teslamate-mcp

Instructions

Implementation Reference

• src/tools.py:62-66 (registration)

  ToolDefinition registering the 'get_drive_summary_per_day' tool with its description and associated SQL query file.
  ToolDefinition( name="get_drive_summary_per_day", description="Get the drive summary per day for each car. Provides daily driving statistics including distance, duration, and efficiency.", sql_file="drive_summary_per_day.sql", ),
• main.py:22-28 (handler)

  Handler factory that creates the execution function for the tool, which runs the SQL query from the specified file synchronously.
  def create_tool_handler(sql_file: str): """Factory function to create tool handlers""" def handler() -> List[Dict[str, Any]]: return db_manager.execute_query_sync(sql_file) return handler
• main.py:31-39 (registration)

  Dynamic registration loop that creates and registers the handler for each tool, including 'get_drive_summary_per_day', with the MCP server.
  # Register all tools from definitions for tool_def in TOOL_DEFINITIONS: tool_func = create_tool_handler(tool_def.sql_file) tool_func.__doc__ = tool_def.description tool_func.__name__ = tool_def.name # Register the tool with the MCP server mcp.tool()(tool_func)
• main_remote.py:118-127 (handler)

  Async handler function for executing predefined tools by name, including 'get_drive_summary_per_day', using the async database query execution.
  async def execute_predefined_tool(tool_name: str) -> List[Dict[str, Any]]: """Execute a predefined tool by name""" if not app_context: raise RuntimeError("Application context not initialized") tool = get_tool_by_name(tool_name) return await app_context.db_manager.execute_query_async( tool.sql_file, app_context.db_pool )
• main_remote.py:178-186 (registration)

  Registration of all predefined tools, including schema (empty input), in the list_tools MCP method for the remote HTTP server.
  # Add all predefined tools for tool_def in TOOL_DEFINITIONS: tools.append( types.Tool( name=tool_def.name, description=tool_def.description, inputSchema={"type": "object", "properties": {}}, ) )