diff --git a/Scripts/staging-gate.sh b/Scripts/staging-gate.sh
index 12ec824..18c6018 100755
--- a/Scripts/staging-gate.sh
+++ b/Scripts/staging-gate.sh
@@ -180,3 +180,23 @@ if [ "$EDGE" != "ok" ]; then
   exit 1
 fi
 echo "DEPLOY OK — prod healthy at the edge; rollback available: docker tag flamenet-relay:previous flamenet-relay:prod && docker compose up -d --no-deps relay"
+
+# 🔴 **Carry the test fleet with production, or the live suites lie.**
+#
+# relay1/2/3 pin `flamenet-relay:prod`, but a container keeps running the image
+# it started with — so every deploy here silently leaves the fleet a build
+# behind. It was ELEVEN DAYS stale when that was first noticed, answering every
+# health check the whole time, while the Android live suite ran against it and
+# reported green about a build production had already moved off.
+#
+# ⚠️ Best-effort on purpose. The fleet is disposable and a production deploy
+# must not fail because a test relay would not come up — but it must also not
+# quietly stop matching, so the outcome is printed either way and
+# `messenger-privacy-check` asserts it independently.
+if [ -d /srv/test/relays ]; then
+  if (cd /srv/test/relays && docker compose up -d >/dev/null 2>&1); then
+    echo "FLEET OK — relay1/2/3 recreated on the image production now runs"
+  else
+    echo "FLEET WARNING — the test fleet did not update; live suites would test the OLD build"
+  fi
+fi