kmcp

kmcp is a development platform and control plane for MCP servers. It provides a CLI toolkit that lets you scaffold new MCP projects, manage tooling, build container images, and run your MCP server locally for testing and development. The system also includes a Kubernetes-oriented controller and a transport adapter layer to expose MCP servers over HTTP, WebSocket, SSE, and other protocols, making it easier to move prototypes into production. With kmcp you can quickly spin up MCP services, package them for deployment, and manage their lifecycle across environments.

To get started, install the kmcp CLI on your machine and verify it’s available on your PATH. Use the toolkit to scaffold a new MCP service, customize its configuration, and then run the server locally for development. When you’re ready for production, you can build container images, deploy to Kubernetes with the pre-configured transport adapters, and manage MCP server deployments via Kubernetes CRDs. The CLI also provides commands to manage dependencies, generate boilerplate code for supported runtimes (such as FastMCP Python and MCP Go SDK), and integrate security and observability tooling into your MCP workflows.

Prerequisites:

• curl or wget
• Node.js (recommended for the kmcp CLI) or a compatible runtime environment
• Internet access to fetch bootstrap scripts and npm packages

1. Install prerequisites

• Ensure Node.js and npm are installed:
  • On macOS/Linux: curl -fsSL https://get.nodejs.org/ | bash -
  • Or use your system package manager to install Node.js (which includes npm)

2. Install the kmcp CLI (via the recommended bootstrap script)

• Run the official bootstrap to install the kmcp CLI: curl -fsSL https://raw.githubusercontent.com/kagent-dev/kmcp/refs/heads/main/scripts/get-kmcp.sh | bash

3. Verify installation

• Check the CLI help to confirm installation: kmcp --help

4. (Optional) Set up a local MCP server project

• Initialize a new MCP server project in a local directory using the kmcp CLI as per the docs: kmcp init my-mcp-server

5. Run or deploy your MCP server locally

• Use the CLI to run the server for development and testing, or deploy to your Kubernetes cluster using the provided transport adapters and CRDs.

Tips and common considerations:

• The kmcp CLI focuses on rapid scaffolding and one-command deployment workflows to Kubernetes via pre-configured transport adapters.
• When deploying to production, ensure you configure security, secrets management, and observability in your Kubernetes environment.
• For local development, you can scaffold a server and run it with the CLI to validate MCP interactions before containerizing.
• If you encounter network or permission issues while installing the kmcp CLI, ensure your firewall allows script downloads and that you have the necessary permissions to install global npm binaries if applicable.
• Refer to the official documentation at kagent.dev/docs/kmcp for runtime-specific guidance (e.g., FastMCP Python, MCP Go SDK) and architecture workflows.