armor-test

Test Before Deploying

Preview what alerts would fire with proposed configurations. Avoid alert fatigue by testing thresholds before enabling.

Prerequisites

• AnomalyArmor API key configured (~/.armor/config.yaml or ARMOR_API_KEY env var)
• Python SDK installed (pip install anomalyarmor)

When to Use

• "Test this freshness threshold before I enable it"
• "What alerts would fire if I set the threshold to 4 hours?"
• "Dry-run the schema drift check"
• "Preview the impact of this rule"
• "Will this configuration cause too many alerts?"

Steps

Dry-Run Freshness Threshold

1. Specify the table and proposed threshold
2. Call client.freshness.dry_run() with proposed config
3. Review predicted alerts over historical data
4. Adjust threshold if too many/few alerts predicted
5. When satisfied, create the actual schedule

Preview Alert Rules

1. Specify the event types to filter
2. Call client.alerts.preview() with event types
3. See what historical events would have triggered alerts
4. Evaluate alert frequency
5. Adjust configuration as needed

Example Usage

Dry-Run Freshness Threshold

from anomalyarmor import Client

client = Client()

# Test what would happen with a 4-hour freshness threshold
# Uses historical data to predict alert frequency
result = client.freshness.dry_run(
    asset_id="asset-uuid",
    table_path="public.orders",
    expected_interval_hours=4,
    lookback_days=7
)

print(f"Configuration: Alert if stale for {result.threshold_hours} hours")
print(f"Historical period: {result.lookback_days} days")
print()
print(f"Total checks analyzed: {result.total_checks}")
print(f"Would alert count: {result.would_alert_count}")
print(f"Alert rate: {result.alert_rate_percent:.1f}%")
print()

if result.would_alert_now:
    print(f"Current status: Would alert NOW (age: {result.current_age_hours:.1f}h)")
else:
    print(f"Current status: OK (age: {result.current_age_hours:.1f}h)")

print(f"\nRecommendation: {result.recommendation}")