Alerts Overview

Quick Access: Manage Alerts

LangMart's alerting system helps you stay informed about your API usage, costs, and potential issues. Set up alerts to receive notifications when important thresholds are crossed.

Why Use Alerts?

Alerts provide proactive monitoring so you can:

  • Control costs - Get notified before overspending
  • Detect issues early - Catch errors before they impact users
  • Monitor performance - Track latency and response times
  • Prevent downtime - Know when quotas are nearly exhausted
  • Spot anomalies - Detect unusual usage patterns

Alert Types

LangMart supports six types of alerts, each designed for different monitoring needs:

Cost Threshold

Monitor your spending and get notified when costs exceed your budget.

Use cases:

  • Daily spending limits
  • Monthly budget caps
  • Per-project cost tracking

Example: Alert when daily spending exceeds $50

Error Rate

Track API error rates and catch problems early.

Use cases:

  • API reliability monitoring
  • Integration health checks
  • Early issue detection

Example: Alert when error rate exceeds 5% in the last hour

Usage Spike

Detect sudden increases in request volume compared to your baseline.

Use cases:

  • Runaway script detection
  • Attack detection
  • Viral traffic handling

Example: Alert when requests increase by 200% above normal

Quota Limit

Get warned before running out of credits or hitting rate limits.

Use cases:

  • Credit balance monitoring
  • Rate limit awareness
  • Capacity planning

Example: Alert when 80% of monthly quota is used

Model Performance

Monitor response times and ensure API performance meets expectations.

Use cases:

  • SLA monitoring
  • User experience optimization
  • Performance regression detection

Example: Alert when average latency exceeds 10 seconds

Unusual Activity

Detect patterns that deviate from your normal usage.

Use cases:

  • Security monitoring
  • Fraud detection
  • Anomaly detection

Example: Alert on unusual request patterns or sources

How Alerts Work

Alert Lifecycle

  1. Create - Define the alert with conditions and notification preferences
  2. Active - Alert monitors your usage continuously
  3. Triggered - When conditions are met, notifications are sent
  4. Tracked - All triggers are recorded in alert history
  5. Acknowledge - Mark alerts as reviewed to track resolution

Notification Channels

Alerts can notify you through:

  • Dashboard - Visible in the LangMart interface
  • Email - Sent to your registered email address

You can enable one or both channels per alert.

Alert States

Each alert has a status:

State Description
Active Alert is monitoring and will trigger when conditions are met
Inactive Alert is paused and will not trigger
Triggered Alert has fired (can still be active)

Getting Started

Quick Setup with Default Alerts

New users can add recommended alerts with one click:

  1. Go to Alerts page
  2. Click Add Recommended Alerts
  3. Review and enable the alerts you want

Default alerts include:

  • Daily Cost Alert ($10 threshold)
  • High Error Rate Alert (5% threshold)
  • Usage Spike Alert (200% increase)
  • Quota Warning (80% used)
  • Slow Model Response (10 second latency)

All default alerts are created disabled, so you can review and customize before enabling.

Custom Alerts

Create alerts tailored to your needs:

  1. Click Create Alert
  2. Choose alert type
  3. Set threshold and period
  4. Select notification channels
  5. Save and enable

Alert Dashboard

The Alerts page shows:

Summary Cards

  • Total Alerts - Number of configured alerts
  • Active Alerts - Alerts currently monitoring
  • Triggered Alerts - Alerts that have fired

Alert List

Each alert displays:

  • Name and description
  • Alert type with icon
  • Status (active/inactive)
  • Trigger condition
  • Notification channels
  • Last triggered time
  • Trigger count

Actions

For each alert, you can:

  • Test - Send a test notification
  • Enable/Disable - Toggle alert status
  • Delete - Remove the alert

Best Practices

Start with Defaults

Use the recommended alerts as a starting point, then customize based on your needs.

Set Realistic Thresholds

  • Too low: Alert fatigue from constant notifications
  • Too high: Missing important issues
  • Just right: Actionable notifications you'll actually respond to

Use Multiple Channels

Enable both dashboard and email notifications for critical alerts to ensure you don't miss important issues.

Review Regularly

Check your alert history periodically to:

  • Identify patterns
  • Adjust thresholds
  • Remove unnecessary alerts

Test Your Alerts

Use the Test button to verify notifications are working before relying on them.

Next Steps