Create a new file called index.js with the following code:
import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js";
import { z } from "zod";
const server = new McpServer({
name: "My First MCP Server",
version: "1.0.0"
}, {
capabilities: {
resources: {}, // Enable resources support
tools: {} // Enable tools support
}
});
const transport = new StdioServerTransport();
await server.connect(transport);💡 Note: The server configuration object specifies which capabilities your server supports. You can enable resources, tools, and other features as needed.
Resources allow your server to expose data that can be used as context for LLM interactions:
// List available resources
server.setRequestHandler(ListResourcesRequestSchema, async () => {
return {
resources: [
{
uri: "file:///data/config.json",
name: "Configuration File",
mimeType: "application/json"
}
]
};
});
// Read resource contents
server.setRequestHandler(ReadResourceRequestSchema, async (request) => {
const uri = request.params.uri;
if (uri === "file:///data/config.json") {
return {
contents: [
{
uri,
mimeType: "application/json",
text: JSON.stringify({ key: "value" })
}
]
};
}
throw new Error("Resource not found");
});💡 Note: Resources can be static files, dynamic data, or templates. They're perfect for providing context to LLMs.
Tools are functions that LLMs can call to perform actions. Here's how to implement them:
// Simple calculator tool
server.tool(
"calculate",
{
operation: z.enum(["add", "subtract", "multiply", "divide"])
.describe("The mathematical operation to perform"),
x: z.number().describe("First number"),
y: z.number().describe("Second number")
},
async ({ operation, x, y }) => {
let result;
switch (operation) {
case "add": result = x + y; break;
case "subtract": result = x - y; break;
case "multiply": result = x * y; break;
case "divide":
if (y === 0) throw new Error("Division by zero");
result = x / y;
break;
}
return {
content: [
{
type: "text",
text: `Result of ${x} ${operation} ${y} = ${result}`
}
]
};
}
);
// Weather tool with error handling
server.tool(
"getWeather",
{
city: z.string().describe("City name"),
country: z.string().optional().describe("Country code")
},
async ({ city, country }) => {
try {
// Implementation would go here
const weather = await fetchWeather(city, country);
return {
content: [
{
type: "text",
text: `Weather in ${city}: ${weather.description}, ${weather.temperature}°C`
}
]
};
} catch (error) {
console.error(`Weather API error: ${error.message}`);
return {
content: [
{
type: "text",
text: `Error fetching weather: ${error.message}`
}
]
};
}
}
);💡 Note: Tools should have clear parameter validation, good error handling, and return meaningful responses.
MCP supports advanced features like resource templates and dynamic updates:
// Resource template example
server.setRequestHandler(ListResourcesRequestSchema, async () => {
return {
resources: [],
templates: [
{
uriTemplate: "user://{userId}/profile",
name: "User Profile",
description: "Access user profile data"
}
]
};
});
// Resource update notifications
function notifyResourceChanged(uri) {
server.sendNotification("notifications/resources/updated", {
uri
});
}
// Subscribe to resource updates
server.setRequestHandler(SubscribeResourceRequestSchema, async (request) => {
const { uri } = request.params;
// Implementation for subscription
return { success: true };
});Here's a real-world example combining tools and resources:
import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js";
import axios from "axios";
import { z } from "zod";
const server = new McpServer({
name: "WooCommerce Service",
version: "1.0.0"
}, {
capabilities: {
resources: {},
tools: {}
}
});
// Tool: Get Recent Orders
server.tool(
"getRecentOrders",
{
status: z.string().optional()
.describe("Filter orders by status (e.g., processing, completed)"),
limit: z.number().optional()
.describe("Number of orders to return"),
after: z.string().optional()
.describe("Get orders after this date (YYYY-MM-DD)"),
before: z.string().optional()
.describe("Get orders before this date (YYYY-MM-DD)")
},
async ({ status, limit, after, before }) => {
try {
const orders = await fetchOrders({ status, limit, after, before });
return {
content: [
{
type: "text",
text: formatOrders(orders)
}
]
};
} catch (error) {
return {
content: [
{
type: "text",
text: `Error: ${error.message}`
}
]
};
}
}
);
// Resource: Order Statistics
server.setRequestHandler(ListResourcesRequestSchema, async () => {
return {
resources: [
{
uri: "woo://stats/daily",
name: "Daily Order Statistics",
mimeType: "application/json"
}
]
};
});
const transport = new StdioServerTransport();
await server.connect(transport);