detection as code

Treating detection rules like software: stored in version control, reviewed, and tested automatically on every change.

In more detail

Each rule lives in a Git repository beside test events that must match (true positives) and must not (true negatives). A pipeline runs the tests before a rule is deployed, so a tuning change that breaks a detection is caught. Every missed attack or false positive becomes a new named test case.

Related terms

Taught in

References