fix: complete pipeline working

User interrupts still need to be tested.

ref: N25B-429

src/control_backend/agents/llm/llm_agent.py

@@ -46,12 +46,17 @@ class LLMAgent(BaseAgent):
         :param msg: The received internal message.
         """
         if msg.sender == settings.agent_settings.bdi_core_name:
-            self.logger.debug("Processing message from BDI core.")
-            try:
-                prompt_message = LLMPromptMessage.model_validate_json(msg.body)
-                await self._process_bdi_message(prompt_message)
-            except ValidationError:
-                self.logger.debug("Prompt message from BDI core is invalid.")
+            match msg.thread:
+                case "prompt_message":
+                    try:
+                        prompt_message = LLMPromptMessage.model_validate_json(msg.body)
+                        await self._process_bdi_message(prompt_message)
+                    except ValidationError:
+                        self.logger.debug("Prompt message from BDI core is invalid.")
+                case "assistant_message":
+                    self.history.append({"role": "assistant", "content": msg.body})
+                case "user_message":
+                    self.history.append({"role": "user", "content": msg.body})
         else:
             self.logger.debug("Message ignored (not from BDI core.")
 
@@ -114,13 +119,6 @@ class LLMAgent(BaseAgent):
         :param goals: Goals the LLM should achieve.
         :yield: Fragments of the LLM-generated content (e.g., sentences/phrases).
         """
-        self.history.append(
-            {
-                "role": "user",
-                "content": prompt,
-            }
-        )
-
         instructions = LLMInstructions(norms if norms else None, goals if goals else None)
         messages = [
             {