1 comments

  • PhantomArmor 9 hours ago
    Author here. Built this after struggling to train detection models on compressed/outdated datasets.

    Key characteristics: - 23-day attack timeline (mirrors real APT dwell time) - 8M+ logs with 0.007% attack signal - Service account hijacking patterns - Defense product logs (EDR, DLP, SIEM)

    Single scenario, intermediate skill. Has limitations.

    Feedback welcome - what other scenarios would be useful?