Problem

When the Feishu WebSocket connection suffers a keepalive ping timeout, the SDK's message loop exits, but the main Hermes agent process does not terminate or reconnect. This leaves the Gateway in a zombie state where it appears "running" to systemd but accepts no messages.

Log output:

[Lark] [ERROR] receive message loop exit, err: sent 1011 (internal error) keepalive ping timeout; no close frame received
[Lark] [WARNING] ping failed, err: sent 1011 (internal error) keepalive ping timeout

Expected behavior: Under a crash-only architecture, the feishu.py integration thread should raise a SystemExit(1) so systemd-level Restart=always can respawn a healthy stack.

Reference

Upstream: NousResearch#10616

Design (from OpenClaw)

OpenClaw's monitor.ts has a complete lifecycle management system:

• monitorSingleAccount() with abort signal support
• Health check probe via fetchBotIdentityForMonitor()
• State management in monitor.state.ts
• Webhook anomaly tracking

Implementation Plan

1. Add a watchdog thread/task in FeishuAdapter.connect() that monitors the SDK's message loop
2. Detect ping timeout conditions and raise SystemExit(1) to trigger systemd restart
3. Add botIdentity pre-fetch to validate connection health on startup
4. Reference: gateway/platforms/feishu.py line ~1042 FeishuAdapter class