RepoAI
Browse MCPs

Kubectl MCP Server

Kubernetes infrastructure management via natural language AI conversations.

Data last scanned today · Reviewed today

522 dangerous tools55 tools★ 941PythonMIT

Overview

The Kubectl MCP Server provides a bridge between AI models and Kubernetes environments. It enables users to perform complex cluster operations—such as debugging pods, deploying applications, auditing security, and managing Helm charts—by issuing natural language commands rather than manual command-line inputs. The server is designed for interactive DevOps workflows, allowing AI assistants to visualize cluster dashboards and optimize resource costs directly. It is part of the CNCF landscape and supports integration across multiple AI interfaces using the Model Context Protocol.

Our verdict

This server carries a high-risk score of 42/100, primarily due to the fact that it is community-maintained and not backed by an official Kubernetes vendor, alongside its status as not yet editorially reviewed. Given that the authentication and read-only access modes remain unreviewed, users should exercise extreme caution when granting this tool access to sensitive production clusters until security audits are performed.

  • Supports zero direct dependencies for a reduced supply-chain surface
  • Comprehensive documentation provided within the repository
  • Maintains an active development status with 941 GitHub stars
  • Licensed under the permissive MIT license
  • Security-critical authentication methods have not been externally reviewed
  • Lacks verification of read-only mode, posing potential configuration risks
  • High-risk score due to reliance on community maintenance
  • Last repository commit occurred over four months ago

Setup

Claude Desktop
{
  "mcpServers": {
    "kubectl-mcp-server": {
      "args": [
        "-y",
        "kubectl-mcp-server"
      ],
      "command": "npx"
    }
  }
}

Paste into ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows), then restart Claude Desktop. Some servers require API keys or tokens as environment variables — check the project's README and add an "env" object if needed.

  • No read-only mode — running this grants full read/write access.
  • No built-in authentication — anyone with access to this config can invoke every tool it exposes.

Tools

ToolDescriptionRisk
get_podsList pods in the clusterlow
get_logsRetrieve logs from a specific podlow
get_pod_eventsFetch events associated with a podlow
check_pod_healthAnalyze the health of a podlow
exec_in_podExecute commands inside a container in a podhigh
cleanup_podsDelete pods according to specified criteriahigh
get_pod_conditionsGet status conditions for a podlow
get_previous_logsRetrieve logs from the previous container instancelow
get_deploymentsList deployments in the clusterlow
create_deploymentCreate a new deploymentmedium

Compatibility

ClientLocalDockerRemoteRead-only
Claude Desktop
Cursor
VS Code
Windsurf
ChatGPT

Frequently asked questions

›What are the language requirements to install this server?

The server requires Python 3.9 or higher if using pip, or a Node.js environment if using npx.

›How can I enable interactive dashboard features?

You can install the server with the optional UI features using the command 'pip install kubectl-mcp-server[ui]'.

›Is it possible to run this server without a permanent installation?

Yes, you can run the server directly using 'npx -y kubectl-mcp-server' for a zero-install experience.

›Does this server support managing multiple Kubernetes clusters?

Yes, the server includes multi-cluster support for managing your infrastructure.

›Can the server be deployed directly inside a Kubernetes cluster?

Yes, the README mentions support for in-cluster deployment.

Changelog

Badge

Kubectl MCP Server security score, rated on RepoAI

Maintain this server? Add the live badge to your README.