Mcp Kubernetes Ro
@patrickdappollonio
mcp-kubernetes-ro is a Model Context Protocol (MCP) server providing read-only access to Kubernetes clusters for AI assistants. It enables AI models to list resources, get resource details, retrieve pod logs, discover API resources, and perform base64 encoding/decoding operations
概要
What is Mcp Kubernetes Ro?
Mcp Kubernetes Ro is a read-only Model Context Protocol (MCP) server that provides AI assistants with safe, read-only access to Kubernetes clusters. It uses your local kubectl configuration to enable comprehensive cluster inspection and troubleshooting without requiring kubectl to be installed.
How to use Mcp Kubernetes Ro?
Install via pre-built binaries, Homebrew, or Docker. Configure it in your MCP client (e.g., editor settings) with the command mcp-kubernetes-ro and optional flags like --kubeconfig, --namespace, or --transport. It reads your default kubeconfig (~/.kube/config) and environment variable KUBECONFIG. Run in stdio mode by default, or use --transport=sse --port=8080 for HTTP SSE mode.
Key features of Mcp Kubernetes Ro
- Read‑only access to prevent destructive operations
- List, get, and filter Kubernetes resources
- Retrieve pod logs with grep, time, and container filtering
- Encode and decode base64 data for secrets
- Supports stdio and SSE transport modes
- Does not require kubectl binary installed
Use cases of Mcp Kubernetes Ro
- AI assistant inspects cluster resources without risk of modification
- Developer troubleshoots pods by retrieving filtered logs
- Security auditor decodes base64 secrets for review
- Operations engineer lists API resources across multiple contexts
- Metrics monitoring when metrics-server is installed in the cluster
FAQ from Mcp Kubernetes Ro
Does Mcp Kubernetes Ro require kubectl to be installed?
No. It uses your local kubectl configuration (kubeconfig) but does not depend on the kubectl binary.
Can Mcp Kubernetes Ro modify cluster resources?
No. The server provides read‑only access and prevents any destructive operations.
What transport modes are supported?
The server supports both stdio (default) and Server‑Sent Events (SSE) transport modes.
How do I specify a different kubeconfig file?
Use the --kubeconfig command‑line flag or set the KUBECONFIG environment variable.
Are node/pod metrics always available?
Metrics tools (get_node_metrics, get_pod_metrics) require the metrics‑server to be installed in the cluster; otherwise, they return an error.
