Mcp Builder

• Python: Use fastmcp package
• Node/TypeScript: Use @modelcontextprotocol/sdk

Phase 3: Review and Test

1. Test each tool individually: Verify correct behavior with various inputs
2. Test error cases: Ensure graceful handling of invalid inputs, network errors, auth failures
3. Test integration: Verify the server works end-to-end with an MCP client
4. Code review: Check for security issues, proper error handling, type safety

Phase 4: Documentation

1. README: Installation, configuration, available tools
2. Examples: Common usage patterns
3. Configuration: Environment variables, auth setup

Python (FastMCP) Quick Start

from fastmcp import FastMCP

mcp = FastMCP("my-service")

@mcp.tool()
def get_data(query: str, limit: int = 10) -> str:
    """Fetch data from the service.
    
    Args:
        query: Search query string
        limit: Maximum number of results (default: 10)
    """
    # Implementation here
    return results

if __name__ == "__main__":
    mcp.run()

TypeScript (MCP SDK) Quick Start

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-service",
  version: "1.0.0",
});

server.tool(
  "get-data",
  "Fetch data from the service",
  {
    query: z.string().describe("Search query string"),
    limit: z.number().default(10).describe("Max results"),
  },
  async ({ query, limit }) => {
    // Implementation here
    return { content: [{ type: "text", text: JSON.stringify(results) }] };
  }
);

const transport = new StdioServerTransport();
await server.connect(transport);

Best Practices

• Tool naming: Use clear, verb-based names (get-users, create-issue, search-docs)
• Descriptions: Write detailed descriptions — these are the AI's documentation
• Error handling: Return informative error messages, never crash
• Authentication: Use environment variables for secrets, never hardcode
• Rate limiting: Implement appropriate rate limiting for external APIs
• Validation: Validate all inputs before making API calls
• Idempotency: Design tools to be safely retried when possible
• Logging: Add structured logging for debugging

Testing

Use run_command to test MCP servers:

# Python
python -m fastmcp dev my_server.py

# Node/TypeScript
npx tsx my_server.ts

Verify tools work correctly by testing with various inputs and edge cases.