How to restart the agent after getting new input from user

[easontsai]

I am running Autogen on python application, and I am calling the agent via a Rest API.
I want the Agent to pause when user intervention is needed (here I intend to use UserAgent as the stop condition). However, I encountered a problem. After the user inputs a new question(via Rest API), how can I get the Agent to resume from where it left off? (Is it possible to maintain the runtime state and continue execution from there?)

class UserAgent(RoutedAgent):
    def __init__(self, description: str, group_chat_topic_type: str) -> None:
        super().__init__(description=description)
        self._group_chat_topic_type = group_chat_topic_type

    @message_handler
    async def handle_message(self, message: GroupChatMessage, ctx: MessageContext) -> None:
        # When integrating with a frontend, this is where group chat message would be sent to the frontend.
        pass

    @message_handler
    async def handle_request_to_speak(self, message: RequestToSpeak, ctx: MessageContext) -> None:
        user_input = input("Enter your message, type 'APPROVE' to conclude the task: ")
        Console().print(Markdown(f"### User: \n{user_input}"))
        await self.publish_message(
            GroupChatMessage(body=UserMessage(content=user_input, source=self.id.type)),
            DefaultTopicId(type=self._group_chat_topic_type),
        )

[ekzhu]

Is it possible to maintain the runtime state

As long as the runtime is stored in your user session or in memory of the server, its state is persisted. You can also use save_state and load_state to use disk-based persistence.

continue execution from there

You can use runtime.publish_message() to publish a message directly to a topic that the UserAgent subscribed to when you get a POST request from the UI sent by the user.

For example:

# In your FastAPI handler or ChainLit handler for user input
await runtime.publish_message(UserInputMessage(content=...), TopicId(user_agent_topic_type, "<user_session_id>")

Your UserAgent should be subscribed to the user_agent_topic_type and has a handler for the UserInputMessage message type.

@type_subscription(user_agent_topic_type)
class UserAgent(RoutedAgent):
    def __init__(self, description: str, group_chat_topic_type: str) -> None:
        super().__init__(description=description)
        self._group_chat_topic_type = group_chat_topic_type

    @message_handler
    async def handle_message(self, message: GroupChatMessage, ctx: MessageContext) -> None:
        # When integrating with a frontend, this is where group chat message would be sent to the frontend.
        pass

    @message_handler
    async def handle_request_to_speak(self, message: RequestToSpeak, ctx: MessageContext) -> None:
        # Send a message to the UI to prompt the user to provide an input and returns
        pass
    
    @message_handler
    async def handle_user_input(self, message: UserInputMessage, ctx: MessageContext) -> None:
        # Publish the message content to the group chat.
        await self.publish_message(GroupChatMessage(content="...", source="..."), DefaultTopicId(self._group_chat_topic_type))

[ekzhu]

@easontsai if you found this useful, could you contribute to #5345 to help others?