diff --git a/AGENTS.md b/AGENTS.md
index ef6cb86f1..7bdd8dfe8 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -238,7 +238,7 @@ This repo encodes invariants as self-declaring gates. The correct response to a
 
 ## PR Readiness Checklist
 - Static gates first: required checks from **Testing Matrix** pass, `pnpm check:fallow --base origin/main` is clean when code quality/dead-code risk is relevant, CI guards are green, and no conflict markers or unmerged paths remain.
-- Do not report a PR as CI-green from a local unit-only run alone: use `pnpm test:unit` for the repo unit bundle, or `vitest run --project unit-core --project android-adb` when invoking Vitest directly. The **Integration Tests** and **Coverage** jobs run the `provider-integration` project, so verify green on the actual PR head across those jobs, not just unit.
+- Do not report a PR as CI-green from a local unit-only run alone: use `pnpm test:unit` for the repo unit bundle, or `vitest run --project unit-core --project subprocess-stub` when invoking Vitest directly. The **Integration Tests** and **Coverage** jobs run the `provider-integration` project, so verify green on the actual PR head across those jobs, not just unit.
 - Command-surface changes preserve CLI, Node.js, daemon, MCP, help, docs, and SkillGym coverage where that surface is affected. Do not duplicate command contracts across layers.
 - Device-facing behavior is not merge-ready until it has real simulator/emulator/device evidence for the changed path. Fixture-backed tests can prove contracts, but they do not replace a live run that creates or observes the artifact/state the feature claims to handle.
 - If live verification is blocked, state the blocker, exact command or device needed, and downgrade the PR to residual risk instead of calling it ready.
diff --git a/package.json b/package.json
index 6c1e09bc4..2a894e10e 100644
--- a/package.json
+++ b/package.json
@@ -140,8 +140,8 @@
     "test-app:maestro": "node scripts/run-test-app-maestro-suite.mjs",
     "test-app:maestro:ios": "node scripts/run-test-app-maestro-suite.mjs --platform ios",
     "test-app:maestro:android": "node scripts/run-test-app-maestro-suite.mjs --platform android",
-    "test": "vitest run --project unit-core --project android-adb",
-    "test:unit": "vitest run --project unit-core --project android-adb",
+    "test": "vitest run --project unit-core --project subprocess-stub",
+    "test:unit": "vitest run --project unit-core --project subprocess-stub",
     "test:coverage": "vitest run --coverage",
     "test:integration:provider": "vitest run --project provider-integration",
     "test:integration:progress": "node --experimental-strip-types scripts/integration-progress.ts",
diff --git a/vitest.config.ts b/vitest.config.ts
index b3eab91b7..d2fd6638c 100644
--- a/vitest.config.ts
+++ b/vitest.config.ts
@@ -1,8 +1,20 @@
 import { defineConfig } from 'vitest/config';
 import slowTestGateReporter from './scripts/vitest-slow-test-reporter.ts';
 
-const ANDROID_ADB_STUB_TESTS =
-  'src/platforms/android/__tests__/{app-lifecycle-install,app-lifecycle-open,device-input-state,input-actions,notifications,settings}.test.ts';
+// Tests that stub a real binary (adb/xcrun) by mutating process.env.PATH and
+// then spawn it, so each case waits real subprocess/retry/poll time. Run at the
+// unit suite's default ~7x file parallelism they contend for CPU and their stub
+// spawns get starved past an internal budget, so production takes a generic
+// failure path and returns a different error than the assertion expects — a
+// contention flake whose failing subset shifts between runs (see
+// docs/agents/testing.md "tests must not wait real time"). Serialized below with
+// per-file isolation so only one such file spawns stubs at a time, the same
+// execution contract the pre-split android index.test.ts aggregation provided.
+const SUBPROCESS_STUB_TESTS = [
+  'src/platforms/android/__tests__/{app-lifecycle-install,app-lifecycle-open,device-input-state,input-actions,notifications,settings}.test.ts',
+  'src/daemon/__tests__/runtime-hints.test.ts',
+  'src/platforms/apple/core/__tests__/index.test.ts',
+];
 
 export default defineConfig({
   test: {
@@ -27,19 +39,19 @@ export default defineConfig({
             // The Maestro conformance oracle runs via `node --test` in its own CI
             // job (scripts/maestro-conformance), like the layering guard.
           ],
-          exclude: [ANDROID_ADB_STUB_TESTS],
+          exclude: SUBPROCESS_STUB_TESTS,
           setupFiles: ['src/__tests__/process-memo-setup.ts'],
         },
       },
       {
-        // The scripted-adb tests stub the adb binary by mutating process.env
-        // (PATH, AGENT_DEVICE_TEST_ARGS_FILE) and wait real retry/poll time,
-        // so the group runs serialized with per-file isolation — the same
+        // The subprocess-stub tests stub adb/xcrun by mutating process.env
+        // (PATH, AGENT_DEVICE_TEST_ARGS_FILE) and wait real subprocess/retry/poll
+        // time, so the group runs serialized with per-file isolation — the same
         // execution contract the pre-split android index.test.ts aggregation
         // provided without leaking module caches between split files.
         test: {
-          name: 'android-adb',
-          include: [ANDROID_ADB_STUB_TESTS],
+          name: 'subprocess-stub',
+          include: SUBPROCESS_STUB_TESTS,
           setupFiles: ['src/__tests__/process-memo-setup.ts'],
           fileParallelism: false,
           isolate: true,