Example of existing API exposed as an MCP Server using annotations

README.md

MCP Server + Client Example (leveraging existing REST APIs)

To expose existing Python code as an MCP server, use fastMCP's decorators to annotate functions as tools or resources. This involves:

• Wrapping existing functions with @mcp.tool for executable actions or @mcp.resource for data access.
• Adding type hints and docstrings for automatic schema generation.
• Handling context if needed (e.g., for logging or nested calls).

Assume existing code in existing_math.py with functions subtract and get_pi. (see file)

To expose functions subtract and get_pi as MCP server:

1. Import and annotate in a new server file annotated_mcp_server.py. (see file)
2. Run: python annotated_mcp_server.py. Now, clients can call subtract_tool or read math://constants/pi.

This approach keeps original code unchanged while exposing it via MCP.

annotated_mcp_server.py

# Filename: annotated_mcp_server.py
from fastmcp import FastMCP, Context
from existing_math import subtract, get_pi  # Import existing functions

mcp = FastMCP(name="Annotated Math Server", description="Exposes existing math functions as MCP tools and resources")

# Annotate existing subtract function as a tool
@mcp.tool
def subtract_tool(x: float, y: float, ctx: Context = None) -> float:
    """Subtract y from x using existing code. Logs via context if available."""
    if ctx:
        ctx.info(f"Subtracting {y} from {x}")
    return subtract(x, y)  # Call existing function

# Annotate existing get_pi as a resource
@mcp.resource("math://constants/pi")
def pi_resource():
    """Return the value of pi from existing code."""
    return str(get_pi())  # Wrap existing function

if __name__ == "__main__":
    mcp.run(transport="http", host="127.0.0.1", port=8001, path="/mcp")

existing_math.py

# Existing code in: existing_math.py
import math

def subtract(x: float, y: float) -> float:
    return x - y

def get_pi() -> float:
    return math.pi