fix: health_check now queries /health endpoint before log analysis

If xmpp_connected=false, force restart immediately instead of
waiting 10 minutes for message timeout. This catches the most
common failure mode: process alive but XMPP dead.

gateway/scripts/health_check_xxm.py

@@ -112,6 +112,25 @@ def main():
         wlog("=== end ===")
         return
 
+    # ── 0. Quick HTTP health check ──────────────────────────
+    # Check /health endpoint: if xmpp_connected=false, restart immediately.
+    # This catches "process alive but XMPP dead" — the most common failure mode.
+    try:
+        import urllib.request, json
+        req = urllib.request.Request("http://127.0.0.1:5802/health")
+        resp = json.loads(urllib.request.urlopen(req, timeout=5).read())
+        if not resp.get("xmpp_connected", True):
+            wlog("CRIT: xmpp_connected=false via /health — forcing restart")
+            kill_all_bots()
+            time.sleep(3)
+            start_bot()
+            wlog("=== end (restarted: XMPP disconnected) ===")
+            return
+    except Exception as e:
+        wlog(f"WARN: /health check failed: {e}")
+        # If /health is unreachable, the bot process might be dead or stuck
+        pass
+
     # Activity analysis
     recent = parse_log_tail(BOT_LOG, 100)
     alive = sum(1 for l in recent if "alive" in l)