Multiple Tools

One server, many capabilities

Nothing stops a server from exposing more than one tool. Each @mcp.tool() call registers another entry on the same mcp object.

mcp = FastMCP("kitchen-server")


@mcp.tool()
def convert_celsius_to_fahrenheit(celsius: float) -> float:
    """Convert a Celsius temperature to Fahrenheit."""
    return celsius * 9 / 5 + 32


@mcp.tool()
def convert_cups_to_milliliters(cups: float) -> float:
    """Convert a volume in US cups to milliliters."""
    return cups * 236.588

tools/list now returns both. A client, or eventually an AI deciding which one to use, sees two distinct entries, each with its own name, description, and schema. Exactly like multiple-tool LangChain lessons, just server-hosted.

Naming matters more here

In LangChain, tool names only needed to be unique within one Python process. An MCP server might eventually run alongside other servers in the same host (Lesson 17 covers this directly), so vague names like convert or get become a real problem: which server's convert did the AI mean? Prefer specific, descriptive names, convert_celsius_to_fahrenheit rather than convert, the same way the official docs name their example tools get_alerts and get_forecast rather than just alerts and forecast.