diff --git a/docs/roadmap.md b/docs/roadmap.md
index 6de3c81..3edb934 100644
--- a/docs/roadmap.md
+++ b/docs/roadmap.md
@@ -1050,6 +1050,51 @@ them name the condition that would change the answer.
 
 ## Done
 
+- [x] **[#441](https://github.com/Gerrrt/HomeLab/issues/441) Alerted on a sensor
+      that stops logging, and closed the Zeek half by deciding it elsewhere.**
+      2026-09-17. `SuricataLogsStopped` landed on 2026-09-12 in
+      `loki/rules/security.rules.yaml`: `absent_over_time({app="suricata"}[9h])`,
+      aggregate across both interfaces, `for: 0s` like the two absence rules it
+      copies. The window was read rather than picked — 22 days of the stream, in
+      which the longest silence across both interfaces was 75 minutes — which is
+      what the issue asked for, because the thirty minutes it opened with would
+      have paged about twice a day.
+
+      **Re-measured on 2026-09-17 over 27.7 days and it holds.** Worst aggregate
+      silence 80 minutes, six gaps past an hour, none past two, so nine hours
+      keeps nearly seven times the headroom `DhcpLeaseLogsStopped` settled on.
+      The aggregate design earned its keep in the interval: `igc0.10` alone went
+      quiet for more than nine hours **seven times**, worst 47.5 hours, so a
+      per-interface rule would have paged seven times in a fortnight. The
+      short-gap counts are not comparable between the two measurements — 135
+      over thirty minutes here against 54 on 2026-09-12, a difference in method,
+      not in Suricata — and the rule's comment says so rather than presenting
+      one series.
+
+      **The Zeek half was not built, and will not be as the issue described it.**
+      `ZeekLogsStopped` was to be the same shape against the lab's Loki once
+      [#437](https://github.com/Gerrrt/HomeLab/issues/437) existed. Three things
+      make that the wrong instrument. #437 puts Zeek's logs on `alexander` and
+      never on `10.0.99.20`, which ADR-0007 requires;
+      [ADR-0020](adr/0020-run-the-lab-stack-in-a-guest-with-its-own-prometheus.md)
+      gives that stack no Alertmanager on purpose, and `stacks/lab/loki/` ships
+      no ruler and no `rules/` because *"a ruler with nowhere to deliver
+      evaluates rules and discards the result, which reads as coverage and is
+      not"*; and
+      [ADR-0028](adr/0028-let-guest-liveness-cross-but-not-guest-telemetry.md)
+      revisited that exact boundary and kept it. #437 already carries the right
+      mechanism — a `homelab_zeek_mirror_active` textfile gauge read on the
+      hypervisor, crossing as guest state under ADR-0028 — so Zeek's liveness is
+      a metric question answered where Zeek is built, and this issue closes
+      rather than holding a slot behind a Wave 2 dependency that itself waits on
+      [#414](https://github.com/Gerrrt/HomeLab/issues/414).
+
+      What this proves and what it does not, unchanged from the merge: Loki
+      rules still have no unit-test harness, so `check_loki_rules.sh` shows the
+      rule parses and the ruler evaluates it, not that it fires. The live ruler
+      reports it `health=ok` and `state=inactive`, which is the good state and
+      not evidence of detection.
+
 - [x] **[#470](https://github.com/Gerrrt/HomeLab/issues/470) The wiki's drift
       check is watched.** 2026-09-14. `Gerrrt/Lemmiwinks/.claude/tools/drift-check`
       reads the wiki's machine-checkable claims against the machine and files a
diff --git a/docs/security.md b/docs/security.md
index 9709cb7..b1d7a90 100644
--- a/docs/security.md
+++ b/docs/security.md
@@ -101,8 +101,26 @@ process is alive, not that it is inspecting anything. `SuricataLogsStopped` in
 once both interfaces have been silent for nine hours, a window read from 22 days
 of the stream whose longest silence was 75 minutes, and it cannot see one
 interface going quiet on its own — the guest segment is silent for days at a
-time ([#441](https://github.com/Gerrrt/HomeLab/issues/441)). The runbook's test
-alert is still the only proof that it detects.
+time ([#441](https://github.com/Gerrrt/HomeLab/issues/441)). Re-measured on
+2026-09-17 over 27.7 days, the worst aggregate silence was 80 minutes and the
+guest segment alone crossed nine hours seven times, so the window holds and the
+decision to aggregate is what keeps the rule quiet. The runbook's test alert is
+still the only proof that it detects.
+
+**A limit none of the three names, and no rule here closes: Suricata is not a
+protocol logger, and the lab is out of reach.** SNI, JA3 and certificate metadata — the
+ground the plaintext limit gives up — are Zeek's, and east-west traffic between
+the lab's domain guests crosses no router, so `morpheus` never sees a packet of
+it ([#437](https://github.com/Gerrrt/HomeLab/issues/437),
+[ADR-0006](adr/0006-detect-at-the-chokepoint.md)). When that sensor is built its
+logs stay on `alexander` and never reach `10.0.99.20`, which ADR-0007 requires
+and [ADR-0020](adr/0020-run-the-lab-stack-in-a-guest-with-its-own-prometheus.md)
+enforces by giving the lab no Alertmanager for a ruler to deliver to. So the
+absence rule this section describes has no Zeek equivalent and will not get one:
+whether that sensor is still running is answered on the hypervisor, as guest
+state crossing under
+[ADR-0028](adr/0028-let-guest-liveness-cross-but-not-guest-telemetry.md), by the
+`homelab_zeek_mirror_active` gauge #437 builds alongside the mirror.
 
 **Device joins are detected as of 2026-09-04**, from the DHCP server rather
 than from the wireless. `morpheus` ships Kea's lease log to Loki, and the first
diff --git a/stacks/observability/loki/rules/security.rules.yaml b/stacks/observability/loki/rules/security.rules.yaml
index 4923756..8231803 100644
--- a/stacks/observability/loki/rules/security.rules.yaml
+++ b/stacks/observability/loki/rules/security.rules.yaml
@@ -616,6 +616,23 @@ groups:
   # SuricataLogsStopped fires only when BOTH interfaces have been silent, and a
   # Degens-only silence is invisible to it by design.
   #
+  # RE-VERIFIED 2026-09-17, five weeks on and over a retention window that has
+  # rolled forward: 27.7 days ending 2026-09-17 22:25 UTC, same five-minute
+  # buckets. The worst aggregate silence is 80 minutes — five minutes longer
+  # than the 2026-09-12 reading — six gaps exceeded an hour, none exceeded two,
+  # and nine hours is still nearly seven times the worst gap. The case for
+  # aggregating got stronger rather than weaker: igc0.10 alone crossed nine
+  # hours SEVEN times, worst 47.5 hours, so a per-interface rule would have
+  # paged seven times in a fortnight.
+  #
+  # The short-gap counts are NOT comparable across the two measurements — this
+  # one counts 135 gaps over thirty minutes where 2026-09-12 counted 54, which
+  # is a difference in method (bucket alignment, and which streams were summed)
+  # rather than in Suricata. The long tail, which is the half that sets the
+  # window, agrees. Note also that `interface` is only stamped from 2026-09-02,
+  # when syslog.alloy began mapping facility local1/local2, so the per-interface
+  # figures cover 13.7 days against the aggregate's 27.7.
+  #
   # Two rules, two questions. SuricataStopped in prometheus/rules/ids.rules.yaml
   # reads the firewall's process table over SNMP (#90) — bsnmpd serves
   # HOST-RESOURCES-MIB hrSWRunTable, the pfsense module fetches the rows named