad's administrator prompt.
-
-Still to check, in order, once it runs:
-
-1. `vcamcaptured.log`: `mic source: model D47, 1 source info(s), provider …,
- hasMicSource=1`, then `serving the microphone-only one`, and no
- cameracaptured crash report.
-2. Voice Memos Record: no `CaptureSessionRecorderError`; cameracaptured's
- `captureSession_SetConfiguration` reads `Cam/Audio:0/1`.
-3. `/var/mobile/vpquery.log` gets a `stream 0: … reads, in … frames` line,
- and the `.m4a` under the Voice Memos app group's `Recordings/` decodes
- (`afconvert` to WAV) to something that is not zeros.
-
-The mic-only provider hands out a source; whether `BWAudioSourceNode` then
-runs in cameracaptured without anything else from the camera device is not
-established until step 2.
+in §5 answers.
+
+With the fallback (bundle `2.4.0-local.473d5071`, deployed by the
+coordinating session):
+
+```
+mic source: no AVCaptureSession.plist for model VPHONE600
+mic source: model D47, 1 source info(s), provider 0x658ca0f00, hasMicSource=1
+mic source: daemon built no provider; serving the microphone-only one
+```
+
+Voice Memos then gets its device and runs the session: `-[AVCaptureSession
+addInput:]: <AVCaptureDeviceInput [iPhone Microphone]>`, an
+`AVCaptureMovieFileOutput` and an `AVCaptureAudioDataOutput`, and
+cameracaptured logs `captureSession_SetConfiguration … Cam/Audio:0/1`,
+`starting source node <AudioDevice, BWAudioSourceNode>`, `AURemoteIO … input
+client: 6 ch, 48000 Hz` and `captureSession_FileSinkStartRecording`. The
+sound plugin delivers (`stream 0: 2.15 s, 24 reads, in 98304 frames …`).
+Every recording then ends about 2 s in with cameracaptured crashing; §7.
+
+## 7. The Audio Mix analysis that crashes cameracaptured
+
+The crash (`cameracaptured-2026-10-03-195510.ips` and three more):
+`EXC_BAD_ACCESS (SIGSEGV) KERN_INVALID_ADDRESS at 0x300` in
+`BNNSGraphContextMakeStreaming`, under `MIL2BNNS::loadContext` →
+`NeuralNet::NeuralNet` → `AUNeuralNet::Initialize` → `DSPGraph` →
+SoundAnalysis → `-[AudioRemixSessionManager startNewSessionBlocking]`.
+Voice Memos sees `AVFoundationErrorDomain -11819` (media services were reset).
+
+### Why there is a remix session at all
+
+* Voice Memos records spatial audio when MobileGestalt answers
+ `DeviceSupportsSpatialAudioCapture` (`RCDeviceSupportsSpatialAudioCapture`,
+ behind `RCSpatialAudioCaptureIsAvailable`, VoiceMemos framework). The
+ guest answers as the D47 it was built from.
+* `CaptureSessionRecorder` then requires first-order ambisonics: in the app
+ binary, right after creating the `AVCaptureDeviceInput`, it calls
+ `isMultichannelAudioModeSupported:2` and, only if YES,
+ `setMultichannelAudioMode:2` (0x1001b274c–0x1001b2760); NO branches to a
+ `swift_allocError` (0x1001b28bc), the same error type its other failures
+ use. There is no stereo fallback in that path.
+* `isMultichannelAudioModeSupported:` ends in
+ `-[AVCaptureFigAudioDevice isAudioCaptureModeSupported:]`, which answers
+ mode 2 from the source attribute `cinematicAudioCaptureSupported` and mode 1
+ from `builtInMicrophoneStereoAudioCaptureSupported` (the two
+ `objc_msgSend` stubs it tail-calls). D47's microphone entry sets both.
+* The movie-file head pipeline (`_buildMovieFileSinkHeadPipeline…`,
+ 0x1b0728ac0) adds a `BWAudioRemixAnalysisMetadataNode` next to the
+ "Cinematic Audio Converter" when its metadata configuration asks for one;
+ `FigCaptureMetadataObjectConfigurationRequiresSpatialAudioMix` keys that on
+ `kCMMetadataIdentifier_QuickTimeMetadataSpatialAudioMix`. That the
+ ambisonic mode is what puts this identifier in Voice Memos' configuration
+ is inferred, not traced.
+ Its start marker runs `startNewSessionBlocking`, which builds a SoundAnalysis
+ `SNMovieRemix` session; that is the neural net that faults.
+
+So stripping `cinematicAudioCaptureSupported` from the source (option a)
+would not give a plain recording: Voice Memos would stop at
+`isMultichannelAudioModeSupported:` with its own error instead.
+
+### Why the neural net faults
+
+Not established. audiomxd loads and runs a neural net through the same
+`MIL2BNNS` path in the same recording (the SpatialCapture wind model,
+`graph size is 1032192 bytes`, `context size is 3752 bytes`, `Successfully
+loaded`), from a `.mil` it compiles to `.ir` in its own cache. cameracaptured's
+remix model is shipped as `.ir` (`MIL2BNNS extension is '.ir'`, `graph size is
+1359872 bytes`) and faults in the next step, making the streaming context.
+That the precompiled graph targets something the VM's CPU or BNNS build does
+not have is a guess.
+
+### What the hook does
+
+The node is written for a session that does not start
+(`-[BWAudioRemixAnalysisMetadataNode renderSampleBuffer:forInput:]`,
+0x1b0600d88):
+
+* a non-zero return from `startNewSessionBlocking` (0x1b06010bc `cbnz w0`)
+ skips setting `_shouldSendData` and goes to `emitSampleBuffer:` — the audio
+ passes through;
+* `submitAudioBuffer:` is called only after `sessionReady` returns YES;
+* `-finishAndGetResultsBlockingWithStartingPTS:andEndingPTS:` returns an
+ error when there is no subscriber (`ldr x0, [x0, #0x10]; cbz`) instead of
+ blocking, and `_sendRemixMetadataSampleBuffer` ignores its result.
+
+`VCamCaptured/Microphone/VCamMicrophoneRemix.m` replaces
+`-[AudioRemixSessionManager startNewSessionBlocking]` (`- (int)`, checked
+against the runtime type encoding before wrapping). While the daemon is
+running on the microphone-only provider, it returns -16992, the code the
+method's own failure path signals, without creating the session; otherwise it
+calls the original. The recording keeps its ambisonic audio and loses only the
+Audio Mix analysis track's data.
+
+Not yet run in a guest. To check: `vcamcaptured.log` has `remix: wrapped …` at load and
+`Audio Mix analysis session not started (-16992)` at the first recording;
+cameracaptured's pid survives a recording; then the `vpquery.log` line and a
+non-silent `.m4a`.
diff --git a/VPhoneGuestComponents/VCamCaptured/Microphone/VCamMicrophoneRemix.m b/VPhoneGuestComponents/VCamCaptured/Microphone/VCamMicrophoneRemix.m
new file mode 100644
index 00000000..7fca029d
--- /dev/null
+++ b/VPhoneGuestComponents/VCamCaptured/Microphone/VCamMicrophoneRemix.m
@@ -0,0 +1,60 @@
+// No Audio Mix analysis on a guest that has no capture hardware.
+//
+// A spatial recording (Voice Memos on iOS 27 records first-order ambisonics,
+// because the microphone source built from the product plist advertises
+// cinematic audio) gets a BWAudioRemixAnalysisMetadataNode in its movie-file
+// pipeline. At the start marker the node calls
+// -[AudioRemixSessionManager startNewSessionBlocking], which builds a
+// SoundAnalysis movie-remix session; initialising that session's neural net
+// faults in cameracaptured on the guest (BNNSGraphContextMakeStreaming,
+// KERN_INVALID_ADDRESS 0x300), about 2 s into every recording.
+//
+// The node already copes with a session that does not start: a non-zero
+// return leaves _shouldSendData clear, buffers are passed through
+// (emitSampleBuffer:), submitAudioBuffer: is skipped while sessionReady is NO,
+// and -finishAndGetResultsBlockingWithStartingPTS:andEndingPTS: returns an
+// error instead of waiting when there is no subscriber. So while the daemon
+// runs on the microphone-only provider (no camera device: a VM), the method
+// returns the error its own failure path signals, without creating the
+// session. With a real capture device it runs unchanged.
+// See Research/Guest/ios27_capture_microphone_source.md §7.
+
+#include "VCamCapturedPrivate.h"
+
+// The code -startNewSessionBlocking signals when its session does not start.
+static const int kVccRemixSessionNotStarted = -16992;
+
+static IMP vcc_remix_start_orig = NULL;
+static BOOL vcc_remix_skip_logged = NO;
+
+static int vcc_remix_start_hook(id self, SEL _cmd) {
+ if (!vcc_microphone_only_source_active()) {
+ return ((int (*)(id, SEL))vcc_remix_start_orig)(self, _cmd);
+ }
+ if (!vcc_remix_skip_logged) {
+ vcc_remix_skip_logged = YES;
+ vcc_log(@" mic source: no capture hardware; Audio Mix analysis session not started (%d)",
+ kVccRemixSessionNotStarted);
+ }
+ return kVccRemixSessionNotStarted;
+}
+
+void vcc_install_remix_session_skip(void) {
+ Class cls = NSClassFromString(@"AudioRemixSessionManager");
+ SEL sel = NSSelectorFromString(@"startNewSessionBlocking");
+ Method m = cls ? class_getInstanceMethod(cls, sel) : NULL;
+ if (!m) {
+ vcc_log(@" remix: -[AudioRemixSessionManager startNewSessionBlocking] missing");
+ return;
+ }
+ // The node branches on a 32-bit status (cbnz w0); refuse any other shape.
+ const char *types = method_getTypeEncoding(m);
+ if (!types || (types[0] != 'i' && types[0] != 'I')) {
+ vcc_log(@" remix: startNewSessionBlocking has type %s, not wrapped",
+ types ?: "?");
+ return;
+ }
+ vcc_remix_start_orig = method_setImplementation(m, (IMP)vcc_remix_start_hook);
+ vcc_log(@" remix: wrapped -[AudioRemixSessionManager startNewSessionBlocking] (orig imp=%p)",
+ vcc_remix_start_orig);
+}
diff --git a/VPhoneGuestComponents/VCamCaptured/Microphone/VCamMicrophoneSource.m b/VPhoneGuestComponents/VCamCaptured/Microphone/VCamMicrophoneSource.m
index 2ab874fb..f92c16e8 100644
--- a/VPhoneGuestComponents/VCamCaptured/Microphone/VCamMicrophoneSource.m
+++ b/VPhoneGuestComponents/VCamCaptured/Microphone/VCamMicrophoneSource.m
@@ -41,7 +41,7 @@ typedef void (*VccSourceInfoFn)(void *device, CFDictionaryRef plist,
static IMP vcc_shared_provider_orig = NULL;
static id vcc_mic_provider = nil;
static BOOL vcc_mic_provider_failed = NO;
-static BOOL vcc_mic_provider_logged_use = NO;
+static BOOL vcc_mic_provider_served = NO;
static BOOL vcc_provider_has_mic(id provider) {
Ivar iv = class_getInstanceVariable(object_getClass(provider), "_hasMicSource");
@@ -153,14 +153,18 @@ static id vcc_shared_provider_hook(id self, SEL _cmd) {
vcc_mic_provider = vcc_build_mic_provider(self);
vcc_mic_provider_failed = (vcc_mic_provider == nil);
}
- if (vcc_mic_provider && !vcc_mic_provider_logged_use) {
- vcc_mic_provider_logged_use = YES;
+ if (vcc_mic_provider && !vcc_mic_provider_served) {
+ vcc_mic_provider_served = YES;
vcc_log(@" mic source: daemon built no provider; serving the microphone-only one");
}
return vcc_mic_provider;
}
}
+BOOL vcc_microphone_only_source_active(void) {
+ return vcc_mic_provider_served;
+}
+
void vcc_install_microphone_source(void) {
Class cls = NSClassFromString(@"FigCaptureSourceBackingsProvider");
if (!cls) {
diff --git a/VPhoneGuestComponents/VCamCaptured/VCamCapturedPrivate.h b/VPhoneGuestComponents/VCamCaptured/VCamCapturedPrivate.h
index 2ea38d68..27b33369 100644
--- a/VPhoneGuestComponents/VCamCaptured/VCamCapturedPrivate.h
+++ b/VPhoneGuestComponents/VCamCaptured/VCamCapturedPrivate.h
@@ -43,6 +43,15 @@ void vcc_swizzle_method(Class cls, SEL sel, IMP newImp, IMP *outOrig);
// in the constructor, before the daemon's first source query.
void vcc_install_microphone_source(void);
+// YES once the daemon has been served the microphone-only provider, which
+// happens only when it has no capture device of its own.
+BOOL vcc_microphone_only_source_active(void);
+
+// While that provider is in use, -[AudioRemixSessionManager
+// startNewSessionBlocking] fails the way it does when its session cannot
+// start, instead of running the Audio Mix neural net that faults on a guest.
+void vcc_install_remix_session_skip(void);
+
void vcc_install_synthetic(void);
void vcc_start_frame_receiver(void);
void vcc_install_endpoint_hook(void);
diff --git a/VPhoneGuestComponents/VCamCaptured/libvcamcaptured.m b/VPhoneGuestComponents/VCamCaptured/libvcamcaptured.m
index beded01c..0cd8e0e7 100644
--- a/VPhoneGuestComponents/VCamCaptured/libvcamcaptured.m
+++ b/VPhoneGuestComponents/VCamCaptured/libvcamcaptured.m
@@ -92,6 +92,7 @@ __attribute__((constructor)) static void vcc_init(void) {
// Not delayed with the camera hooks below: the daemon asks for its
// capture sources as soon as a client connects.
vcc_install_microphone_source();
+ vcc_install_remix_session_skip();
// Schedule install after the daemon has run its own init. The delay
// gives FigCaptureSourceServerStart's `dispatch_once` block time to