async def overwrite_refactored_search_results(callback_context: CallbackContext) -> Optional[types.Content]: """ Owerwrite the response from the refactored_search_agent. :param callback_context: the context :return: a confirmation message that the search was successful or not. """ # Get the name of the agent calling this callback agent_name = callback_context.agent_name print(f"[Callback] After model call for agent: {agent_name}") # List the existing state variables for debugging print(f"[Callback] Existing state variables: {callback_context.state.to_dict().keys()}") # check if the 'refactored_search_results' variable can be found in the state if 'refactored_search_results' in callback_context.state and 'search_results' in callback_context.state: return types.Content( parts=[types.Part(text="The original search results and the refactored search results have been preserved")], role="model" ) else: return types.Content( parts=[types.Part(text="Sorry, I was not able to retrieve anything intersting from the Web.")], role="model" ) class informationChunk(BaseModel): companies: List[str] = Field(description="The companies mentioned in the news") news: str = Field(description="The news content") date: str = Field(description="The date of the news") origin: str = Field(description="The origin of the news") class informationChunks(BaseModel): informations: List[informationChunk] = Field(description="List of information chunks") date: str = Field(description="Today's date") refactor_search_agent = LlmAgent( model='gemini-2.5-flash', name='refactor_search_agent', description='A helpful assistant for refactoring the information found on the web following instructions received from the user', instruction=REFACTOR_SEARCH_PROMPT, output_schema=informationChunks, output_key='refactored_search_results', after_agent_callback=overwrite_refactored_search_results # type: ignore )