Your first MCP server (Python)

The Python MCP server is the equivalent of the Node version. Same pattern; different language.

The boilerplate

pip install mcp

A skeleton:

from mcp.server import Server
from mcp.server.stdio import stdio_server

app = Server("my-server")

# Tool definitions

async def main():
    async with stdio_server() as (read_stream, write_stream):
        await app.run(read_stream, write_stream, app.create_initialization_options())

Tool definition

from mcp.types import Tool, TextContent

@app.list_tools()
async def list_tools() -> list[Tool]:
    return [
        Tool(
            name="get_weather",
            description="Get current weather for a city",
            inputSchema={
                "type": "object",
                "properties": {
                    "city": {"type": "string"}
                },
                "required": ["city"]
            }
        )
    ]

@app.call_tool()
async def call_tool(name: str, arguments: dict):
    if name == "get_weather":
        weather = await fetch_weather(arguments["city"])
        return [TextContent(type="text", text=weather)]

Reviewer ritual

Same discipline as the Node version:

• Tool naming.
• Schema precision.
• Error handling.
• Tests.

When Python wins

Python over Node when:

• Existing tooling is in Python (data science, ML, scripting).
• Server logic involves heavy data processing.
• Integration with Python-native services (numpy, pandas, etc.).

Trade-offs

• Python: more verbose, ecosystem advantage in some domains.
• Node: lighter weight, faster cold start, ecosystem advantage in other domains.

The choice often comes down to team familiarity.

A real server

A team's analytics MCP server in Python:

• Tools for querying internal data warehouse.
• Schema validation via pydantic.
• Tests via pytest.
• 400 lines.

Used by the team's AI assistants for ad-hoc analytics queries.

What we won't do

Default to Node when the team is Python-native (or vice versa).

Skipping the schema validation in either language.

Close

The Python MCP server is the Node server's equivalent. Same protocol; same disciplines. The choice is language preference; the engineering is the same.

Related reading

• Your first MCP server (Node) — Node equivalent.
• MCP tool schemas — schema discipline.
• Tool design like APIs — surrounding discipline.

We build AI-enabled software and help businesses put AI to work. If you're building MCP servers in Python, we'd love to hear about it. Get in touch.