Skip to main content

Recommendations

K8Cost generates actionable, prioritized recommendations for every optimization opportunity it finds in your clusters. Each recommendation tells you exactly what to change, why, and how much you will save.

What You Get​

Every recommendation includes:

  • Title and description -- A clear explanation of the issue (e.g., "CPU request is 4x higher than p95 usage")
  • Severity -- Critical, warning, or informational, based on the potential impact
  • Category -- Resource sizing, idle workloads, storage, HPA, replicas, security, or performance
  • Estimated monthly savings -- Dollar value calculated using your cloud provider's pricing
  • Current vs. recommended values -- The exact change needed (e.g., "Reduce CPU request from 2000m to 500m")
  • Risk assessment -- Cost reduction potential, performance impact, and risk level for each recommendation
  • Source rule -- Which optimization rule generated the recommendation, with the threshold that triggered it

Key Capabilities​

  • Filter and sort by cluster, namespace, category, severity, or estimated savings
  • Bulk approve multiple recommendations at once for batch application
  • Impact simulation -- Preview the aggregate effect of applying selected recommendations before committing
  • One-click GitOps export -- Generate patches in YAML, Helm, Kustomize, JSON Patch, or Strategic Merge Patch format
  • PR integration -- Create pull requests on GitHub or merge requests on GitLab directly from the recommendations page
  • Auto-resolve -- Recommendations are automatically marked as resolved when the underlying issue is fixed in a subsequent agent sync
  • Approve / reject workflow -- Track which recommendations your team has accepted, rejected, or deferred

Recommendation Lifecycle​

Generated  -->  Reviewed  -->  Approved  -->  Applied  -->  Resolved
                    |
                    +--> Rejected (with reason)

Recommendations are generated fresh on each agent sync. If you apply a recommendation and the next sync shows the issue is resolved, K8Cost automatically marks it as resolved. If a rejected issue reappears, the recommendation is regenerated so nothing falls through the cracks.

Sign up free to see recommendations for your clusters.