Model Context Protocol moderate risk

Kubectl

kubectl-mcp-server enables natural language interaction with Kubernetes clusters, allowing users to manage and troubleshoot their infrastructure via AI assistants.

Connections & Capabilities

Connects To

GitHubRedisDockerKubernetesVercel

Capabilities

readwriteexecadmin

Quickstart

Install

npm install -g skillkit

Config

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

Exposed MCP Tools (5)

safe

get_pods

Retrieves a list of pods in a specified namespace.

Read-only operation, no modification of cluster state.

moderate

create_deployment

Creates a new deployment in the Kubernetes cluster.

Creates resources, but doesn't directly impact existing critical components.

moderate

scale_deployment

Scales the number of replicas for a given deployment.

Modifies the number of pods, potentially impacting application availability.

high

exec_pod

Executes a command inside a specified pod.

Allows arbitrary command execution within a pod, potentially leading to security breaches.

critical

delete_pod

Deletes a pod from the cluster.

Destructive operation that can cause service disruption.

Safety Assessment

kubectl-mcp-server offers a convenient way to interact with Kubernetes using natural language. It's relatively safe for read-only operations and exploration. However, caution is advised when performing write operations, and proper RBAC configuration is crucial to mitigate risks.

Supports non-destructive mode to prevent accidental changes
RBAC validation ensures actions are authorized
Secret masking protects sensitive information
Provides read-only access for safe exploration

Can perform write operations on the cluster
Requires kubectl to be configured correctly, which can be complex
Potential for privilege escalation if RBAC is misconfigured
Exposes Kubernetes API to AI assistant, increasing attack surface
Incorrect natural language commands could lead to unintended actions