Kmcp is an MCP server development and deployment tool created by Kagent Dev, operating on the mcp platform. As a subproject of kagent, an AI agent platform for Kubernetes, Kmcp accelerates local development and manages the lifecycle of MCP servers in cloud-native environments. It supports scaffolding projects in Python (FastMCP) and Go (MCP Go), with plans to support TypeScript, Golang, Java, and JavaScript frameworks. The tool is distributed as a CLI and a Kubernetes Controller, integrating deeply with Kubernetes for deployment and management.

What does Kmcp connect to?

Kmcp connects to and manages Model Context Protocol (MCP) servers, deploying them to Kubernetes environments. It functions as a platform to accelerate local development and manage the lifecycle of MCP servers in cloud-native environments. Kmcp is optimized for AI Tool Integration, with services having built-in support for agentgateway, part of the kagent AI agent platform. This allows it to work with various Large Language Model (LLM) providers such as OpenAI, Anthropic, Google Gemini, xAI, Azure OpenAI, Vertex AI, and Ollama.

Core features and tools

Kmcp exposes several command-line tools to manage MCP server development and deployment. These include kmcp init for initializing new MCP projects, kmcp build for creating production-ready container images, and kmcp deploy for Kubernetes deployments. The kmcp install command sets up the necessary Kubernetes controller components. Other callable tools are kmcp run, kmcp secrets, kmcp add-tool, kmcp completion, and kmcp help. It supports stdio and HTTP Streamable transport protocols.

Who is it for?

Kmcp is designed for developers building MCP servers, AI/ML engineers packaging existing prototypes, DevOps engineers constructing MCP infrastructure in Kubernetes, and platform engineers. It supports rapid local development and production deployments in cloud-native environments. Skip this if you are not working with Kubernetes-native MCP server development or deployment.

How to get started

Access is granted through a standard OAuth authorization flow, integrated with agentgateway for security. Kmcp services can read environment variables from Kubernetes secrets for sensitive information, and supports HTTP, WebSocket, and Server-Sent Events transport protocols.

Use cases

• Scaffolding and local development: Initialize new MCP projects (e.g., "kmcp init python my-weather-server") with pre-configured templates that include source code, tests, Dockerfile, dependencies, and Kmcp configuration, allowing for rapid local development and testing.
• Building and packaging for production: Create production-ready container images for MCP services, including cross-platform options using the "kmcp build" command.
• Kubernetes-native deployment and lifecycle management: After a one-time "kmcp install" to set up the controller, deploy and update MCP services to Kubernetes using "kmcp deploy", managed as first-class Kubernetes resources via MCPServer custom resources, with automatic handling of service discovery and port mapping.

Kmcp provides a platform for building and managing Model Context Protocol servers on Kubernetes. It simplifies local development and orchestrates cloud-native deployments, integrating with AI agent platforms. Its toolset supports scaffolding, building, and deploying MCP services, handling critical lifecycle aspects. The focus on Kubernetes and AI agent platforms makes it a specialized offering. Start by initializing a new Python-based MCP project using "kmcp init python my-server".