fpd -workshop

This MCP server project provides a hands-on setup to connect external tools and services to AI via the MCP (Model Context Protocol) framework. The workshop focuses on building a simple MCP server that can be extended to integrate tools (such as Slack, Zapier, or Claude) and expose capabilities to an AI assistant for orchestration and automation. The included simple server serves as a minimal host that you can run locally, then expand by adding tools, authentication, and command handlers following the MCP guidelines.

To use the server, start the Node.js process as configured in the mcp_config. Once running, you can point an MCP-enabled client (like Claude or another orchestrator) to the server endpoint, load available tools, and issue prompts that leverage those tools. The workshop emphasizes practical integration patterns: wiring authentication for the tools you already use, testing interactions in a chat-style interface, and iterating on tool capabilities such as sending messages to Slack, retrieving channel data, or triggering external workflows via Zapier.

If you’re following along the Hands-On Exercise #3 pathway, you’ll clone this repo, navigate to the simple-server folder, and run the server to observe how the MCP protocol enables tool invocation and result handling within an AI-driven chat session.

Prerequisites:

• Node.js (LTS version) and npm installed on your machine
• Git to clone the repository
• Basic familiarity with running command-line processes

Install steps:

1. Clone the repository: git clone https://github.com/modelcontextprotocol/mor10-fpd-mcp-workshop.git
2. Navigate to the project directory: cd mor10-fpd-mcp-workshop
3. Open the simple-server folder README for any folder-specific setup instructions (if required).
4. Install Node dependencies (if a package.json exists in the root or simple-server): npm install
5. Run the MCP server as configured in the documentation (the example uses a Node server at simple-server/server.js): node simple-server/server.js
6. Verify the server is listening on the expected port (commonly 3000 or 8080) and that any required environment variables are set as needed.

Notes:

• If the server expects environment variables (e.g., API keys for tools), set them in your shell or an .env file and ensure they are loaded by your Node process.
• If you modify the server, you may need to restart the process to apply changes.

Tips and common considerations:

• This workshop demonstrates a minimal MCP server scaffold. You can extend it by adding authentication, tool adapters, and new endpoints as you grow your MCP capabilities.
• If you plan to connect to external services (Slack, Zapier, Claude, etc.), ensure you have valid accounts and grant the necessary permissions during the authentication flows.
• Check the simple-server/README for any folder-specific run instructions or required environment variables.
• When testing, use small, deterministic prompts to verify tool invocations and responses before scaling to more complex scenarios.
• If you run into port conflicts, change the server port in the configuration or environment and update any client configurations accordingly.
• Keep your MCP client and server versions aligned with the MCP protocol version used in the workshop materials.