Confirmed on macOS 26.5.1 (25F80) - still present in the .1 update, not just 25F71. Your snippet returns 512 bytes on a machine that's currently hitting the hang: pipe capacity: 512 bytes Same signature you describe: xcodebuild and the IDE both stall at "Planning" before any compilation, with the clang feature-detection probe (clang -v -E -dM /dev/null) blocked in write() against the shrunken pipe. Two data points that line up with the kernel-leak theory rather than a stuck helper: logout/login restores capacity too, not only a full reboot - it returns to ~64 KB and then degrades again within about one build cycle. Userspace remedies do not help: pkill of the build helpers (SWBBuildService, sourcekit-lsp, SourceKitService, swift-plugin-server), and wiping ModuleCache / DerivedData / SPM caches. The hang also reproduces with the IDE fully closed via plain xcodebuild, so it isn't a hung GUI helper. So from here it looks macOS-version-bound (26.5.x) rather than Xcode-bound, consistent with your finding that an Xcode downgrade doesn't move it.