Math MCP Server

Instructions

Implementation Reference

• src/math_server.py:38-56 (handler)

  The handler function for the 'add' MCP tool. It sums the provided list of numbers using NumPy's sum with 64-bit float precision, rounds to the precision limit, logs the operation, and returns the result. Raises ValueError for empty lists.

  @math_mcp.tool
  def add(
      numbers: list[int | float],
  ) -> float:
      """Adds a list of positive and/or negative numbers with 64 bit floating point precision and returns a 64 bit float.
         You need to provide them in the format of a list. For example, [1, 2, 3] would return 6.0.
         You can also use fractions if you want to, like [1/2, 1/3, 1/4].
      """
  
      # This is technically allowed by Fast MCP, but it is an error here
      if not numbers:
          logging.error("Received an empty list for addition.")
          raise ValueError("""The list of numbers cannot be empty. Try wrapping the numbers in brackets, like [1, 2, 3], if this is not the case.
          """)
  
      # Use numpy for fast addition
      result = np.round(np.sum(numbers, dtype=np.float64), decimals=SIXTY_FOUR_BIT_FLOAT_DECIMAL_PLACES)
      logging.info(f"Adding numbers: {numbers} -> Result: {result}")
      return result

• src/math_server.py:39-45 (schema)

  Schema definition via type annotations (input: list[int | float], output: float) and docstring describing usage and examples for the 'add' tool.

  def add(
      numbers: list[int | float],
  ) -> float:
      """Adds a list of positive and/or negative numbers with 64 bit floating point precision and returns a 64 bit float.
         You need to provide them in the format of a list. For example, [1, 2, 3] would return 6.0.
         You can also use fractions if you want to, like [1/2, 1/3, 1/4].
      """

• src/math_server.py:38-38 (registration)

  Registration of the 'add' tool to the FastMCP server instance 'math_mcp' using the @tool decorator.

  @math_mcp.tool