Problem

When the WhatsApp health monitor detects a stale socket and restarts the provider, messages in flight are dropped. No retry or delivery confirmation occurs.

Timeline (from logs)

12:23:19 — User sends message
12:23:41 — Gateway detects stale socket, restarts WhatsApp provider
         → "[whatsapp:default] health-monitor: restarting (reason: stale-socket)"
12:23:43 — WhatsApp reconnects
         → "Listening for personal WhatsApp inbound messages"
Result:   Agent response generated but never delivered to WhatsApp

Expected Behavior

• Messages should be queued during reconnection and sent once the channel is healthy
• OR: Send a system message to the user indicating temporary unavailability

Actual Behavior

Silent message drop. User sees no response until they send a follow-up.

Environment

• OpenClaw version: 2026.3.13
• Channel: WhatsApp (personal)
• Model: anthropic/claude-sonnet-4-5
• Host: Gateway (local bind)

Reproduction

1. Wait for WhatsApp health monitor to detect stale socket (~5 min idle)
2. Send a message that triggers a tool call + response
3. Observe: response never reaches WhatsApp

Suggested Fix

Add message delivery retry logic with exponential backoff when channel state is reconnecting. Queue outbound messages until channel confirms readiness.