feat: add feature flag for new message async logic (#4805)

* feat: add feature flag for new message async logic

* remove from api sync

fern/scripts/prepare-openapi.ts

@@ -98,7 +98,12 @@ for (const path of Object.keys(lettaAgentsAPI.paths)) {
       ][method].parameters.filter(
         (param: Record<string, string>) =>
           param.in !== 'header' ||
-          (param.name !== 'user_id' && param.name !== 'User-Agent' && param.name !== 'X-Project-Id'),
+          (
+            param.name !== 'user_id' &&
+            param.name !== 'User-Agent' &&
+            param.name !== 'X-Project-Id' &&
+            !param.name.startsWith('X-Experimental')
+          ),
       );
     }
   }

letta/server/rest_api/dependencies.py

@@ -7,24 +7,35 @@ if TYPE_CHECKING:
     from letta.server.server import SyncServer
 
 
+class ExperimentalParams(BaseModel):
+    """Experimental parameters used across REST API endpoints."""
+
+    message_async: Optional[str] = None
+
+
 class HeaderParams(BaseModel):
     """Common header parameters used across REST API endpoints."""
 
     actor_id: Optional[str] = None
     user_agent: Optional[str] = None
     project_id: Optional[str] = None
+    experimental_params: Optional[ExperimentalParams] = None
 
 
 def get_headers(
     actor_id: Optional[str] = Header(None, alias="user_id"),
     user_agent: Optional[str] = Header(None, alias="User-Agent"),
     project_id: Optional[str] = Header(None, alias="X-Project-Id"),
+    message_async: Optional[str] = Header(None, alias="X-Experimental-Message-Async"),
 ) -> HeaderParams:
     """Dependency injection function to extract common headers from requests."""
     return HeaderParams(
         actor_id=actor_id,
         user_agent=user_agent,
         project_id=project_id,
+        experimental_params=ExperimentalParams(
+            message_async=message_async,
+        ),
     )
 
 

letta/server/rest_api/routers/v1/agents.py

@@ -1669,7 +1669,7 @@ async def send_message_async(
     )
     run = await server.job_manager.create_job_async(pydantic_job=run, actor=actor)
 
-    if settings.temporal_endpoint:
+    if headers.experimental_params.message_async and settings.temporal_endpoint:
         agent_state = await server.agent_manager.get_agent_by_id_async(
             agent_id, actor, include_relationships=["memory", "multi_agent_group", "sources", "tool_exec_environment_variables", "tools"]
         )