Five channels, three severity levels, deduplication, and escalation routing. Configurable per interface.
Alert Channels
| Channel | Watch | Insight | Command |
|---|
| Console output | Yes | Yes | Yes |
| File log | Yes | Yes | Yes |
| Webhook (Slack, Teams, etc.) | Yes | Yes | Yes |
| Email (SMTP) | — | Yes | Yes |
| PagerDuty | — | Yes | Yes |
Use the webhook channel to integrate with Slack, Microsoft Teams, or any service that accepts incoming webhooks.
Advanced Alert Features
- Deduplication — Suppresses repeat alerts for the same issue within a configurable window
- Escalation — Routes alerts based on severity to different channels or on-call groups
- Detailed payloads — Each alert includes file path, severity, error/warning counts, message type, sending application, and full error list
Severity Levels
| Level | Description |
|---|
critical | Interface down or failure rate exceeds threshold |
warning | Elevated error rate or anomalous throughput |
info | Informational events (interface started, config changed) |
Alert Lifecycle
Alerts follow a lifecycle: triggered → acknowledged → resolved.
# List recent alerts
pidgeon loft alerts --since 6h
# List critical alerts only
pidgeon loft alerts --severity critical
# Acknowledge an alert via API
curl -X POST https://api.pidgeon.health/api/loft/alerts/{id}/acknowledge
Example Alert
CRITICAL: Validation failure rate exceeded threshold
Interface: ER Interface
Message: 12.3% failure rate in the last 15 minutes (threshold: 5%)
Triggered: 2026-02-22T14:15:00Z
Failed messages: 47
Top errors: PID.5 empty (23), PV1.2 invalid (18), DG1 missing (6)
Acknowledging an alert records that a human reviewed it but does not resolve the underlying issue. Use
AI Triage for recommended fixes.
