feat(health): recent XMPP conversation log panel + stale error fix

User requirement: health tab should show recent XMPP conversations.
- bot hooks log_xmpp on inbound (on_msg) and outbound (_deliver_loop)
  so real chats land in xmpp_messages.jsonl (was: only cron/scanner)
- index.html health tab: new '最近对话' panel (last 10 msgs, dir arrow,
  preview, status, time); refreshHealth updates it incrementally
- last_error now shows age and resolved state: once a successful
  outbound happens after an error, it's shown gray as '已恢复'
  instead of alarming red forever; unresolved errors still red
- health() status no longer degraded by errors that were later
  resolved by successful outbound

deploy/bot/xmpp_agent_core.py

@@ -147,6 +147,14 @@ _inbound_lock = threading.Lock()
 
 RECENT_SENT_MAX = 50
 
+# ── XMPP 消息日志（dashboard 健康Tab「最近对话」数据源）──
+try:
+    sys.path.insert(0, os.path.join(os.path.dirname(os.path.abspath(__file__)), '..', '..'))
+    from xmpp_logger import log_xmpp as _log_xmpp
+except Exception:
+    def _log_xmpp(*a, **kw):
+        pass
+
 
 def _rs(p):
     """Parse agent arg from sys.argv, returns agent name string."""
@@ -283,6 +291,7 @@ class XmppAgent(slixmpp.ClientXMPP):
                 log.info(f"📩 收到图片(URL body): from={msg['from']} url={body[:80]}")
                 body = f"[IMAGE] {body}"
             log.info(f"📩 收到: from={msg['from']} type={msg['type']} body={body[:60]}")
+            _log_xmpp('in', sender_str := str(msg['from']), f"{AGENT_NAME}@yoin.fun", body)
             if ('[executor]' in body and 'gateway_zhiwei' in body):
                 log.info(f"过滤 executor 消息: {body[:80]}...")
                 return
@@ -322,8 +331,10 @@ def _deliver_loop(bot):
                         bot.send_message(mto=to, mbody=body, mtype=mtype)
                     asyncio.run(_send(target, text, msg_type))
                     bot.mark_sent(text)
+                    _log_xmpp('out', f"{AGENT_NAME}@yoin.fun", target, text)
                     log.info(f"  已发送到 {target}: {text[:80]}")
                 except Exception as e:
+                    _log_xmpp('out', f"{AGENT_NAME}@yoin.fun", target, text, status='error', error=str(e)[:150])
                     log.error(f"  发送到 {target} 失败: {e}")
             time.sleep(0.3)
         except Exception as e:

scripts/verify_health_fix.py

@@ -0,0 +1,14 @@
+import urllib.request, json
+d = json.loads(urllib.request.urlopen('http://127.0.0.1:8899/api/xmpp/health', timeout=120).read())
+ba = d.get('bot_activity', {})
+print('status:', d.get('status'))
+print('last_error:', (ba.get('last_error') or '')[:80])
+print('last_error_age_sec:', ba.get('last_error_age_sec'))
+print('last_error_resolved:', ba.get('last_error_resolved'))
+print('last_outbound_age_sec:', ba.get('last_outbound_age_sec'))
+print('---')
+m = json.loads(urllib.request.urlopen('http://127.0.0.1:8899/api/xmpp/messages', timeout=30).read())
+msgs = m.get('messages', [])
+print('messages count:', len(msgs))
+for x in msgs[:5]:
+    print(' ', x.get('timestamp'), x.get('direction'), (x.get('body_preview') or '')[:50])

static/index.html

@@ -1462,6 +1462,9 @@ async function renderHealth() {
     // XMPP 通道健康（独立 fetch，不阻塞主流程）
     let xmpp = null;
     try { xmpp = await fetchJSON('/api/xmpp/health'); } catch(e) {}
+    // 最近 XMPP 对话日志（bot 挂钩 log_xmpp 后的真实聊天记录）
+    let xmppMsgs = [];
+    try { const md = await fetchJSON('/api/xmpp/messages'); xmppMsgs = md.messages || []; } catch(e) {}
 
     const svcs = services.services || [];
     const summary = services.summary || { ok: 0, total: 0 };
@@ -1534,7 +1537,12 @@ async function renderHealth() {
       html += '<div><span class="text-slate-500">⚠️ 错误</span><br><span class="font-mono ' + (ba.errors ? 'text-[#f85149]' : 'text-slate-500') + '" id="hBot_err">' + (ba.errors || 0) + '</span></div>';
       html += '</div>';
       if (ba.last_error) {
-        html += '<div class="text-xs text-[#f85149] mt-2 p-2 bg-red-900/20 rounded" id="hBot_lastErr">' + ba.last_error + '</div>';
+        const errAge = ba.last_error_age_sec >= 0 ? Math.round(ba.last_error_age_sec/60) + '分钟前' : '';
+        if (ba.last_error_resolved) {
+          html += '<div class="text-xs text-slate-500 mt-2 p-2 bg-slate-800/30 rounded" id="hBot_lastErr">⚠️ 历史错误（' + errAge + '，已恢复）: ' + ba.last_error + '</div>';
+        } else {
+          html += '<div class="text-xs text-[#f85149] mt-2 p-2 bg-red-900/20 rounded" id="hBot_lastErr">⚠️ ' + errAge + ': ' + ba.last_error + '</div>';
+        }
       }
 
       // Hermes Gateway + LLM Provider
@@ -1549,6 +1557,27 @@ async function renderHealth() {
       html += '<div><span class="text-slate-500">🧠 LLM Provider</span><br><span class="font-mono ' + (llm.status === 'ok' ? 'text-[#3fb950]' : 'text-[#f85149]') + '" id="hLlmStatus">' + (llm.status || '?') + '</span>';
       if (llm.error) html += '<br><span class="text-[#f85149]" id="hLlmErr">' + llm.error + '</span>';
       html += '</div></div>';
+
+      // ── 最近 XMPP 对话日志 ──
+      html += '<div class="mt-3 pt-3 border-t border-slate-800/50">';
+      html += '<div class="text-xs text-slate-500 mb-2">💬 最近对话（' + xmppMsgs.length + ' 条）</div>';
+      if (xmppMsgs.length === 0) {
+        html += '<div class="text-xs text-slate-600 p-2">暂无记录 — bot 重启后对话开始采集</div>';
+      } else {
+        html += '<div class="space-y-1 max-h-64 overflow-y-auto" id="hXmppMsgs">';
+        xmppMsgs.slice(0, 10).forEach(m => {
+          const isIn = m.direction === 'in';
+          const arrow = isIn ? '📩' : '📤';
+          const dirColor = isIn ? 'text-[#58a6ff]' : 'text-[#3fb950]';
+          const statusMark = m.status === 'ok' ? '' : ' <span class="text-[#f85149]">[' + m.status + ']</span>';
+          const who = isIn ? (m.from || '').split('@')[0] : '知微';
+          const t = (m.timestamp || '').slice(11, 16);
+          const preview = (m.body_preview || '').replace(/</g, '&lt;').slice(0, 80);
+          html += '<div class="text-xs p-1.5 rounded bg-slate-800/40 flex gap-2"><span class="' + dirColor + ' shrink-0">' + arrow + ' ' + who + '</span><span class="text-slate-400 flex-1 break-all">' + preview + statusMark + '</span><span class="text-slate-600 shrink-0 font-mono">' + t + '</span></div>';
+        });
+        html += '</div>';
+      }
+      html += '</div>';
       html += '</div>';
     }
     el.innerHTML = html;
@@ -1563,9 +1592,10 @@ async function refreshHealth() {
   const el = document.getElementById('tab-health');
   if (!el || el.classList.contains('hidden') || !el.dataset.rendered) return;
   try {
-    const [services, xmpp] = await Promise.all([
+    const [services, xmpp, msgData] = await Promise.all([
       fetchJSON('/api/services'),
-      fetchJSON('/api/xmpp/health')
+      fetchJSON('/api/xmpp/health'),
+      fetchJSON('/api/xmpp/messages').catch(() => null)
     ]);
     const svcs = (services.services || []);
     const sum = services.summary || {};
@@ -1613,12 +1643,40 @@ async function refreshHealth() {
       if (bi) bi.textContent = ba.inbound || 0;
       if (bo) { bo.textContent = ba.outbound || 0; bo.className = 'font-mono ' + (ba.outbound ? 'text-[#3fb950]' : (ba.inbound ? 'text-[#f85149]' : 'text-slate-500')); }
       if (be) { be.textContent = ba.errors || 0; be.className = 'font-mono ' + (ba.errors ? 'text-[#f85149]' : 'text-slate-500'); }
-      if (bl) bl.textContent = ba.last_error || '';
+      if (bl) {
+        const errAge = ba.last_error_age_sec >= 0 ? Math.round(ba.last_error_age_sec/60) + '分钟前' : '';
+        if (ba.last_error && ba.last_error_resolved) {
+          bl.textContent = '⚠️ 历史错误（' + errAge + '，已恢复）: ' + ba.last_error;
+          bl.className = 'text-xs text-slate-500 mt-2 p-2 bg-slate-800/30 rounded';
+        } else if (ba.last_error) {
+          bl.textContent = '⚠️ ' + errAge + ': ' + ba.last_error;
+          bl.className = 'text-xs text-[#f85149] mt-2 p-2 bg-red-900/20 rounded';
+        } else {
+          bl.textContent = '';
+        }
+      }
       // LLM Provider
       const llm = xmpp.llm_provider || {};
       const ls = document.getElementById('hLlmStatus'), le = document.getElementById('hLlmErr');
       if (ls) { ls.textContent = llm.status || '?'; ls.className = 'font-mono ' + (llm.status === 'ok' ? 'text-[#3fb950]' : 'text-[#f85149]'); }
       if (le) le.textContent = llm.error || '';
+
+      // 最近对话列表
+      const msgBox = document.getElementById('hXmppMsgs');
+      if (msgBox && msgData && msgData.messages) {
+        let mh = '';
+        msgData.messages.slice(0, 10).forEach(m => {
+          const isIn = m.direction === 'in';
+          const arrow = isIn ? '📩' : '📤';
+          const dirColor = isIn ? 'text-[#58a6ff]' : 'text-[#3fb950]';
+          const statusMark = m.status === 'ok' ? '' : ' <span class="text-[#f85149]">[' + m.status + ']</span>';
+          const who = isIn ? (m.from || '').split('@')[0] : '知微';
+          const t = (m.timestamp || '').slice(11, 16);
+          const preview = (m.body_preview || '').replace(/</g, '&lt;').slice(0, 80);
+          mh += '<div class="text-xs p-1.5 rounded bg-slate-800/40 flex gap-2"><span class="' + dirColor + ' shrink-0">' + arrow + ' ' + who + '</span><span class="text-slate-400 flex-1 break-all">' + preview + statusMark + '</span><span class="text-slate-600 shrink-0 font-mono">' + t + '</span></div>';
+        });
+        msgBox.innerHTML = mh;
+      }
     }
 
     // 更新时间戳

xmpp_logger.py

@@ -306,6 +306,7 @@ def health():
 
         last_inbound_ts = max((_line_epoch(l) for l in inbound), default=0)
         last_outbound_ts = max((_line_epoch(l) for l in outbound), default=0)
+        last_error_ts = max((_line_epoch(l) for l in errors), default=0)
         result["bot_activity"] = {
             "inbound": len(inbound), "outbound": len(outbound), "errors": len(errors),
             "last_error": errors[-1][:250] if errors else None,
@@ -313,9 +314,14 @@ def health():
             "last_outbound": outbound[-1][:200] if outbound else None,
             "last_inbound_age_sec": int(now_epoch - last_inbound_ts) if last_inbound_ts else -1,
             "last_outbound_age_sec": int(now_epoch - last_outbound_ts) if last_outbound_ts else -1,
+            "last_error_age_sec": int(now_epoch - last_error_ts) if last_error_ts else -1,
+            # 错误之后已有成功出站 → 该错误已被覆盖，不算当前问题
+            "last_error_resolved": bool(last_error_ts and last_outbound_ts > last_error_ts),
         }
-        if errors and not outbound: result["status"] = "degraded"
+        # 只有未覆盖的错误才降级状态
-        if inbound and not outbound: result["status"] = "degraded"
+        unresolved_error = last_error_ts and not (last_outbound_ts > last_error_ts)
+        if errors and not outbound and unresolved_error: result["status"] = "degraded"
+        if inbound and not outbound and unresolved_error: result["status"] = "degraded"
     except Exception:
         result["bot_activity"] = {"error": "journalctl failed"}