Detections are only useful if they survive contact with a real adversary. This is an ongoing library of detection-engineering write-ups that start from attacker behaviour and work backwards to the telemetry and logic that would actually catch it.

Every piece is mapped to MITRE ATT&CK so defenders can see which techniques are covered and, just as importantly, which gaps remain.

The approach

Most detection content starts from a product and lists what it can alert on. This library inverts that. Each write-up begins with how an attack actually unfolds — what the adversary needs to achieve, in what order, and what traces each step unavoidably leaves — and only then asks which log source captures it and what logic separates it from normal activity.

Working that direction exposes something vendor-led content usually hides: the difference between behaviour that is hard to avoid emitting and behaviour that is merely commonly observed. Detections built on the former survive an attacker changing tooling; detections built on the latter break the moment someone swaps a binary or renames a service.

What each write-up covers

In the library

Tooling

Several write-ups pair with tools on this site so the techniques can be practised immediately: the email header parser for authentication tracing, the URL scanner for phishing structure, and the hash analyzer for triage.

Read the full series in the research feed.