Escalation Alert

When a critical issue needs escalation:

1. Assess severity — Determine if this is genuinely P0/P1:
   • P0: service outage, data loss, security breach
   • P1: major feature broken, many users affected
2. Compose alert — Format a clear incident message:

   🔴 **[P0/P1] — [Brief Title]**
   **Impact:** [What's affected, how many users]
   **Started:** [When it began]
   **Status:** Investigating
   **Details:** [1-2 sentence description]
   

3. Send — Post via send_message to the configured alert channel
4. Log — Record the escalation in memory with timestamp
5. Follow up — On next heartbeat, check if the issue is resolved and post an update

Only escalate genuinely critical issues. For P2/P3, batch into daily digest instead.