Alerts

Amlexia notifies you when latency, errors, cost, or smart conditions cross thresholds you define in the dashboard.

Create an alert

1. Open app.amlexia.com → your project → Alerts
2. Create alert
3. Choose condition type and threshold
4. Set severity (low → critical)
5. Pick a channel and destination
6. Save

Condition types

Type	Use
latency	P95 or avg latency above threshold (ms)
error_rate	Error ratio 0–1
cost	Cost threshold (USD)
anomaly	Active anomaly count
provider_outage	Provider error rate
retry_storm	Retries per hour
cost_explosion	Cost per hour
token_spike	Tokens per hour
webhook_degradation	Inbound webhook failures per hour

Notification channels

Channel	Configuration
email	Recipient email address
slack	Slack incoming webhook URL
webhook	Any HTTPS URL — JSON POST on trigger (same URL field as Slack)

Alert evaluation runs on a schedule (platform cron). Cooldowns prevent alert storms (default 30 minutes per rule).

Alert history

View recent triggers on the Alerts page (GET /api/alerts/history). Each entry includes rule name, severity, timestamp, and JSON details.

Email delivery

Welcome and alert emails use Resend when configured on the platform. Without Resend, triggers still appear in-dashboard and via webhook/Slack.

Best practices

• Start with error rate and latency on critical routes
• Use separate projects or AMLEXIA_ENVIRONMENT=staging so test traffic does not page production
• Pair alerts with Live during incidents
• Use Operations for SLO context

SDK and alerts

Alerts are dashboard-only — no SDK code changes. The SDK sends events; rules run server-side.

Related

• Dashboard API — alerts
• Operations