Instructions

Implementation Reference

• src/mcp_server_baidu_digitalhuman/dhserver.py:483-506 (handler)

  The MCP tool handler function that executes the logic: obtains DH API client and calls the underlying get_lite2d_general_status method.

  async def getLite2dGeneralStatus(
          figureId: Annotated[str, Field(description="人像ID", default=None)],
          systemFigure: Annotated[Optional[bool], \
                  Field(description="是否是平台公共人像，true返回平台公共人像，false返回定制人像，空查询全部", default=None)],
          trainSuccess: Annotated[bool, \
                  Field(description="是否查询训练完成:true：只返回可用人像,false：只返回排队中、训练中或训练失败的定制人像,为空不进行过滤",
                        default=None)]) \
          -> MCPLite2DStatusResponse:
      """
      Retrieve the status of a lite2d General video via the DH API.
  
      Args:
          figureId: 人像ID
  
      Returns:
          任务状态
      """
      try:
          client = await getDhClient()
          ret = await client.get_lite2d_general_status(figureId, systemFigure, trainSuccess)
          return ret
      except Exception as e:
          return MCPVideoStatusResponse(error=str(e))

• src/mcp_server_baidu_digitalhuman/dhserver.py:463-482 (registration)

  MCP tool registration with name, description, and input schema defined via function parameters.

  @mcp.tool(
      name="getLite2dGeneralStatus",
      description=(
      """
  #工具说明：根据2D小样本数字人对应的人像ID，查询该任务目前的状态，也可以用于查询有哪些可用的2D人像。
  # 样例1：
  用户输入：查一下id为xxx的数字人好了没有。
  思考过程:
  1.用户想要查询人像生成任务的状态，需要使用“getLite2dGeneralStatus”工具。
  2.工具需要，figureId，systemFigure，trainSuccess，pageNo，ppageSize这些参数。
  3.用户提到了ID为xxx，所以figureId的值为xxx，现在不清楚这个任务的状态，所以trainSuccess的值不需要填，系统人像不需要生成过程，所以systemFigure值为false，其他为默认值。
  # 样例2：
  用户输入：我可以用哪些人像
  思考过程:
  1.用户想要查询哪些人像ID可以使用，需要使用“getLite2dGeneralStatus”工具。
  2.工具需要，figureId，systemFigure，trainSuccess，pageNo，ppageSize这些参数。
  3.查询可用人像，所以figureId为空，syste
  Figure为空，trainSuccess为ture，pageNo默认为1，避免漏查pageSize为最大值100。
      """)
  )

• src/mcp_server_baidu_digitalhuman/dhapi.py:314-328 (schema)

  Pydantic model defining the output schema for the tool response.

  class MCPLite2DStatusResponse(BaseDHResponse):
      """ MCP 2D小样本数字人状态响应 """
      figureId: Optional[str] = None
      name: Optional[str] = None
      customizeType: Optional[str] = None
      systemFigure: Optional[bool] = None
      keepBackground: Optional[bool] = None
      gender: Optional[str] = None
      resolutionWidth: Optional[int] = None
      resolutionHeight: Optional[int] = None
      templateVideoUrl: Optional[str] = None
      maskVideoUrl: Optional[str] = None
      status: Optional[str] = None
      failedMessage: Optional[str] = None

• src/mcp_server_baidu_digitalhuman/dhapi.py:758-779 (helper)

  Underlying DH API client helper method that performs the actual API call to query lite2d general status.

  async def get_lite2d_general_status(self, figureId: str, systemFigure: bool = None, trainSuccess: bool = None) \
          -> MCPLite2DStatusResponse:
      """ 2D 小样本数字人生成状态 """
      async def api_call():
          # fix in the future: systemFigure, trainSuccess
          endpoint = f"api/digitalhuman/open/v1/figure/lite2d/query?figureId={figureId}"
          return await self._make_request(endpoint)
  
      ret = await self._handle_api_request(
          api_call=api_call,
          response_model_class=Lite2dStatusResponse,
          mcp_response_class=MCPLite2DStatusResponse,
          error_msg="No video generation data returned.",
          figureId=lambda d: d.result[0].figureId,
          name=lambda d: d.result[0].name,
          status=lambda d: d.result[0].status,
          templateVideoUrl=lambda d: d.result[0].templateVideoUrl,
          failedMessage=lambda d: d.result[0].failedMessage,
          systemFigure=lambda d: d.result[0].systemFigure,
          gender=lambda d: d.result[0].gender,
      )
      return ret