Kost
Go agent that flags over-provisioned
pods and hands you the kubectl fix.
Findthewaste.
Over-Provisioning
Spots pods that request more CPU or memory than they actually use, and shows exactly what it's costing you every month.
Learn more →
The fix, ready
Every finding comes with a ready kubectl set resources command — copy, paste, apply, no digging through YAML to figure out the fix.
Learn more →
Slack Alerts
Wire up a webhook and waste notifications land straight in your team channel. Push over pull — you don't go check a dashboard, the alert comes to you.
Learn more →
Oneagent,onaloop.
Every pod, actual usage.
Live CPU and memory from the Metrics API — not what was requested.
Only flags what's wasted.
Compares requests against real usage — nothing gets flagged without clearing both thresholds.
One finding, four destinations.
Same JSON, sent to stdout, Prometheus, the dashboard, and Slack.
Security&Isolation.
Nothing leaves your cluster.
Zero third-party API calls. Slack webhook is the only outbound, and it's optional.
Runs as non-root.
runAsNonRoot: true, UID 65534. No path to root inside the pod.
Read-only by design.
ClusterRole grants only get/list/watch — pods, nodes, PVCs, Metrics API. No write, no exec, no secrets.
TryKostnow.
The project's open and ready for contributors — dig into the code or open an issue anytime.
View on GitHub