Building a Minimal MCP Server

Model Context Protocol lets agents plug into your tools. A 40-line server exposes a real capability to Claude.

55 min · Reviewed 2026

One Protocol, Many Tools

An MCP server exposes tools over stdio or HTTP. Any MCP-capable client (Claude Desktop, Cursor, Claude Code) can call them. Build once, use everywhere.

npm install @modelcontextprotocol/sdk zodOfficial SDK plus Zod for schemas.

// server.ts
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: "word-tools",
  version: "0.1.0",
});

server.tool(
  "word_count",
  "Count words in a string.",
  { text: z.string().min(1) },
  async ({ text }) => {
    const count = text.trim().split(/\s+/).length;
    return {
      content: [{ type: "text", text: String(count) }],
    };
  }
);

server.tool(
  "reverse",
  "Reverse a string.",
  { text: z.string() },
  async ({ text }) => ({
    content: [{ type: "text", text: [...text].reverse().join("") }],
  })
);

const transport = new StdioServerTransport();
await server.connect(transport);Two tools, typed inputs, string outputs. Run with `node server.js` from an MCP-capable client config.

// claude_desktop_config.json snippet
{
  "mcpServers": {
    "word-tools": {
      "command": "node",
      "args": ["/absolute/path/to/server.js"]
    }
  }
}Point Claude Desktop at your script. Restart, and the tools appear.

Understanding "Building a Minimal MCP Server" in practice: AI-assisted coding shifts work from syntax recall to design thinking — models handle boilerplate so you focus on architecture. Model Context Protocol lets agents plug into your tools. A 40-line server exposes a real capability to Claude — and knowing how to apply this gives you a concrete advantage.

Apply MCP in your ai-coding workflow to get better results
Apply tools in your ai-coding workflow to get better results
Apply stdio transport in your ai-coding workflow to get better results
Apply schema in your ai-coding workflow to get better results

Use AI to generate unit tests for an existing function
Ask AI to refactor a messy function and explain the changes
Have AI suggest a code review for a recent pull request

The big idea: MCP is the USB port for agents. Write a tiny server once, and every MCP client can use your tools.

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-progx-mcp-server-creators

What does MCP stand for in the context of AI agent tool integration?
1. Modular Code Processor
2. Message Control Pipeline
3. Model Context Protocol
4. Machine Communication Port
Which transport mechanism uses stdout to send JSON-RPC messages?
1. HTTP transport
2. WebSocket transport
3. stdio transport
4. stdio transport
A developer adds a console.log() statement to their MCP server for debugging. What problem does this cause when using stdio transport?
1. The server will run slower due to extra processing
2. The server will crash immediately
3. The JSON-RPC message stream becomes corrupted
4. The client will receive duplicate messages
Where should a developer send log messages in an MCP server using stdio transport?
1. stderr
2. The client console
3. A log file
4. stdout
What analogy does the lesson use to describe MCP?
1. A firewall for tool access
2. A highway for data transfer
3. A bridge between AI models and databases
4. The USB port for agents
Which three clients are mentioned as examples of MCP-capable clients in the lesson?
1. Chrome, Firefox, and Safari
2. Visual Studio, IntelliJ, and Atom
3. Python, Node.js, and Rust
4. Claude Desktop, Cursor, and Claude Code
What is JSON-RPC in the context of MCP servers?
1. A remote procedure call protocol using JSON format
2. A web framework for building HTTP servers
3. A programming language for writing MCP servers
4. A data storage format for tool configurations
What is the primary benefit of MCP's standardization described in the lesson?
1. Faster tool execution speed
2. Automatic tool optimization
3. Built-in security features
4. Write once, use across multiple MCP clients
Why does stdout need to be reserved exclusively for JSON-RPC messages in stdio transport?
1. To enable faster communication speeds
2. To support multiple concurrent clients
3. Because the JSON-RPC protocol requires clean message boundaries
4. For security and authentication
A student builds an MCP server with exactly three tools. Based on the lesson, why might this be better than a version with ten tools?
1. Three tools are easier to write than ten tools
2. Three tools run significantly faster on the server
3. The server will use less memory with three tools
4. The model will make better tool selection decisions with fewer options
What happens when an MCP-capable client connects to an MCP server?
1. The client automatically downloads and installs the server
2. The server sends a copy of its code to the client
3. The client can call the tools exposed by the server
4. The client compiles the server's code for local execution
What is a key characteristic of tools exposed by an MCP server using stdio transport?
1. They are stored in a cloud database
2. They are accessed over a network connection
3. They require a web browser to access
4. They communicate using JSON-RPC messages over stdout
What problem does the lesson identify with building an MCP server that exposes every possible tool?
1. The tools become less accurate
2. The server takes too long to start up
3. The model freezes on choice due to too many options
4. The server becomes too complex to maintain
What distinguishes focused tools from a comprehensive tool collection in MCP server design?
1. Focused tools are faster to execute
2. Focused tools are more secure
3. Focused tools use less memory
4. Focused tools help the model make clearer selection decisions
A developer wants to make their custom tool available to multiple different AI assistants. Based on the lesson, what is the most efficient approach?
1. Use a web API that each assistant must program against
2. Create an MCP server that exposes the tool
3. Build a separate integration for each AI assistant
4. Wait for each AI assistant to add native support

← Back to interactive lesson

Tendril · Creators · AI-Assisted Coding

Building a Minimal MCP Server

Model Context Protocol lets agents plug into your tools. A 40-line server exposes a real capability to Claude.

55 min · Reviewed 2026

One Protocol, Many Tools

An MCP server exposes tools over stdio or HTTP. Any MCP-capable client (Claude Desktop, Cursor, Claude Code) can call them. Build once, use everywhere.

npm install @modelcontextprotocol/sdk zodOfficial SDK plus Zod for schemas.

// server.ts
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: "word-tools",
  version: "0.1.0",
});

server.tool(
  "word_count",
  "Count words in a string.",
  { text: z.string().min(1) },
  async ({ text }) => {
    const count = text.trim().split(/\s+/).length;
    return {
      content: [{ type: "text", text: String(count) }],
    };
  }
);

server.tool(
  "reverse",
  "Reverse a string.",
  { text: z.string() },
  async ({ text }) => ({
    content: [{ type: "text", text: [...text].reverse().join("") }],
  })
);

const transport = new StdioServerTransport();
await server.connect(transport);Two tools, typed inputs, string outputs. Run with `node server.js` from an MCP-capable client config.

// claude_desktop_config.json snippet
{
  "mcpServers": {
    "word-tools": {
      "command": "node",
      "args": ["/absolute/path/to/server.js"]
    }
  }
}Point Claude Desktop at your script. Restart, and the tools appear.

Apply MCP in your ai-coding workflow to get better results
Apply tools in your ai-coding workflow to get better results
Apply stdio transport in your ai-coding workflow to get better results
Apply schema in your ai-coding workflow to get better results

Use AI to generate unit tests for an existing function
Ask AI to refactor a messy function and explain the changes
Have AI suggest a code review for a recent pull request

The big idea: MCP is the USB port for agents. Write a tiny server once, and every MCP client can use your tools.

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-progx-mcp-server-creators

What does MCP stand for in the context of AI agent tool integration?
1. Modular Code Processor
2. Message Control Pipeline
3. Model Context Protocol
4. Machine Communication Port
Which transport mechanism uses stdout to send JSON-RPC messages?
1. HTTP transport
2. WebSocket transport
3. stdio transport
4. stdio transport
A developer adds a console.log() statement to their MCP server for debugging. What problem does this cause when using stdio transport?
1. The server will run slower due to extra processing
2. The server will crash immediately
3. The JSON-RPC message stream becomes corrupted
4. The client will receive duplicate messages
Where should a developer send log messages in an MCP server using stdio transport?
1. stderr
2. The client console
3. A log file
4. stdout
What analogy does the lesson use to describe MCP?
1. A firewall for tool access
2. A highway for data transfer
3. A bridge between AI models and databases
4. The USB port for agents
Which three clients are mentioned as examples of MCP-capable clients in the lesson?
1. Chrome, Firefox, and Safari
2. Visual Studio, IntelliJ, and Atom
3. Python, Node.js, and Rust
4. Claude Desktop, Cursor, and Claude Code
What is JSON-RPC in the context of MCP servers?
1. A remote procedure call protocol using JSON format
2. A web framework for building HTTP servers
3. A programming language for writing MCP servers
4. A data storage format for tool configurations
What is the primary benefit of MCP's standardization described in the lesson?
1. Faster tool execution speed
2. Automatic tool optimization
3. Built-in security features
4. Write once, use across multiple MCP clients
Why does stdout need to be reserved exclusively for JSON-RPC messages in stdio transport?
1. To enable faster communication speeds
2. To support multiple concurrent clients
3. Because the JSON-RPC protocol requires clean message boundaries
4. For security and authentication
A student builds an MCP server with exactly three tools. Based on the lesson, why might this be better than a version with ten tools?
1. Three tools are easier to write than ten tools
2. Three tools run significantly faster on the server
3. The server will use less memory with three tools
4. The model will make better tool selection decisions with fewer options
What happens when an MCP-capable client connects to an MCP server?
1. The client automatically downloads and installs the server
2. The server sends a copy of its code to the client
3. The client can call the tools exposed by the server
4. The client compiles the server's code for local execution
What is a key characteristic of tools exposed by an MCP server using stdio transport?
1. They are stored in a cloud database
2. They are accessed over a network connection
3. They require a web browser to access
4. They communicate using JSON-RPC messages over stdout
What problem does the lesson identify with building an MCP server that exposes every possible tool?
1. The tools become less accurate
2. The server takes too long to start up
3. The model freezes on choice due to too many options
4. The server becomes too complex to maintain
What distinguishes focused tools from a comprehensive tool collection in MCP server design?
1. Focused tools are faster to execute
2. Focused tools are more secure
3. Focused tools use less memory
4. Focused tools help the model make clearer selection decisions
A developer wants to make their custom tool available to multiple different AI assistants. Based on the lesson, what is the most efficient approach?
1. Use a web API that each assistant must program against
2. Create an MCP server that exposes the tool
3. Build a separate integration for each AI assistant
4. Wait for each AI assistant to add native support

← Back to interactive lesson