Aegis: Self-Healing Kubernetes Platform
A Go controller that remediates unhealthy pods, with metrics-gated canary rollouts.
Why it exists
Crash-looping and stuck-Pending pods need someone to notice and intervene. Aegis detects and remediates them automatically, and makes every deploy prove itself before it gets full traffic.
- Event-driven Go controller built on client-go informers that remediates CrashLoopBackOff pods and sweeps for pods stuck Pending past a timeout
- Argo Rollouts canary (25% → analysis → 50% → 100%) gated on a Prometheus success-rate query
