Platform for stateful agents: AI with advanced memory that can learn and self-improve over time.
fix: summarization includes tool call message before truncation (#2084)
Andy Lia45739444fe1
diff --git a/letta/llm_api/helpers.py b/letta/llm_api/helpers.pyindex 6d6f77c13..ed497e09f 100644--- a/letta/llm_api/helpers.py+++ b/letta/llm_api/helpers.py@@ -337,6 +337,10 @@ def calculate_summarizer_cutoff(in_context_messages: List[Message], token_counts)break+ # includes the tool response to be summarized after a tool call so we don't have any hanging tool calls after trimming.+ if i + 1 < len(in_context_messages_openai) and in_context_messages_openai[i + 1]["role"] == "tool":+ cutoff += 1+logger.info(f"Evicting {cutoff}/{len(in_context_messages)} messages...")return cutoff + 1diff --git a/letta/services/agent_manager.py b/letta/services/agent_manager.pyindex 96b891ddd..190ca8a81 100644--- a/letta/services/agent_manager.py+++ b/letta/services/agent_manager.py@@ -866,9 +866,8 @@ class AgentManager:@enforce_typesdef trim_older_in_context_messages(self, num: int, agent_id: str, actor: PydanticUser) -> PydanticAgentState:message_ids = self.get_agent_by_id(agent_id=agent_id, actor=actor).message_ids- newer_messages = self._trim_tool_response(agent_id=agent_id, actor=actor, message_ids=message_ids[num:])- trimmed_messages = [message_ids[0]] + newer_messages # 0 is system message- return self.set_in_context_messages(agent_id=agent_id, message_ids=trimmed_messages, actor=actor)+ new_messages = [message_ids[0]] + message_ids[num:] # 0 is system message+ return self.set_in_context_messages(agent_id=agent_id, message_ids=new_messages, actor=actor)@enforce_typesdef trim_all_in_context_messages_except_system(self, agent_id: str, actor: PydanticUser) -> PydanticAgentState:@@ -877,16 +876,6 @@ class AgentManager:new_messages = [message_ids[0]] # 0 is system messagereturn self.set_in_context_messages(agent_id=agent_id, message_ids=new_messages, actor=actor)- def _trim_tool_response(self, agent_id: str, actor: PydanticUser, message_ids: list[str]) -> PydanticAgentState:- """- Trims the tool response from the in-context messages if there is no tool call present in trimmed messages.- """- if message_ids:- messages = self.message_manager.get_messages_by_ids(message_ids=[message_ids[0]], actor=actor)- if messages and messages[0].role == "tool":- return message_ids[1:]- return message_ids-@enforce_typesdef prepend_to_in_context_messages(self, messages: List[PydanticMessage], agent_id: str, actor: PydanticUser) -> PydanticAgentState:message_ids = self.get_agent_by_id(agent_id=agent_id, actor=actor).message_idsdiff --git a/letta/services/summarizer/summarizer.py b/letta/services/summarizer/summarizer.pyindex 6fd2d1e3e..55dd62f8c 100644--- a/letta/services/summarizer/summarizer.py+++ b/letta/services/summarizer/summarizer.py@@ -107,13 +107,9 @@ class Summarizer:self.summarizer_agent.update_message_transcript(message_transcripts=formatted_evicted_messages + formatted_in_context_messages)# Add line numbers to the formatted messages- line_number = 0- for i in range(len(formatted_evicted_messages)):- formatted_evicted_messages[i] = f"{line_number}. " + formatted_evicted_messages[i]- line_number += 1- for i in range(len(formatted_in_context_messages)):- formatted_in_context_messages[i] = f"{line_number}. " + formatted_in_context_messages[i]- line_number += 1+ offset = len(formatted_evicted_messages)+ formatted_evicted_messages = [f"{i}. {msg}" for (i, msg) in enumerate(formatted_evicted_messages)]+ formatted_in_context_messages = [f"{i + offset}. {msg}" for (i, msg) in enumerate(formatted_in_context_messages)]evicted_messages_str = "\n".join(formatted_evicted_messages)in_context_messages_str = "\n".join(formatted_in_context_messages)diff --git a/tests/integration_test_summarizer.py b/tests/integration_test_summarizer.pyindex b38e03ba9..6c0f74c10 100644--- a/tests/integration_test_summarizer.py+++ b/tests/integration_test_summarizer.py@@ -147,7 +147,7 @@ def test_cutoff_calculation_with_tool_call(mocker, client: LocalClient, agent_stclient.server.agent_manager.trim_older_in_context_messages(agent_id=agent_state.id, num=3, actor=client.user)test1 = mock_set_messages.call_args_list[0][1]- assert len(test1["message_ids"]) == 4+ assert len(test1["message_ids"]) == 5mock_set_messages.reset_mock()