aegisnet

A self-hosted network threat detection lab. It ingests Suricata telemetry, runs five deterministic detection rules and groups the alerts into incidents for an analyst dashboard.

State
releasedrelease v1.0.0 2026-09-21last commit 2026-10-21repositoryas of
Decision
The first run against real Suricata output showed that two of the five rules could not see real traffic, though every synthetic test passed: the data generators shared the detectors' wrong assumptions. I fixed the normaliser and corrected the generators so the synthetic data models a real sensor.
Limits
Detection accuracy on real traffic is unmeasured. The published numbers come from synthetic data written in the same repository, and one rule has never judged real traffic.
Stack
Python, FastAPI, PostgreSQL, Redis, Dramatiq, Next.js, TypeScript, Docker Compose, Suricata

ProofREADME.md:8README.md:105ADR-022:9ADR-022:72README.md:606STATUS.md:26issue #12pyproject.toml:15frontend/package.jsondocker-compose.yml:38docker-compose.lab.yml:93

cd ..