workshop s

This workshop provides a hands-on exploration of Model Context Protocol (MCP) servers, focusing on Node.js/TypeScript implementations for stdio and streamable HTTP transports, as well as a Java/Spring AI example. The MCP client (e.g., a VS Code extension or an HTTP client) interacts with the MCP servers to issue intent-driven requests that are handled by the server's endpoint logic. The Node.js stdio server demonstrates a simple Hello World tool (greetPerson) and can be extended to greet audiences via greetAudience. The streamable HTTP variant showcases a scalable transport that supports multiple concurrent clients. The Java Spring AI example illustrates how to implement the same patterns with a different tech stack, including how to integrate with the MCP protocol using standard IO or HTTP-based streams. To use the servers locally, run the configured commands for each server and then exercise the provided MCP workflows from your client of choice (VS Code Copilot integration is a primary focus of the workshop).

Prerequisites:

• Node.js 24 (for the TypeScript/Node.js MCP servers)
• Java 25 (for the Java/Spring AI MCP server)
• Git and a local development environment (VS Code recommended)

Installation steps:

1. Clone the repository containing the workshop MCP servers.
2. Install Node.js dependencies for the Node-based servers (in each Node server folder, run npm install if a package.json is present).
3. Ensure Java is installed and available in your PATH for the Java/Spring AI server.
4. Start the MCP servers using the commands described in the mcp_config (e.g., node 01-typescript-mcp-hello-world-stdio/server.js). If a build step is required for the TypeScript projects, run the usual tsc or npm run build in the respective folders prior to starting.
5. Verify connectivity with your MCP client by invoking a simple greetPerson or greetAudience request and confirming the responses.

Note: The workshop content relies on local execution; there is no cloud deployment required. You can follow the folder-by-folder workflow to learn the MCP server development lifecycle.

Tips and considerations:

• The Node.js stdio variant uses a local process I/O channel; the Streamable HTTP version is suitable for larger-scale scenarios with multiple clients.
• For Java/Spring AI, ensure the correct Spring AI dependencies are present and that the server jar can be executed with java -jar.
• If you encounter port or transport binding issues, double-check that the MCP client is configured to use the same transport (stdio vs. HTTP) as the target server.
• When debugging, VS Code with the GitHub Copilot extension is commonly used in this workshop to simulate MCP client interactions; however, you can also test with curl or a minimal MCP client implemented in your language of choice.
• Environment variables may be used to configure endpoints, authentication, or feature flags in more advanced scenarios; add them under the env object in each mcpServers entry if needed.