MCP K8s Eye

You run an MCP server called MCP K8s Eye to manage Kubernetes workloads and quickly diagnose cluster health. It connects to Kubernetes, supports common and custom resources, and provides diagnostics and resource usage insights to help you keep clusters healthy and workloads running smoothly.

How to use

You connect your MCP client to the MCP K8s Eye HTTP endpoint to access Kubernetes workload analysis and diagnostics. Use the provided HTTP transport endpoint to receive structured data about pods, deployments, services, statefulsets, cronjobs, ingresses, network policies, webhooks, nodes, and more. You can request resource descriptions, lists, and diagnostic analyses, then act on the results to repair misconfigurations or optimize resources.

How to install

# prerequisites
- Go 1.23 or higher
- kubectl configured

# clone the MCP K8s Eye project
git clone https://github.com/wenhuwang/mcp-k8s-eye.git
cd mcp-k8s-eye

# build the binary
go build -o mcp-k8s-eye

Additional content

Configure your MCP client to connect via HTTP to the endpoint described below. If you prefer a local, embedded usage pattern, you can use the SSE transport URL when starting or configuring your MCP client.

{
  "mcpServers": {
    "k8s eye": {
      "url": "http://localhost:8080/sse",
      "env": {}
    }
  }
}

Available tools

resource_get

Get detailed information about a specific resource in a namespace.

resource_list

List detailed information about all resources in a namespace.

resource_create_or_update

Create or update a resource in a namespace.

resource_delete

Delete a resource in a namespace.

resource_describe

Describe a resource in a namespace with detailed information.

deployment_scale

Scale a deployment in a namespace.

pod_exec

Execute a command inside a pod in a namespace.

pod_logs

Fetch logs from a pod in a namespace.

pod_analyze

Diagnose all pods in a namespace to assess status and resource usage.

deployment_analyze

Diagnose all deployments in a namespace for availability and readiness.

statefulset_analyze

Diagnose all statefulsets for service, PVCs, and replica availability.

service_analyze

Diagnose services for selector correctness, endpoints, and events.

cronjob_analyze

Diagnose cronjobs for schedule and last run status.

ingress_analyze

Diagnose ingress configurations and related TLS secrets.

networkpolicy_analyze

Diagnose network policies and their effect on pods.

validatingwebhook_analyze

Diagnose validating webhooks and their referenced services and pods.

mutatingwebhook_analyze

Diagnose mutating webhooks and their referenced services and pods.

node_analyze

Diagnose node conditions and health across the cluster.

workload_resource_usage

Get CPU and memory usage for pods, deployments, replicasets, statefulsets, and daemonsets.