fix: move ConversationAgent after BaseAgent definition

Co-authored-by: aider (openai/editor-model) <aider@aider.chat>

app.py

@@ -2565,7 +2565,7 @@ class AgentReputationEngine:
         return rep
 
 
-# --- V2 PHASE 4 CONVERSATION & NOTIFICATION ENGINES ---
+# --- V2 PHASE 4 NOTIFICATION ENGINE ---
 class NotificationEngine:
     """Central notification center managing persisted system alerts and WS events."""
 
@@ -2580,34 +2580,6 @@ class NotificationEngine:
         return notif
 
 
-class ConversationAgent(BaseAgent):
-    """Dedicated Conversation Agent for human interaction, clarification, failure reporting, and approvals."""
-
-    def __init__(self, db: DatabaseManager, message_bus: MessageBus, event_bus: EventBus, model_manager: ModelManager):
-        super().__init__("agent-conversation-01", "Mnemosyne Chat", "Human Interface & Conversation Specialist", db, message_bus, event_bus)
-        self.model_manager = model_manager
-
-    async def process_user_message(self, user_message: str, user_id: str = "human-operator") -> str:
-        # Record user message in conversation history
-        user_log = ConversationMessageModel(user_id=user_id, sender="Human Operator", message=user_message)
-        await self.db.save_conversation_log(user_log)
-
-        # Context lookup from memory vault
-        memories = await self.db.search_memories(user_message)
-        ctx_str = "\n".join([m["content"] for m in memories[:3]]) if memories else "No direct memory match."
-
-        prompt = f"User said: '{user_message}'\nRelevant Memory Context:\n{ctx_str}\nProvide a helpful, polite, and strategic response as Spark Colony OS Operator Assistant."
-        resp = await self.model_manager.generate_response(LogicalModel.MDL_FST, prompt)
-        reply_text = resp["content"]
-
-        # Record agent reply
-        agent_log = ConversationMessageModel(user_id=user_id, sender=self.name, message=reply_text)
-        await self.db.save_conversation_log(agent_log)
-        await self.record_memory("system", f"Human Chat Interaction: {user_message} -> {reply_text}", ["conversation", "human_interaction"])
-
-        return reply_text
-
-
 # --- CONFIDENCE ENGINE & COST TRACKER ---
 class ConfidenceEngine:
     """Calculates objective confidence metrics based on evidence attributes."""
@@ -3078,6 +3050,34 @@ class DynamicWorkerAgent(BaseAgent):
         return res
 
 
+class ConversationAgent(BaseAgent):
+    """Dedicated Conversation Agent for human interaction, clarification, failure reporting, and approvals."""
+
+    def __init__(self, db: DatabaseManager, message_bus: MessageBus, event_bus: EventBus, model_manager: ModelManager):
+        super().__init__("agent-conversation-01", "Mnemosyne Chat", "Human Interface & Conversation Specialist", db, message_bus, event_bus)
+        self.model_manager = model_manager
+
+    async def process_user_message(self, user_message: str, user_id: str = "human-operator") -> str:
+        # Record user message in conversation history
+        user_log = ConversationMessageModel(user_id=user_id, sender="Human Operator", message=user_message)
+        await self.db.save_conversation_log(user_log)
+
+        # Context lookup from memory vault
+        memories = await self.db.search_memories(user_message)
+        ctx_str = "\n".join([m["content"] for m in memories[:3]]) if memories else "No direct memory match."
+
+        prompt = f"User said: '{user_message}'\nRelevant Memory Context:\n{ctx_str}\nProvide a helpful, polite, and strategic response as Spark Colony OS Operator Assistant."
+        resp = await self.model_manager.generate_response(LogicalModel.MDL_FST, prompt)
+        reply_text = resp["content"]
+
+        # Record agent reply
+        agent_log = ConversationMessageModel(user_id=user_id, sender=self.name, message=reply_text)
+        await self.db.save_conversation_log(agent_log)
+        await self.record_memory("system", f"Human Chat Interaction: {user_message} -> {reply_text}", ["conversation", "human_interaction"])
+
+        return reply_text
+
+
 # --- SPECIALIZED COLONY AGENTS ---
 class CommanderAgent(BaseAgent):
     """Commander Agent: Manages workflow, assigns tasks, measures confidence. NEVER searches, browses, or writes reports."""