Prune stale container strike counts, not just skip-list entries

The 2026-07-19 prune covered the skip list and intentional stops but not the
per-container strike counts, which have the same failure: set_strikes 0 only
fires when a container is seen running again, so an uninstalled one keeps its
count forever and every consumer reads the host as unhealthy. Reserved daemon_*
keys are excluded.

Watchdogs/docker_watchdog.sh

@@ -137,6 +137,17 @@
 #   All docker commands wrapped in timeout. Daemon hangs cannot stall the
 #   watchdog and leave containers unmonitored between cycles.
 #
+# Stale Entry Pruning
+#   Every cycle, entries naming a container that no longer exists are removed from
+#   DOCKER_WATCHDOG_FAILED_FILE, DOCKER_WATCHDOG_INTENTIONAL_FILE, and the per-container
+#   strike counts in WATCHDOG_STATE_FILE. None of these can clear themselves: every
+#   clearing path requires seeing the container running again, which never happens once
+#   it is uninstalled. Without pruning, one removed container makes this host report
+#   unhealthy permanently. Reserved daemon_* keys in WATCHDOG_STATE_FILE are never pruned.
+#   Pruning runs under --dry-run as well: the documented contract for that flag is that
+#   nothing gets restarted, and dropping a record of a container that no longer exists is
+#   housekeeping, not an action. It is idempotent — repeated runs converge.
+#
 # Notification Batching
 #   Events collected across a full cycle and sent as a single summary.
 #   Prevents notification floods when a shared dependency failure cascades.
@@ -710,6 +721,28 @@ CYCLE_START=$(date +%s)
     done
     unset _prune_file _prune_snapshot _prune_container
 
+    # Same failure, different file. WATCHDOG_STATE_FILE cannot join the loop above: it stores
+    # key:value rather than one name per line, and it holds reserved daemon_* keys alongside the
+    # per-container strike counts. A strike entry for an uninstalled container never clears on its
+    # own — set_strikes "$container" 0 only fires when the container is seen running again, which
+    # by definition never happens — so the count sits there and every consumer that treats a
+    # non-zero strike as unhealthy reports this host unhealthy forever. Confirmed live 2026-08-02:
+    # claudeclaw_docker at 2119 strikes long after removal, which is what kept the monitor page's
+    # watchdog summary from ever going green.
+    if [[ -s "$WATCHDOG_STATE_FILE" ]]; then
+        mapfile -t _prune_snapshot < "$WATCHDOG_STATE_FILE"
+        for _prune_line in "${_prune_snapshot[@]}"; do
+            _prune_container="${_prune_line%%:*}"
+            [[ -z "$_prune_container" ]] && continue
+            [[ "$_prune_container" == daemon_* ]] && continue   # reserved keys, not containers
+            if ! timeout "$DOCKER_TIMEOUT" docker inspect "$_prune_container" &>/dev/null; then
+                sed -i "/^${_prune_container}:/d" "$WATCHDOG_STATE_FILE" 2>/dev/null
+                warn "$_prune_container no longer exists — removed from $(basename "$WATCHDOG_STATE_FILE")"
+            fi
+        done
+        unset _prune_snapshot _prune_line _prune_container
+    fi
+
     _skip_contents=$(cat "$DOCKER_WATCHDOG_FAILED_FILE" 2>/dev/null | tr '\n' ' ' | xargs)
     [[ -n "$_skip_contents" ]] && warn "$ICON_SKIP Skip list active: $_skip_contents — manual intervention needed"
     _intentional_contents=$(cat "$DOCKER_WATCHDOG_INTENTIONAL_FILE" 2>/dev/null | tr '\n' ' ' | xargs)