create_generation

Instructions

Implementation Reference

• src/luma_ai_mcp_server/server.py:256-304 (handler)

  The handler function that implements the core logic for the 'create_generation' tool, validating inputs, making API request to Luma, and returning generation details.

  async def create_generation(params: dict) -> str:
      """Create a new generation."""
      if "prompt" not in params:
          raise ValueError("prompt parameter is required")
  
      if "model" in params:
          model = params["model"]
          if isinstance(model, str):
              if model not in [m.value for m in VideoModel]:
                  raise ValueError(f"Invalid model: {model}")
          elif isinstance(model, VideoModel):
              params["model"] = model.value
  
      if "aspect_ratio" in params:
          aspect_ratio = params["aspect_ratio"]
          if isinstance(aspect_ratio, str):
              if aspect_ratio not in [a.value for a in AspectRatio]:
                  raise ValueError(f"Invalid aspect ratio: {aspect_ratio}")
          elif isinstance(aspect_ratio, AspectRatio):
              params["aspect_ratio"] = aspect_ratio.value
  
      if "keyframes" in params:
          keyframes = params["keyframes"]
          if not isinstance(keyframes, dict):
              raise ValueError("keyframes must be an object")
          if not any(key in keyframes for key in ["frame0", "frame1"]):
              raise ValueError("keyframes must contain frame0 or frame1")
  
      input_data = CreateGenerationInput(**params)
      request_data = input_data.model_dump(exclude_none=True)
      response = await _make_luma_request("POST", "/generations", request_data)
  
      if input_data.keyframes:
          output = [
              f"Created advanced generation with ID: {response['id']}",
              f"State: {response['state']}",
          ]
          if "frame0" in input_data.keyframes:
              output.append("starting from an image")
          if "frame1" in input_data.keyframes:
              output.append("ending with an image")
      else:
          output = [
              f"Created text-to-video generation with ID: {response['id']}",
              f"State: {response['state']}",
          ]
  
      return "\n".join(output)

• src/luma_ai_mcp_server/server.py:121-134 (schema)

  Pydantic BaseModel defining the input schema and validation for the create_generation tool parameters.

  class CreateGenerationInput(BaseModel):
      """
      Input parameters for video generation.
      """
  
      prompt: str
      model: VideoModel = VideoModel.RAY_2
      resolution: Optional[Resolution] = None
      duration: Optional[Duration] = None
      aspect_ratio: Optional[AspectRatio] = None
      loop: Optional[bool] = None
      keyframes: Optional[dict] = None
      callback_url: Optional[str] = None

• src/luma_ai_mcp_server/server.py:503-507 (registration)

  Tool registration in the list_tools() function, specifying name, description, and input schema.

  Tool(
      name=LumaTools.CREATE_GENERATION,
      description="Creates a new video generation from text, image, or existing video",
      inputSchema=CreateGenerationInput.model_json_schema(),
  ),

• src/luma_ai_mcp_server/server.py:559-561 (registration)

  Dispatch/case statement in call_tool() that invokes the create_generation handler.

  case LumaTools.CREATE_GENERATION:
      result = await create_generation(arguments)
      return [TextContent(type="text", text=result)]

• src/luma_ai_mcp_server/server.py:94-105 (helper)

  Enum defining tool names, including CREATE_GENERATION constant.

  class LumaTools(str, Enum):
      PING = "ping"
      CREATE_GENERATION = "create_generation"
      GET_GENERATION = "get_generation"
      LIST_GENERATIONS = "list_generations"
      DELETE_GENERATION = "delete_generation"
      UPSCALE_GENERATION = "upscale_generation"
      ADD_AUDIO = "add_audio"
      GENERATE_IMAGE = "generate_image"
      GET_CREDITS = "get_credits"
      GET_CAMERA_MOTIONS = "get_camera_motions"