METHOD AND PRODUCT BOUNDARIES
Interpret the output before acting on it.
GreenKube brings cost, energy and carbon estimates into dashboards and reports, with evidence-backed recommendations and reviewable GitOps proposals.
Inputs, source coverage and allocation shape what estimates and post-change outcomes can say.
From telemetry to action
Telemetry becomes an estimate and a report.
The workflow combines visibility and reporting with evidence-backed recommendations and GitOps proposals. Exact metric requirements, model details and configuration belong in the documentation.
Kubernetes context + supported Prometheus telemetry
↓
Cost / energy / carbon estimates
↓
Dashboards and reports
↓
Evidence-backed recommendations
↓
GitOps pull request and review
↓
Post-change signal checks
Estimation
Inputs, allocation and uncertainty travel together.
Cost values are estimates or allocations, not billing guarantees. Energy and carbon values are model outputs; the operational energy model is CPU-based and is not a direct hardware power measurement.
Workload allocation can distribute shared values, but it does not establish exact physical consumption by a pod. Missing metrics, contextual gaps, source freshness and geographic coverage can change what a report represents.
Use the Carbon page for the source families, proxy limits, fallback questions and allocation boundaries that should accompany an interpretation.
Optimization workflow
From evidence to a reviewed change.
GreenKube connects recommendation evidence, GitOps proposals and post-change checks while keeping production decisions with the operator.
- 01
Evidence and recommendation ranking
Gather evidence, expose confidence or risk, and rank possible work for operator review.
- 02
GitOps pull-request bot
Propose supported manifest changes as pull requests; target support depends on the configured integration.
- 03
Apply detection and verification
Detect applied changes and check post-change signals against defined evidence or health gates.
- 04
Measured outcomes and savings
Compare observed outcomes with estimates while treating attribution and savings as evidence-dependent.
- 05
VPA and Karpenter connectors
Connect supported VPA and Karpenter configurations to resource recommendations.
Human control
Keep production changes under operator control.
GreenKube proposes changes through pull requests for supported targets; teams review and merge them through their existing controls. The software does not automatically merge changes, roll back deployments or guarantee a healthy outcome.
A projected cost or carbon impact is not a measured result. A lower CPU request does not automatically reduce energy use or an invoice.
Use the canonical documentation for implementation detail.
This marketing page avoids reproducing setup instructions, exact model defaults or configuration-dependent integration behavior. Check the documentation before configuring a source or interpreting a report.