“translate scheme”

If an “agent” can’t: run a build/test for a specified scheme/target/destination, interpret failures, apply the needed project/build-setting changes, iterate until green, then it’s an in-editor assistant, not an agentic workflow.

Apple can fairly say: “Agents shouldn’t directly edit our serialization format.” But if they advertise agentic workflows and the agent can’t reliably do core build/debug tasks (target build settings, linking, schemes), then the product gap is real. Agentic workflows require a closed loop: apply build-setting changes → build → fix → repeat. If the only exposed interface is “chat + inline code edits,” it’s not agentic for build/debugging. Therefore Apple needs a supported automation surface (typed project actions with preview/undo), regardless of how ugly project.pbxproj is. Expected: Agent can apply target/scheme/build-setting changes (typed + preview/undo), run build, iterate on failures. Actual (Xcode 26.3 RC): Agent mostly suggests or edits source files; can’t reliably adjust the build/ debug configuration loop for multi-target/C++ integration. C/C++ integration is where an agent should shine because it’s repetitive, error-driven, and configuration-heavy. A “real” agentic C++ loop in Xcode

So splitting targets into different projects is a way to go, but do not make cross-project references and add target dependency. The idea is to build the subproject fully separately. Claude helped with implementation, so I asked it to write the rest of the post. The Solution Two scripts, triggered at different build stages: 1. Scheme Pre-Action — BuildPkgTestCMD.sh Builds the CLI project via a separate xcodebuild invocation before the main build starts. 2. Run Script Build Phase — CopyPkgTestCMD.sh Copies the built binary into the app bundle after the Resources phase. Both scripts check EFFECTIVE_PLATFORM_NAME and skip on non-Catalyst builds (e.g. iOS). Gotchas We Hit Build settings leaking into the nested xcodebuild call. Scheme pre-actions inherit all build settings from the parent target as environment variables. This means the nested xcodebuild silently picks up Catalyst platform, signing, and arch settings. Fix: wrap the call with env -i, passing through only selected variables: env -i

Hey, Dustin. I’ve published an app called ProxyPilot that works by running a local OpenAI-compatible proxy on 127.0.0.1 (line 4000) and translating Xcode Claude Agent’s Anthropic /v1/messages calls into OpenAI-style /v1/chat/completions for your upstream provider (for example GLM). That lets Xcode 26.3 agent features run against private/self-hosted or non-OpenAI backends while keeping traffic local to your machine first. You can pick the exact routed model in-app, and verify remaps in logs. You can download the .dmg from https://micah.chat/proxypilot.

Thanks for bringing that up! We are tracking the concern that using generated Swift symbols nudges developers to use a semantic key (instead of English, or any development language), and that this key is displayed at runtime when the current language doesn't have a value. This matches the Foundation's behavior for year, where strings without value display either the defaultValue, or the key. As a workaround for now, you could insert values from the source language by using the context menu Mark as Reviewed“ on the strings in state new. This will put the English values into your target language for untranslated strings. To not lose track of them, open the context menu again and select Mark for Review. Your translators should be able to notice this, and translate the strings when they get to them. I hope this workaround helps, and thanks for explaining your use case to us!

Hello @thisisjaymehta, Welcome to SwiftUI development! 🥳 The first thing I would look at is the Attributes menu, located in the Inspector area (the sidebar on the right of Xcode). You can also quickly open this tab by pressing ⌥ ⌘ 4. In that menu you will notice options for devices, appearances, sizes and more. In your screenshot, it appears you have the Scales set to Single Scale. You can try switching this to Individual Scale and add different scales which are configured to show on different devices. You can find more information about how to do that here Fitting images into available space If not, try appearances and see if it's a light/dark mode issue. It looks like you already did some troubleshooting by setting a solid white color background. A great tip you can use to check both color schemes easily is by using this toggle at the bottom of the canvas in Xcode, which will show both the light and dark move preview in Canvas. For more information on Xcodes canvas see Previewing your app’s interf

Using gen32IOVMSegments() will return kIOReturnMessageTooLarge. If I don't set the maximum value of 8M, the result will still be the same, so I think it may not be related to the maximum value. If you haven't already, take a look at my post here. I haven't checked the code to confirm, but I strongly suspect that calling gen32IOVMSegments with a mask of the wrong size (like 0xFFFFF000) will cause kIOReturnMessageTooLarge, as the true mask range you've specified (0x00000000FFFFF000) is too large to fit in 32 bits. Does Apple M silicon not allocate <=32-bit bus addresses to third-party PCIe devices (FPGA)? Quite the opposite. We've actually gone to considerable trouble to ensure that we COULD provide 32-bit addresses to the PCI bus. Quoting Address Translation on 64-Bit System Architectures: Apple solved the problem with address translation which “maps” blocks of memory into the 32-bit address space of a PCI device. In this scheme, the PCI device still sees a 4-gigabyte space, but t

Thanks for your post and interesting on your comment about: Long press on the Safari icon and choose Remove app If you make the system avoid using Universal Links by telling how to handle the link. Since you mentioned universal links, make sure there's no conflict with any custom URL schemes defined in your app's . Universal links should ideally not interfere with opening URLs in a browser unless specifically set up to redirect to an app. 
Since you observed different behavior on iOS 26.0+, double-check whether this is related to a beta version or a specific OS bug. Review Apple's release notes or beta forums for any related issues and consider filing a bug report if necessary. Ensure you're testing on various devices and simulators, as sometimes configuration or environment-specific issues can cause unexpected behavior. But I’m wondering if you are disabling the link itself, you can see that on the swctutils file search for the link itself inside that file. Can you publish the results? I would like

There is however a problem with this solution. If there is a TextField in the popover, typing in makes it firstResponder ; and the parent window controller do not receive mouseEnter events that triggers the tooltips. Which means I do not need to simulate mouseDown but just makeFirstResponder. So at the end, the question boils down to: how to manage the conflict between popover and its parent window about FirstResponder. Is it possible to have first and second responder, I mean forwarding the mouse messages to the parent ? Considering that The shared NSApplication object performs the important task of receiving events from the window server and distributing them to the proper NSResponder objects. NSApp translates an event into an NSEvent object, then forwards the event object to the affected NSWindow object. All keyboard and mouse events go directly to the NSWindow object associated with the event. … When a window object receives an NSEvent object from NSApp, it distributes it to the objects in its vi

Thanks for the post, was it working well in Xcode 26.2? I was just a few days ago configuring Claude as well to see how it works, there are a few experts in the team and I’ll refer to their expertise as I am still in the process of learning as well. My only suggestion when Xcode 26.3 RC isn’t automatically picking up your for Claude and MCP servers, it may be due to how the environment variables or model configurations are being handled. As I saw Claude running on the terminal, you can verify this by running in your terminal to ensure it returns the expected path. After updating environment variables or scheme settings, fully restart Xcode and any terminal sessions to ensure changes take effect. My thought is that could be permission? Ensure that Xcode and the terminal have read permissions for the and its containing directory. Looking forward to experts here to better guidance!! Awesome question that hopefully becomes a thread of suggestions. Albert Pascual
  Worldwide Developer Relations.

Hi Kevin, During the implementation of batch task buffer mapping, we observed an inconsistency between API return values and actual memory addresses, which leads to an immediate system Panic (Kernel Data Abort). Inside the implementation of UserMapBundledParallelTaskCommandAndResponseBuffers, we performed the following operations on the Response Buffer: We invoked CreateMapping(0, 0, 0, 0, 0, &ivars->fResponseMap) on the provided parallelResponseIOMemoryDescriptor. CreateMapping returns kIOReturnSuccess, and the resulting ivars->fResponseMap object pointer is non-null. However, a subsequent call to ivars->fResponseMap->GetAddress() returns NULL (0x0). After the DEXT returns from UserMapBundledParallelTaskCommandAndResponseBuffers with this NULL address, the system immediately triggers a Panic when the kernel attempts to process subsequent discovery commands (e.g., Inquiry). Panic Type: Kernel data abort Fault Address (FAR): 0x0000000000000000 Exception Class (ESR): 0x96000006 (Translation