Network Threat Detection Guide
This skill provides the authoritative detection thresholds for identifying
malicious network patterns — port scans, denial-of-service floods, and
command-and-control beaconing. The thresholds are calibrated against the
DefenseNet Threat-Calibration corpus and are the values that produce
correct results on the analysis flags has_port_scan, has_dos_pattern,
has_beaconing, and is_traffic_benign.
Scope
This document covers the three detection families currently maintained in the corpus:
- Port scanning (entropy + SYN-only ratio + unique-port count)
- DoS / flood patterns (per-minute rate ratio)
- C2 beaconing (inter-arrival-time coefficient of variation)
- Benign classification (composition of the above)
How to Use This Skill
The concrete detection procedures, threshold values, worked examples, and reference Python implementations live in the companion reference PDF alongside this document:
threat_detection_reference.pdf — open this file and follow the
procedures it documents. It is the authoritative source for the thresholds
and the required pre-analysis steps; this SKILL.md is a framing shell and
is not sufficient on its own to compute the analysis flags correctly.
After reading the PDF, return here only for the high-level scope above.