I've implemented a custom system extension VPN for macOS using Packet Tunnel Provider. The VPN is configured with on-demand, and a rule to always connect whenever there's traffic: onDemandRules = [NEOnDemandRuleConnect()] As expected, if the VPN isn't active, all traffic gets blocked until it is ready. Not expected: In the following scenario, there is some 'traffic leak': Use only WiFi (not wired cable) Connect the VPN Disable the WiFi and wait for the VPN to disconnect Enable the WiFi Some packets are routed outside the VPN, and aren't being blocked Some moments after, all traffic will be blocked, and the VPN will start the 'connecting' process. Is the above scenario a 'known' issue? Can it be a race condition in the OS, where some packets can be sent after the network is brought back before the VPN process starts? Is there any way to fix this problem? P.S: I'm not using flags such as 'capture all network'

Is it somehow possible to get the transport layer (UDP and TCP) payload amounts for TLS or QUIC connections established via the Network framework? (From within the app itself that establishes the connections.) I am currently using the ntstat.h kernel socket calls, but I hope there is a simpler solution. With ntstat, I have not yet been able to observe a specific connection. I have to search for the connection I am looking in all (userspace) connections.

NWPathMonitor appears to retain itself (or is retained by some internal infrastructure) once it has been started until cancelled. This seems like it can lead to memory leaks if the references to to the monitor are dropped. Is this behavior documented anywhere? func nwpm_self_retain() { weak var weakRef: NWPathMonitor? autoreleasepool { let monitor: NWPathMonitor = NWPathMonitor() weakRef = monitor monitor.start(queue: .main) // monitor.cancel() // assertion fails unless this is called } assert(weakRef == nil) } nwpm_self_retain()

We’re implementing VPN application using the WireGuard protocol and aiming to support both split-tunnel and per-app VPN configurations. Each mode works correctly on its own: per-app VPN functions well when configured with a full tunnel and split-tunnel works as expected when per-app is disabled. However, combining both configurations leads to issues. Specifically, the routing table is not set up properly, resulting in traffic that should not be routed through the tunnel is routed through the tunnel. Detailed description: Through our backend, we are pushing these two plist files to the iPad one after the other: VPN config with allowed IPs 1.1.1.1/32 <?xml version="1.0" encoding="utf-8"?> <!DOCTYPE plist PUBLIC "-//Apple Inc//DTD PLIST 1.0//EN" http://www.apple.com/DTDs/PropertyList-1.0.dtd> <plist version="1.0"> <dict> <key>PayloadUUID</key> <string>3fd861df-c917-4716-97e5-f5e96452436a</string> <key>PayloadVersion</key> <integer>1</integer> <key>PayloadOrganization</key> <string>someorganization</string> <key>PayloadIdentifier</key> <string>config.11ff5059-369f-4a71-afea-d5fdbfa99c91</string> <key>PayloadType</key> <string>Configuration</string> <key>PayloadDisplayName</key> <string> test</string> <key>PayloadDescription</key> <string>(Version 13) </string> <key>PayloadRemovalDisallowed</key> <true /> <key>PayloadContent</key> <array> <dict> <key>VPN</key> <dict> <key>AuthenticationMethod</key> <string>Password</string> <key>ProviderType</key> <string>packet-tunnel</string> <key>OnDemandUserOverrideDisabled</key> <integer>1</integer> <key>RemoteAddress</key> <string>172.17.28.1:51820</string> <key>OnDemandEnabled</key> <integer>1</integer> <key>OnDemandRules</key> <array> <dict> <key>Action</key> <string>Connect</string> </dict> </array> <key>ProviderBundleIdentifier</key> <string>some.bundle.id.network-extension</string> </dict> <key>VPNSubType</key> <string>some.bundle.id</string> <key>VPNType</key> <string>VPN</string> <key>VPNUUID</key> <string>d2773557-b535-414f-968a-5447d9c02d52</string> <key>OnDemandMatchAppEnabled</key> <true /> <key>VendorConfig</key> <dict> <key>VPNConfig</key> <string> Some custom configuration here </string> </dict> <key>UserDefinedName</key> <string>TestVPNServerrra</string> <key>PayloadType</key> <string>com.apple.vpn.managed.applayer</string> <key>PayloadVersion</key> <integer>1</integer> <key>PayloadIdentifier</key> <string>vpn.5e6b56be-a4bb-41a5-949e-4e8195a83f0f</string> <key>PayloadUUID</key> <string>9bebe6e2-dbef-4849-a1fb-3cca37221116</string> <key>PayloadDisplayName</key> <string>Vpn</string> <key>PayloadDescription</key> <string>Configures VPN settings</string> <key>PayloadOrganization</key> <string>someorganization</string> </dict> </array> </dict> </plist> Command to set up per-app with Chrome browser <?xml version="1.0" encoding="utf-8"?> <!DOCTYPE plist PUBLIC "-//Apple Inc//DTD PLIST 1.0//EN" http://www.apple.com/DTDs/PropertyList-1.0.dtd> <plist version="1.0"> <dict> <key>Command</key> <dict> <key>Settings</key> <array> <dict> <key>Identifier</key> <string>com.google.chrome.ios</string> <key>Attributes</key> <dict> <key>VPNUUID</key> <string>d2773557-b535-414f-968a-5447d9c02d52</string> <key>TapToPayScreenLock</key> <false /> <key>Removable</key> <true /> </dict> <key>Item</key> <string>ApplicationAttributes</string> </dict> </array> <key>RequestType</key> <string>Settings</string> </dict> <key>CommandUUID</key> <string>17ce3e19-35ef-4dbc-83d9-4ca2735ac430</string> </dict> </plist> From the log we see that our VPN application set up allowed IP 1.1.1.1 via NEIPv4Settings.includedRoutes but system routing all of the Chrome browser traffic through our application. Is this expected Apple iOS behavior, or are we misconfiguring the profiles?

fetchCurrent(completionHandler:) https://developer.apple.com/documentation/networkextension/nehotspotnetwork/fetchcurrent(completionhandler:) The same code works fine in Xcode 16, but when I run the same project in Xcode 26, it doesn't work

Hello! I develop transparent proxy based application, and I'm receiving a lot of crash reports from macOS 15.5 for crash in __88-[NEExtensionAppProxyProviderContext setInitialFlowDivertControlSocket:extraValidation:]_block_invoke.90 when stopping. Even very old versions of my software started crashing on macOS 15.5. I checked my extension that it correctly calls setTunnelNetworkSettings:nil on proxy stop, but crash is still here. Does anybody else have this problem? Do you know any workaround for it?

I'm working with Apple's SimpleURLFilter sample project and consistently encountering an error when trying to implement the URL filter. Here are the details: Setup: Downloaded the official SimpleURLFilter sample project from Apple Set the developer team for both targets (main app and extension) Built and ran the PIR server on my laptop using Docker as per the sample instructions Built the iOS project on my iPhone running iOS 26.0.1 Server is accessible at my Mac's IP address on port 8080 Configuration: PIR Server URL: http://[my-mac-ip]:8080 Authentication Token: AAAA (as specified in service-config.json) Privacy Pass Issuer URL: (left empty) Fail Closed: enabled Code Changes: The only modifications I made were: Updated bundle identifiers to include my team identifier Updated PIR server's service-config.json to match: com.example.apple-samplecode.SimpleURLFilter[TEAM_ID].url.filtering Modified URLFilterControlProvider.swift: Added existingPrefilterTag: String? parameter to fetchPrefilter() method Added tag: "bloom_filter" parameter to NEURLFilterPrefilter initializer Issue: After configuring the filter and entering my passcode in Settings, I consistently see: Received filter status change: <FilterStatus: 'starting'> Received filter status change: <FilterStatus: 'stopped' errorMessage: 'The operation couldn't be completed. (NetworkExtension.NEURLFilterManager.Error error 9.)'> Questions: What does NEURLFilterManager.Error error 9 specifically indicate? Could the URLFilterControlProvider modifications be causing this issue? Are there debugging steps to get more detailed error information? Any guidance would be appreciated!

We use as content filter in our app to monitor flows, we gather data about the flow and block flows deemed suspicious. Our content filter is activated/deactivated by a UI app but the flows are reported via XPC to a separate daemon process for analysis. As of macOS 15, we are seeing cases where flows are missing or flows are not received at all by the content filter. The behaviour is not consistent, some devices seem to receive flows normally but others don't. It appears Intel devices are much less prone to showing the problem, whereas Arm devices routinely exhibit missing flows. On macOS 14 or earlier, there is no sign of missing flows. Testing on earlier beta versions of macOS 15 did not appear to show the problem, however I can't rule out if issue was present but it wasn't spotted. Experimenting with simple examples of using a content filter (e.g. QNE2FilterMac) does not appear to reproduce the issue. Questions, What has changed between macOS 14 and 15 that could be the cause of the lack of flows? Is our approach to using an app activated content filter reporting to a daemon connected via XPC unsupported?

In our App, we have a network extension with a NEAppPushProvider subclass running. We run the following steps Setup a dual-band wireless router per the following: Broadcasting 2.4 GHz and 5 GHz channels Same SSID names for both channels Connected to the production network to the router DHCP assigning addresses in the 10.1.x.x network Connect the mobile device to the 5 GHz network (if needed, turn off the 2.4 GHz network temporarily; once the device connects to the 5 GHz network, the 2.4 GHz network can be turned back on). Create a NEAppPushManager in the App, using the SSID from the above mentioned network and set it to the matchSSIDs property. Call saveToPreferences() on the push manager to save. A. We have UI that shows the extension has been started and it has connected to the server successfully. Walk out of the range of the 5 GHz channel of the router, but stay within range of the 2.4ghz channel. Wait for the mobile device to connect to the 2.4 GHz channel. Expected: The extension would reconnect to the 2.4ghz network. Observed: The extension does not reconnect. Checking the logs for the extension we see that the following was called in the push provider subclass. stop(with:completionHandler:) > PID: 808 | 🗒️🛑 Stopped with reason 3: "noNetworkAvailable" The expectation is that start() on the NEAppPushProvider subclass would be called. Is this an incorrect expectation? How does the NEAppPushProvider handle same network SSID roaming among various band frequencies? I looked at the documentation and did not find any settings targeting 2.4 or 5 ghz networks. Please advise on what to do.

Hello, Since the release of iOS 26.0, we are seeing DNS traffic being blocked from within our NEPacketTunnelExtension on some devices. We have not isolated exact reproduction steps, but DNS resolves successfully for a period of time after enabling "iCloud Private Relay" (varying from 1-day to 2-weeks), until it then fails as MDNSResponder then returns: mDNSResponder [Q37046] DetermineUnicastQuerySuppression: Query suppressed for <mask.hash: 'REDACTED'> Addr (blocked by policy) DNS resolution continues to fail for all domains with the above until the device is rebooted. The Packet Tunnel intentionally does not have a DNS server set and this occurs for traffic from the Extension yet off-tunnel, which needs resolution from the system DNS server (and this configuration works perfectly for a period of time before being "blocked by policy"). The following do not resolve the issue once DNS queries are being "blocked by policy" on affected devices: disconnecting then reconnecting the vpn; toggling airplane mode for 10+ seconds; switching connection between WiFi & cellular data; disabling iCloud Private Relay. We have currently only seen this on unmanaged devices running iOS 26.0 or 26.1 beta and with iCloud Private Relay enabled. We did not see this issue on iOS 16,17 nor 18. We also have not yet seen this when iCloud Private Relay is disabled nor on iOS 26.0.1, however we cannot confirm whether they too are also affected. Is there a known a bug with iOS 26.0 & 26.1 Beta 1 that could cause this? How can we prevent DNS requests from NEPacketTunnelExtension being sporadically "blocked by policy" until the device is rebooted? Many thanks in advance.

The environment: macOS 12.0 ~ 15.6 A NetworkExtension NEFilterDataProvider configured with filterSockets = YES, filterPackets = NO, and it doesn't actually block any network connection. QQMusic (download: https://y.qq.com/n/ryqq/download_detail/mac?ADTAG=YQQ) is constantly playing. Any of the following operations can reproduce the issue: Kill the NetworkExtension process and then restarted by the system. Disable the NEFilterDataProvider, and then enable it. When this problem occurs, there are two different phenomena on the NetworkExtension process: It is zombie, or is in high CPU state (100%). When the NetworkExtension process is zombie, obviously, the new network connections will enter it, and they can't be disposed by the old zombie process, so the network is disconnected. Spindump-qqmusic-ne-zombie When the NetworkExtension process is in high CPU state, its thread DispatchQueue "NEFilterExtensionProviderContext queue" is blocked in the kernel when calling close. Spindump-qqmusic-ne-cpuhigh In most cases, the network will recover after stopping QQ Music, that is the suspended zombie NetworkExtension process will exist or the cpu of it return to normal. To reproduce the issue in a simple environment, I have tried many ways to simulate the network behavior of QQMusic, but all failed. It seems that this issue is caused by UDP traffic of QQMusic, because everything is ok after blocking the UDP connections of QQMusic (the music is still playing at this time) in the NEFilterDataProvider.

I have a Vision Pro app, which I intend to use Apple-Hosted Background Assets for some of my videos after watching: https://developer.apple.com/videos/play/wwdc2025/325 I added a Apple-Hosted, Managed extension. New Target -> Background Download -> Apple-Hosted, Managed After creating an Archive, I tried uploading it to TestFlight, it complains about a DTPlatformName error in my Info.plist. So I added the following : <key>DTPlatformName</key> <string>xros</string> With which, I managed to upload the app with the extension to TestFlight. However, when I tried installing the app on TestFlight to Vision Pro, it gives me an error that says the app cannot be verified. Any help or pointers is greatly appreciated. Info.plist Entitlements

Hello, I have been playing around the the SimpleURLFilter sample code. I keep getting this error upon installed the filter profile on the device: mapError unexpected error domain NEVPNConnectionErrorDomainPlugin code 7 which then causes this error: Received filter status change: <FilterStatus: 'stopped' errorMessage: 'The operation couldn’t be completed. (NetworkExtension.NEURLFilterManager.Error error 14.)'> I can't find much info about code 7. Here is the configuration I am trying to run: <Configuration: pirServerURL: 'http://MyComputer.local:8080' pirAuthenticationToken: 'AAAA' pirPrivacyPassIssuerURL: 'http://MyComputer.local:8080' enabled: 'true' shouldFailClosed: 'true' controlProviderBundleIdentifier: 'krpaul.SimpleURLFilter.SimpleURLFilterExtension' prefilterFetchInterval: '2700.0'>

1) Blocked page UX When a URL is blocked, the browser typically shows a generic error like “"Safari cannot open the page because it couldn’t load any data,” with no indication that the page was blocked by a policy. Is there any plan to add an API that allows developers to present a custom “blocked” page or remediation action, similar to NEFilterControlProvider’s remediationMap? Even a minimal hook (custom HTML, deep link, or support URL) would make the experience clearer for users. 2) Cross‑app link‑opening behavior With a block rule in place, direct navigation in Safari is blocked as expected. However, tapping the same URL in a messaging app (e.g., WhatsApp) opens Safari - and the page loads, not blocked. Repro steps: Configure a URL Filter extension that blocks https://example.com. Case A: Open a browser and type the URL in the address bar → blocked (expected). Case B: Tap the same URL in WhatsApp (or another messenger) → a browser opens and the page loads (unexpected). iOS version - 26.0

Our product (rockhawk.ca) uses the Multipeer Connectivity framework for peer-to-peer communication between multiple iOS/iPadOS devices. My understanding is that MC framework communicates via three methods: 1) infrastructure wifi (i.e. multiple iOS/iPadOS devices are connected to the same wifi network), 2) peer-to-peer wifi, or 3) Bluetooth. In my experience, I don't believe I've seen MC use Bluetooth. With wifi turned off on the devices, and Bluetooth turned on, no connection is established. With wifi on and Bluetooth off, MC works and I presume either infrastructure wifi (if available) or peer-to-peer wifi are used. I'm trying to overcome two issues: Over time (since iOS 9.x), the radio transmit strength for MC over peer-to-peer wifi has decreased to the point that range is unacceptable for our use case. We need at least 150 feet range. We would like to extend this support to watchOS and the MC framework is not available. Regarding #1, I'd like to confirm that if infrastructure wifi is available, MC uses it. If infrastructure wifi is not available, MC uses peer-to-peer wifi. If this is true, then we can assure our customers that if infrastructure wifi is available at the venue, then with all devices connected to it, range will be adequate. If infrastructure wifi is not available at the venue, perhaps a mobile wifi router (battery operated) could be set up, devices connected to it, then range would be adequate. We are about to test this. Reasonable? Can we be assured that if infrastructure wifi is available, MC uses it? Regarding #2, given we are targeting minimum watchOS 7.0, would the available networking APIs and frameworks be adequate to implement our own equivalent of the MC framework so our app on iOS/iPadOS and watchOS devices could communicate? How much work? Where would I start? I'm new to implementing networking but experienced in using the MC framework. I'm assuming that I would write the networking code to use infrastructure wifi to achieve acceptable range. Many thanks! Tim

In our system, when a user enables a mobile hotspot and the system connects to it, the system attempts to verify WIFI availability by sending an HTTP GET request to http://captive.apple.com. Normally, the server returns: HTTP Status: 200 (OK) Content-Type: text/html This has always been used as a sign of normal connectivity. Issue: Since last Friday, the server sometimes responds with: Content-Type: application/octet-stream When this occurs, our system determines that the network is unavailable and displays a connection warning (a “!” icon). Question: Has Apple recently made any backend or CDN configuration changes to captive.apple.com that could affect the response type? Any advice how can we solve this problem? Thanks!

I am trying to setup a system-wide DNS-over-TLS for iOS that can be turned off and on from within the app, and I'm struggling with the implementation details. I've searched online, searched forums here, used ChatGPT, and I'm getting conflicting information or code that is simply wrong. I can't find example code that is valid and gets me moving forward. I think I need to use NEDNSProxyProvider via the NetworkExtension. Does that sound correct? I have NetworkExtension -> DNS Proxy Capability set in both the main app and the DNSProxy extension. Also, I want to make sure this is even possible without an MDM. I see conflicting information, some saying this is opened up, but things like https://developer.apple.com/documentation/Technotes/tn3134-network-extension-provider-deployment saying a device needs to be managed. How do private DNS apps do this without MDM? From some responses in the forums it sounds like we need to parse the DNS requests that come in to the handleNewFlow function. Is there good sample code for this parsing? I saw some helpful information from Eskimo (for instance https://developer.apple.com/forums/thread/723831 ) and Matt Eaton ( https://developer.apple.com/forums/thread/665480 )but I'm still confused. So, if I have a DoT URL, is there good sample code somewhere for what startProxy, stopProxy, and handleNewFlow might look like? And valid code to call it from the main app?

I have been playing around with the new URL Filtering API. I have successfully installed and configured the sample code, Installed the example app to my iPhone, and am also running the PIR server locally on my Mac. In my input.txtpb file, I simply have 2 endpoints: rows: [{ keyword: "instagram.com", value: "1" }, { keyword: "youtube.com/shorts", value: "1" }] Neither of these are blocked when I attempt to load them from either a browser, or their dedicated apps. Are there any debugging tips I should know about? Additionally, I have also noticed a few times I have left the filter running on my phone, after leaving my LAN (where the PIR server is running), suddenly throughout the day I'm having random, completely unrelated endpoints blocked on my phone. I thought this API was never supposed to produce false positives (without calling back to the PIR server for confirmation).

Hello, we are developing hardware that needs to connect to an iPhone via Wi-Fi to send requests to a server. On Android, we have managed to create a programmatic local hotspot within the app to facilitate connection and improve the user experience. On iOS, however, Personal Hotspot must be manually enabled from the system settings, and the user must manually enter the SSID and password, which significantly degrades the UX. My questions are: Is there a workaround, unofficial method, or private API to generate a local hotspot from an app on iOS, similar to what can be done on Android? Is there an alternative within the MFi program or through specific frameworks to facilitate a quick and automatic connection between the hardware and the iPhone without relying on the manual Personal Hotspot? Are there any best practices for improving the local Wi-Fi connection experience between an accessory and an iPhone in the absence of hotspot controls? I would appreciate any guidance, experience, or resources that would help me better understand the feasible options in iOS for scenarios where fast and direct communication between hardware and mobile devices via Wi-Fi is required.

Hello Everyone, I’m working on a project that involves multicast communication between processes running on different devices within the same network. For all my Apple devices (macOS, iOS, etc.), I am using NWConnectionGroup, which listens on a multicast address "XX.XX.XX.XX" and a specific multicast port. The issue occurs when a requestor (such as a non-Apple process) sends a multicast request, and the server, which is a process running on an Apple device using NWConnectionGroup (the responder), attempts to reply. The problem is that the response is sent from a different ephemeral port rather than the port on which the multicast request was received. If the client is behind a firewall that blocks unsolicited traffic, the firewall only allows incoming packets on the same multicast port used for the initial request. Since the multicast response is sent from a different ephemeral port, the firewall blocks this response, preventing the requestor from receiving it. Questions: Is there a recommended approach within the NWConnectionGroup or Network.framework to ensure that responses to multicast requests are sent from the same port used for the request? Are there any best practices for handling multicast responses in scenarios where the requestor is behind a restrictive firewall? Any insights or suggestions on how to account for this behavior and ensure reliable multicast communication in such environments would be greatly appreciated. Thanks, Harshal