mcpField
K
k8s-resource-optimizer

k8s-resource-optimizer

github.com/jorgealvesCloud & DevOps2213 installsUpdated 3d ago
Repository

About k8s-resource-optimizer

k8s-resource-optimizer helps teams right-size Kubernetes container resource requests and limits by comparing configuration with actual usage metrics. It is intended for cost optimization, reliability reviews, and pre-production planning, not real-time autoscaling.

Key features

  1. 01Compare resource settings with usage metrics
  2. 02Identify under-provisioned containers
  3. 03Identify excess resource allocation
  4. 04Work from text or JSON metrics
  5. 05Review infrastructure-as-code files

Use cases

  1. 01Find waste during cluster cost reviews
  2. 02Set memory thresholds after OOMKills
  3. 03Plan resource requests and limits before launch

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/jorgealves/agent_skills --skill k8s-resource-optimizer

After installing, type/k8s-resource-optimizerin Claude Code to confirm it loaded.

Bundled filesskill.yaml

Details

From the project README

K8s Resource Optimizer

Purpose and Intent

The k8s-resource-optimizer helps teams balance performance and cost. It identifies containers that are either "starving" (causing crashes) or "bloated" (wasting money) by comparing their configuration against actual usage patterns.

When to Use

  • Cloud Cost Optimization: Run monthly to identify waste in your clusters.
  • Reliability Engineering: Use after a production incident involving OOMKills to find the correct memory threshold.
  • Pre-production Scaling: Set realistic requests/limits before a major launch.

When NOT to Use

  • Initial Development: Don't over-optimize before you have real traffic patterns.
  • Real-time Autoscaling: This tool is for "right-sizing" configuration; use a HPA/VPA for real-time adjustments.

Security and Data-Handling Considerations

  • No live cluster interaction required if metrics are passed as text/json.
  • Safe to run on infrastructure-as-code files.

Comments

Markdown supported