A programming framework for agentic AI
Add last_speaker to GroupChatManager (#3318)
Li Jiang1971f85ded0e
diff --git a/autogen/agentchat/groupchat.py b/autogen/agentchat/groupchat.pyindex 48f11d526..bcb8c30e2 100644--- a/autogen/agentchat/groupchat.py+++ b/autogen/agentchat/groupchat.py@@ -965,6 +965,7 @@ class GroupChatManager(ConversableAgent):# Store groupchatself._groupchat = groupchat+ self._last_speaker = Noneself._silent = silent# Order of register_reply is important.@@ -1006,6 +1007,52 @@ class GroupChatManager(ConversableAgent):if (recipient != agent or prepare_recipient) and isinstance(agent, ConversableAgent):agent._prepare_chat(self, clear_history, False, reply_at_receive)+ @property+ def last_speaker(self) -> Agent:+ """Return the agent who sent the last message to group chat manager.++ In a group chat, an agent will always send a message to the group chat manager, and the group chat manager will+ send the message to all other agents in the group chat. So, when an agent receives a message, it will always be+ from the group chat manager. With this property, the agent receiving the message can know who actually sent the+ message.++ Example:+ ```python+ from autogen import ConversableAgent+ from autogen import GroupChat, GroupChatManager+++ def print_messages(recipient, messages, sender, config):+ # Print the message immediately+ print(+ f"Sender: {sender.name} | Recipient: {recipient.name} | Message: {messages[-1].get('content')}"+ )+ print(f"Real Sender: {sender.last_speaker.name}")+ assert sender.last_speaker.name in messages[-1].get("content")+ return False, None # Required to ensure the agent communication flow continues+++ agent_a = ConversableAgent("agent A", default_auto_reply="I'm agent A.")+ agent_b = ConversableAgent("agent B", default_auto_reply="I'm agent B.")+ agent_c = ConversableAgent("agent C", default_auto_reply="I'm agent C.")+ for agent in [agent_a, agent_b, agent_c]:+ agent.register_reply(+ [ConversableAgent, None], reply_func=print_messages, config=None+ )+ group_chat = GroupChat(+ [agent_a, agent_b, agent_c],+ messages=[],+ max_round=6,+ speaker_selection_method="random",+ allow_repeat_speaker=True,+ )+ chat_manager = GroupChatManager(group_chat)+ groupchat_result = agent_a.initiate_chat(+ chat_manager, message="Hi, there, I'm agent A."+ )+ """+ return self._last_speaker+def run_chat(self,messages: Optional[List[Dict]] = None,@@ -1034,6 +1081,7 @@ class GroupChatManager(ConversableAgent):a.previous_cache = a.client_cachea.client_cache = self.client_cachefor i in range(groupchat.max_round):+ self._last_speaker = speakergroupchat.append(message, speaker)# broadcast the message to all agents except the speakerfor agent in groupchat.agents: