MCP Mix Server

Instructions

Implementation Reference

• mcp_server/tools/parquet_tools.py:4-16 (handler)

  The main tool handler function for 'summarize_parquet_file'. Decorated with @mcp.tool() for automatic registration upon module import. Delegates the logic to the read_parquet_summary helper.

  @mcp.tool()
  def summarize_parquet_file(filename: str) -> str:
      """
      Summarise a Parquet file by reporting its number of rows and columns.
  
      Args:
          filename (str): Name of the Parquet file in the /data directory.
  
      Returns:
          str: A string describing the file's dimensions.
      """
      
      return read_parquet_summary(filename)

• mcp_server/main.py:3-6 (registration)

  Imports the parquet_tools module (and csv_tools), which triggers registration of the decorated tool functions before starting the MCP server with mcp.run().

  # import and register tools decorated in tools.py
  # before running mcp.run()
  import tools.csv_tools
  import tools.parquet_tools

• mcp_server/utils/file_reader.py:24-40 (helper)

  Supporting utility function that loads the Parquet file using pandas.read_parquet and computes the summary of rows and columns. Handles missing file gracefully.

  def read_parquet_summary(filename: str) -> str:
      """
      Read a Parquet file and return a simple summary.
  
      Args:
          filename (str): Name of the Parquet file
  
      Returns:
          str: A string describing the file's contents.
      """
      file_path = DATA_DIR / filename
      if not file_path.exists():
          return f"File {filename} does not exist."
  
      df = pd.read_parquet(file_path)
      
      return f"Parquet file '{filename}' has {len(df)} rows and {len(df.columns)} columns."