← Glossary / 98 of 100
n. / Safety and operations
Anomaly detection
Definition
Anomaly detection finds values that differ sharply from what is normal for a metric, without you setting a fixed threshold for each one.
In more detail
It compares current behavior with a baseline, often adjusted for time of day and week, and flags large departures. That helps with metrics that vary naturally, like traffic. It can still be noisy, so use it to point you at runs worth inspecting, and keep simple threshold alerts for the limits that must never be crossed.
Example
Average tokens per run is usually 3,000 but jumps to 9,000 on Tuesday afternoon. The system flags it, and a trace shows a new prompt that included a whole document.
Related terms
