MCP Data Analyzer

Instructions

Implementation Reference

• src/mcp_data_analyzer/server.py:174-198 (handler)

  The core handler function in ScriptRunner class that implements the load_file tool logic: loads CSV or XLSX files into a pandas DataFrame, stores it in self.data with a given or auto-generated name, logs notes, and returns success or error TextContent.

  def load_file(self, file_path: str, df_name:str = None, sheet_name: Optional[str] = None):
      self.df_count += 1
      if not df_name:
          df_name = f"df_{self.df_count}"
      try:
          file_extension = os.path.splitext(file_path)[1].lower()
          if file_extension == ".csv":
              self.data[df_name] = pd.read_csv(file_path)
              self.notes.append(f"Successfully loaded CSV into dataframe '{df_name}' from '{file_path}'")
          elif file_extension == ".xlsx":
              self.data[df_name] = pd.read_excel(file_path, sheet_name=sheet_name)
              self.notes.append(f"Successfully loaded XLSX into dataframe '{df_name}' from '{file_path}' (sheet: {sheet_name or 'first'})")
          else:
              raise ValueError(f"Unsupported file type: {file_extension}. Only .csv and .xlsx are supported.")
          
          return [
              TextContent(type="text", text=f"Successfully loaded data into dataframe '{df_name}'")
          ]
      except Exception as e:
          error_message = f"Error loading file: {str(e)}"
          self.notes.append(f"ERROR: {error_message}")
          return [
              TextContent(type="text", text=f"Error: {error_message}")
          ]

• src/mcp_data_analyzer/server.py:141-144 (schema)

  Pydantic schema (BaseModel) defining the input parameters for the load_file tool: file_path (required), optional df_name and sheet_name.

  class LoadFile(BaseModel):
      file_path: str
      df_name: Optional[str] = None
      sheet_name: Optional[str] = None

• src/mcp_data_analyzer/server.py:299-299 (registration)

  Registration of the load_file tool in the MCP server's list_tools handler, providing name, description, and input schema.

  Tool(name=DataExplorationTools.LOAD_FILE, description=LOAD_FILE_TOOL_DESCRIPTION, inputSchema=LoadFile.model_json_schema()),

• src/mcp_data_analyzer/server.py:305-306 (registration)

  Dispatch logic in the MCP server's call_tool handler that routes calls to the load_file tool to the ScriptRunner instance's load_file method.

  if name == DataExplorationTools.LOAD_FILE:
      return script_runner.load_file(arguments.get("file_path"), arguments.get("df_name"), arguments.get("sheet_name"))

• src/mcp_data_analyzer/server.py:125-126 (helper)

  Enum defining the tool names, including LOAD_FILE = 'load_file' used in registration and dispatch.

  class DataExplorationTools(str, Enum):
      LOAD_FILE = "load_file"