refactor(agent): extract magic numbers to constants

backend/src/v1/agent/service.py

@@ -32,6 +32,9 @@ if TYPE_CHECKING:
 
 logger = get_logger("v1.agent.service")
 
+DEFAULT_RATE_LIMIT = 60
+EMPTY_USAGE = {"input_tokens": 0, "output_tokens": 0, "cost": "0"}
+
 
 class ResumeDecisionResult(BaseModel):
     applied: bool
@@ -83,8 +86,8 @@ class AgentChatService(BaseService):
         await enforce_rate_limit(
             scope="agent_run",
             identifier=str(user_id),
-            limit=60,
+            limit=DEFAULT_RATE_LIMIT,
-            window_seconds=60,
+            window_seconds=DEFAULT_RATE_LIMIT,
         )
         now = datetime.now(timezone.utc)
 
@@ -272,26 +275,17 @@ class AgentChatService(BaseService):
         self, *, message: str, context: dict[str, object]
     ) -> dict[str, object]:
         context["intent"] = "default"
-        return {
+        return {"content": message, "usage": EMPTY_USAGE}
-            "content": message,
-            "usage": {"input_tokens": 0, "output_tokens": 0, "cost": "0"},
-        }
 
     async def _execution_stage(
         self, *, message: str, context: dict[str, object]
     ) -> dict[str, object]:
-        return {
+        return {"content": message, "usage": EMPTY_USAGE}
-            "content": message,
-            "usage": {"input_tokens": 0, "output_tokens": 0, "cost": "0"},
-        }
 
     async def _organization_stage(
         self, *, message: str, context: dict[str, object]
     ) -> dict[str, object]:
-        return {
+        return {"content": message, "usage": EMPTY_USAGE}
-            "content": message,
-            "usage": {"input_tokens": 0, "output_tokens": 0, "cost": "0"},
-        }
 
     async def get_state_snapshot(self, session_id: UUID) -> dict | None:
         stmt = select(AgentChatSession).where(AgentChatSession.id == session_id)