# Auto-Remediation: act on a recommendation in one click

> ZopNight introduces end-to-end auto-remediation. Click Remediate on a cost recommendation and the platform checks live cloud state, optionally routes to an admin for approval, performs the action, and validates the result, with 20 rules certified end-to-end on real AWS, GCP, and Azure. This release also adds 123 Kubernetes workload recommendations and names the exact target SKU on every rightsizing recommendation, pricing savings against live cloud rates.

Source: https://zop.dev/changelog/v1-7-0
Published: 2026-05-07 · Tags: zopnight, zopday, auto-remediation, recommendations, kubernetes

---

Click Remediate; ZopNight performs the action.

## New

- Auto-Remediation: click Remediate on a recommendation to stop compute, scale a service to zero, or pause it, with a live precondition check, optional admin approval, and result validation (20 rules certified across AWS, GCP, and Azure; errors grouped into clear categories)
- 123 Kubernetes workload recommendations across EKS, GKE, and AKS covering reliability (missing requests / limits, single replica without an HPA, missing probes), security (privileged or root containers, missing TLS), idle and orphaned workloads, and rightsizing

## Improved

- Every rightsizing recommendation now names a concrete target SKU (downsize to `m5.large`, step `db.t3.medium` -> `db.t3.small`) and prices savings as current minus target from live cloud rates. If a target isn't priceable in the region, the recommendation is skipped rather than shown with a guessed number
- Live per-container metrics extended to the Deployments, StatefulSets, DaemonSets, and ReplicaSets detail drawers
- ZopDay: Add Existing clusters (EKS, GKE, AKS) now work end-to-end for component installs, datastore connections, and database provisioning
- ZopDay: cluster reachability is now persisted and gates actions: a definitively unreachable cluster disables Sync, Connect Datastore, and Install, while transient blips are retried silently
- ZopDay: database creation now runs inside your own cluster, so master credentials never leave it and ZopDay needs no outside network path to your datastore (MySQL, PostgreSQL)