diff --git a/.github/workflows/ios.yml b/.github/workflows/ios.yml
index b8a544293..6b079c2f4 100644
--- a/.github/workflows/ios.yml
+++ b/.github/workflows/ios.yml
@@ -1,8 +1,10 @@
 name: iOS
 
 on:
+  # No push trigger for t3/ios-mvf: PR #10 (t3/ios-mvf -> main) already tests every merge into
+  # it, and carries full-ui so the audit runs there. A push run would take a second macOS slot.
   push:
-    branches: [main, t3/ios-mvf]
+    branches: [main]
     paths:
       - "ios/**"
       - "!ios/**/*.md"
@@ -65,9 +67,9 @@ jobs:
           xcodebuild -version
           swift --version
 
-      # Booting takes 2 to 3 minutes and needs nothing from the build, so it runs alongside the
-      # steps below; the UI test step waits for it.
-      - name: Boot the Simulator in the background
+      # Booting in the background was tried and dropped: on the runner's 3 cores it slowed
+      # swift test and the builds by about 6 minutes to save 2:40. xcodebuild boots it at test time.
+      - name: Find the Simulator
         run: |
           runtime="com.apple.CoreSimulator.SimRuntime.iOS-${SIM_OS//./-}"
           udid="$(xcrun simctl list devices available -j \
@@ -79,7 +81,6 @@ jobs:
             exit 1
           fi
           echo "SIM_UDID=$udid" >> "$GITHUB_ENV"
-          nohup xcrun simctl bootstatus "$udid" -b > "$RUNNER_TEMP/simulator-boot.log" 2>&1 &
 
       # The package's drift test compares the vendored server schemas with the server's own copies:
       # the checkout's files once that branch merges, else the t3/server branch, which the shallow
@@ -141,12 +142,6 @@ jobs:
           -destination "generic/platform=iOS" CODE_SIGNING_ALLOWED=NO COMPILER_INDEX_STORE_ENABLE=NO
           HOUSESCAN_GIT_COMMIT="${GITHUB_SHA::12}" build
 
-      - name: Wait for the Simulator
-        timeout-minutes: 10
-        run: |
-          xcrun simctl bootstatus "$SIM_UDID" -b
-          echo "Booted $SIM_NAME on iOS $SIM_OS ($SIM_UDID)."
-
       - name: UI test the whole flow in the Simulator
         # Replays the committed synthetic wall (ios/HouseScanUITests/Fixtures/synthetic-wall) with
         # -autopilot and the bundled sample result.
diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
index d8d9706e0..96183e649 100644
--- a/CONTRIBUTING.md
+++ b/CONTRIBUTING.md
@@ -33,7 +33,7 @@ The `size:*` label describes the effective diff. It is a review signal, not a me
 | Label PR size | a pull request opens or updates | none |
 | TestFlight | someone runs it from the Actions tab | none |
 
-The iOS UI tests skip the every-state accessibility audit on pull requests, because it adds about 10 minutes and macOS runners are scarce. Add the `full-ui` label when a pull request changes screens or copy; pushes to `t3/ios-mvf` and `main` always run it.
+The iOS UI tests skip the every-state accessibility audit on pull requests, because it adds about 10 minutes and macOS runners are scarce. Add the `full-ui` label when a pull request changes screens or copy; pushes to `main` always run it, and PR #10 (`t3/ios-mvf` into `main`) carries the label, so every merge into `t3/ios-mvf` runs it once.
 
 `make check` runs the three local suites. They need uv, Node 24 with pnpm, and Xcode 26 or newer; each directory's README has details. No check is required by branch rules yet. Don't call one required until the rules require its status.