chore: bump version 0.7.19 (#2643)

Co-authored-by: Andy Li <55300002+cliandy@users.noreply.github.com>
Co-authored-by: Kevin Lin <klin5061@gmail.com>
Co-authored-by: Sarah Wooders <sarahwooders@gmail.com>
Co-authored-by: jnjpng <jin@letta.com>

letta/__init__.py

@@ -1,4 +1,4 @@
-__version__ = "0.7.18"
+__version__ = "0.7.19"
 
 # import clients
 from letta.client.client import LocalClient, RESTClient, create_client

letta/groups/sleeptime_multi_agent_v2.py

@@ -231,7 +231,7 @@ class SleeptimeMultiAgentV2(BaseAgent):
             # Update job status
             job_update = JobUpdate(
                 status=JobStatus.completed,
-                completed_at=datetime.now(timezone.utc),
+                completed_at=datetime.now(timezone.utc).replace(tzinfo=None),
                 metadata={
                     "result": result.model_dump(mode="json"),
                     "agent_id": sleeptime_agent_id,
@@ -242,7 +242,7 @@ class SleeptimeMultiAgentV2(BaseAgent):
         except Exception as e:
             job_update = JobUpdate(
                 status=JobStatus.failed,
-                completed_at=datetime.now(timezone.utc),
+                completed_at=datetime.now(timezone.utc).replace(tzinfo=None),
                 metadata={"error": str(e)},
             )
             self.job_manager.update_job_by_id(job_id=run_id, job_update=job_update, actor=self.actor)

letta/llm_api/google_vertex_client.py

@@ -235,6 +235,8 @@ class GoogleVertexClient(GoogleAIClient):
                             )
 
                         except json.decoder.JSONDecodeError:
+                            if candidate.finish_reason == "MAX_TOKENS":
+                                raise ValueError(f"Could not parse response data from LLM: exceeded max token limit")
                             # Inner thoughts are the content by default
                             inner_thoughts = response_message.text
 

letta/services/job_manager.py

@@ -72,7 +72,7 @@ class JobManager:
                 setattr(job, key, value)
 
             if update_data.get("status") == JobStatus.completed and not job.completed_at:
-                job.completed_at = get_utc_time()
+                job.completed_at = get_utc_time().replace(tzinfo=None)
                 if job.callback_url:
                     self._dispatch_callback(session, job)
 
@@ -96,7 +96,7 @@ class JobManager:
                 setattr(job, key, value)
 
             if update_data.get("status") == JobStatus.completed and not job.completed_at:
-                job.completed_at = get_utc_time()
+                job.completed_at = get_utc_time().replace(tzinfo=None)
                 if job.callback_url:
                     await self._dispatch_callback_async(session, job)
 

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "letta"
-version = "0.7.18"
+version = "0.7.19"
 packages = [
     {include = "letta"},
 ]

tests/integration_test_batch_api_cron_jobs.py

@@ -16,7 +16,6 @@ from anthropic.types.beta.messages import (
     BetaMessageBatchSucceededResult,
 )
 from dotenv import load_dotenv
-from letta_client import Letta
 
 from letta.config import LettaConfig
 from letta.helpers import ToolRulesSolver
@@ -75,12 +74,6 @@ def server():
     return SyncServer()
 
 
-@pytest.fixture(scope="session")
-def client(server_url):
-    """Creates a REST client for testing."""
-    return Letta(base_url=server_url)
-
-
 # --- Dummy Response Factories --- #
 
 
@@ -263,7 +256,7 @@ def mock_anthropic_client(server, batch_a_resp, batch_b_resp, agent_b_id, agent_
 # End-to-End Test
 # -----------------------------
 @pytest.mark.asyncio(loop_scope="session")
-async def test_polling_simple_real_batch(client, default_user, server):
+async def test_polling_simple_real_batch(default_user, server):
     # --- Step 1: Prepare test data ---
     # Create batch responses with different statuses
     # NOTE: This is a REAL batch id!
@@ -404,7 +397,7 @@ async def test_polling_simple_real_batch(client, default_user, server):
 
 
 @pytest.mark.asyncio(loop_scope="session")
-async def test_polling_mixed_batch_jobs(client, default_user, server):
+async def test_polling_mixed_batch_jobs(default_user, server):
     """
     End-to-end test for polling batch jobs with mixed statuses and idempotency.