fix(hub): use LlmAgent with explicit Vertex AI client

agents/core-logic/agent.py

@@ -14,7 +14,7 @@ import google.auth.transport.requests
 from typing import Literal
 
 
-from google.adk.agents import Agent
+from google.adk.agents import LlmAgent
 from google.adk.tools import FunctionTool
 from google.adk.runners import Runner
 from gemini_guard import guarded_run, CircuitOpen, RateLimitExceeded
@@ -254,12 +254,13 @@ def build_agent(mode: str = "light", message: str = "") -> tuple:
     models = get_models_for_mode(mode)
     actual_model = route_model(message=message, mode=mode, flash_model=ORCHESTRATOR_MODEL, pro_model=models["orchestrator"])
     instruction = _INSTRUCTION_TEMPLATE.format(mode=mode, model=actual_model)
-    agent = Agent(
+    agent = LlmAgent(
         model=actual_model,
-        name="jason_vauger",
-        description=f"Jason Vauger — OPAX enterprise agent [{mode}]",
         instruction=instruction,
         tools=OPAX_TOOLS,
+        vertexai=True,
+        project="propane-will-491900-m5",
+        location="us-central1",
     )
     return agent, actual_model, instruction