SOP-005: Building Your First MCP Server

Fresh

Source: Introduction to Model Context Protocol

Document Control

Field	Value
SOP ID	SOP-005
Version	1.0
Status	Active
Last Updated	2024-12

Overview

Build an MCP (Model Context Protocol) server to expose tools, resources, and prompts to Claude.

Prerequisites

• Python 3.10+
• Basic async/await knowledge
• Understanding of API concepts

MCP Architecture

Step 1: Install MCP SDK

pip install mcp

Step 2: Create Basic Server

# server.py
from mcp.server import Server
from mcp.server.stdio import stdio_server
from mcp.types import Tool, TextContent

# Create server instance
server = Server("my-mcp-server")

# Define a tool
@server.list_tools()
async def list_tools():
    return [
        Tool(
            name="hello",
            description="Say hello to someone",
            inputSchema={
                "type": "object",
                "properties": {
                    "name": {
                        "type": "string",
                        "description": "Name to greet"
                    }
                },
                "required": ["name"]
            }
        )
    ]

# Implement the tool
@server.call_tool()
async def call_tool(name: str, arguments: dict):
    if name == "hello":
        return [TextContent(
            type="text",
            text=f"Hello, {arguments['name']}!"
        )]

# Run the server
async def main():
    async with stdio_server() as (read, write):
        await server.run(read, write)

if __name__ == "__main__":
    import asyncio
    asyncio.run(main())

Step 3: Test with MCP Inspector

# Install inspector
npx @anthropic-ai/mcp-inspector

# Run your server
python server.py

MCP Primitives

Primitive	Control	Description	Examples
Tools	Model-controlled	Actions/functions the model can call	Search database, Send email, Create file
Resources	Application-controlled	Data accessed via URIs	config://settings, db://users/123
Prompts	User-controlled	Pre-defined templates/workflows	Summarize document, Generate tests

Step 4: Add to Claude Desktop

Edit claude_desktop_config.json:

{
  "mcpServers": {
    "my-server": {
      "command": "python",
      "args": ["/path/to/server.py"]
    }
  }
}

Verification Checklist

• [ ] MCP SDK installed
• [ ] Server runs without errors
• [ ] Tools listed in inspector
• [ ] Tool execution returns expected result
• [ ] Server connects to Claude

Troubleshooting

Issue	Solution
Import error	Check mcp package installed
Connection refused	Check server is running
Tool not found	Verify tool name matches
Timeout	Check async handling

See Also

• SOP-006: MCP Transports
• Workflow-005: MCP Integration