Kubeopt

Name: Kubeopt
Author: kubeopt

@kubeopt

关于 Kubeopt

Kubernetes cluster cost analysis and optimization. Query costs, find savings opportunities, get rightsizing recommendations, and analyze pod spend across AKS, EKS, and GKE directly from Claude.

#kubernetes #devops #cost-optimization #cloud #finops #k8s #aws #azure #gcp

基本信息

分类

其他

传输方式

stdio

发布者

kubeopt

提交者

SrinivasK

配置

使用下面的配置,将此服务器添加到你的 MCP 客户端。

{
  "mcpServers": {
    "kubeopt": {
      "command": "python3",
      "args": [
        "-m",
        "mcp_server.server"
      ],
      "env": {
        "KUBEOPT_API_URL": "http://localhost:5001",
        "KUBEOPT_USERNAME": "kubeopt",
        "KUBEOPT_PASSWORD": "your-password"
      }
    }
  }
}

工具

未检测到工具

工具是从 README 中自动提取的。维护者可以在 ## Tools 标题下列出工具,即可填充这部分内容。

概览

What is Kubeopt?

Kubeopt is the MCP server component of KubeOpt, a Kubernetes cost engineering tool. It exposes six tools over stdio transport, allowing AI assistants like Claude Desktop, Cursor, and Windsurf to query cluster cost data, optimization recommendations, and trigger fresh analyses from a running KubeOpt instance.

How to use Kubeopt?

Run KubeOpt locally (python main.py) or deploy it, then configure your MCP client to execute the server script (python -m mcp_server.server) with the environment variables KUBEOPT_API_URL, KUBEOPT_USERNAME, and KUBEOPT_PASSWORD. After restarting the client, ask questions in plain English about Kubernetes costs.

Key features of Kubeopt

Six MCP tools for querying cost and cluster data
Connects to Azure AKS, AWS EKS, and Google GKE
Runs 16 optimization algorithms (rightsizing, HPA, storage, etc.)
Generates actionable implementation plans with kubectl commands
Provides per-pod cost breakdowns filterable by namespace
Supports on-demand fresh analyses with polling

Use cases of Kubeopt

Query top cost savings opportunities across all clusters
Identify the most expensive pods in a specific namespace
Get a summary of total Kubernetes spend for the current month
Check optimization scores for a staging cluster
Trigger a fresh analysis on a specific cluster and receive results

FAQ from Kubeopt

What tools does the MCP server expose?

It exposes six tools: list_clusters, get_cost_summary, get_cluster_analysis, get_recommendations, analyze_cluster, and get_pod_costs.

What are the prerequisites for using the MCP server?

A running KubeOpt instance (locally via python main.py or deployed on Railway) and a Python virtual environment with dependencies installed (pip install -r requirements.txt).