Observed arrival · 2026-09-04
Hephaisto — Kubernetes incident response with brakes
Hephaisto is an incident-response agent for Kubernetes that investigates alerts, proposes bounded actions, verifies their results, and escalates when it cannot prove a fix.
Field notes
The documented workflow separates diagnosis, planning, and execution: the model gathers evidence and proposes a structured action, while ordinary C# code handles the closed action list. A fresh install starts in observe mode with no actionable namespaces or enabled action types. The page also describes three independent kill-switch arms and verification checkpoints at 60 seconds, five minutes, and 15 minutes, with rollback and escalation when the stated predicate fails.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue