silenceper/mcp-k8s
FreeAI 驱动的 Kubernetes 资源管理工具，允许通过自然语言操作 K8s 集群中的任何资源 (原生/CRD)。
FreeFree tier
Inputs: text
About silenceper/mcp-k8s
mcp-k8s is a Kubernetes MCP (Model Control Protocol) server that enables natural language interaction with Kubernetes clusters. It supports querying and performing CRUD operations on all Kubernetes resource types, including Custom Resource Definitions (CRDs), and provides comprehensive Helm release and repository management. Each operation can be independently enabled or disabled for fine-grained control. Built in Go using the mcp-go SDK and Kubernetes client-go library, it connects to clusters via kubeconfig and communicates over stdio. The tool is designed for use with LLM-powered assistants to simplify Kubernetes management, automation, debugging, and education.
Key Features
Query supported Kubernetes resource types, including built-in resources and CRDs
CRUD operations on Kubernetes resources (create, read, update, delete) with independent enable/disable control
Full Helm support: release management (list, get, install, upgrade, uninstall) and repository management (list, add, remove)
Connects to Kubernetes clusters using kubeconfig
Each operation can be independently enabled or disabled for fine-grained security
Supports all Kubernetes resource types, including custom resources
Stdio-based MCP server implemented in Go
Integrates with LLM assistants for natural language interaction
Pros & Cons
Pros
- Supports all Kubernetes resource types including CRDs, offering full cluster management
- Fine-grained control over read and write operations, enhancing security
- Integrates Helm release and repository management, covering common packaging workflows
- Natural language interface lowers the barrier for Kubernetes operations
- Open source (MIT license) with active community and clear documentation
Cons
- Requires a Model Context Protocol (MCP) compatible AI client to use
- Depends on kubeconfig for cluster access, which may require additional configuration for some environments
- Command-line/stdio interface may not suit all deployment scenarios
Best For
Interactive Kubernetes resource management via natural language, eliminating kubectl complexityBatch operations: describe complex tasks in natural language and let LLM translate them into resource operationsCluster status queries and problem diagnosis through conversational interfacesDevelopment and testing: quick prototype validation, environment management, and configuration generationEducation and training: interactive learning of Kubernetes concepts, best practice guidance, and error explanation
FAQ
What is mcp-k8s?
mcp-k8s is a Kubernetes MCP (Model Control Protocol) server that allows LLM-powered assistants to interact with Kubernetes clusters through natural language. It supports CRUD operations on all resource types (including CRDs) and Helm management.
How do I connect mcp-k8s to my Kubernetes cluster?
mcp-k8s uses your local kubeconfig file to connect to the Kubernetes cluster, similar to kubectl. Ensure your kubeconfig is correctly configured and accessible.
Can I disable write operations for safety?
Yes, each operation (create, update, delete) can be independently enabled or disabled via configuration, allowing fine-grained control over cluster modifications.
Does mcp-k8s support Helm?
Yes, it supports Helm release management (list, get, install, upgrade, uninstall) and Helm repository management (list, add, remove), with independent enable/disable options.