update

src/interface.jl

@@ -203,6 +203,7 @@ function planner_mistral_openorca(a::agentReflex)
     Plan: first you should always think about your conversation with the user and your earlier work thoroughly then extract and devise a complete, step by step plan to achieve your objective (pay attention to correct numeral calculation and commonsense).
     P.S.1 each step of the plan should be a single action.
     P.S.2 ask the user if you don't have info.
+    P.S.3 mark a completed step with done keyword.
     <|/s|>
     $conversation
     <|assistant|>
@@ -557,6 +558,24 @@ function conversation(a::agentReflex, usermsg::String; attemptlimit::Int=3)
   return response
 end
 
+""" Direct conversation is not an agent, messages does not pass through logic loop
+  but goes directly to LLM.
+  
+"""
+function directconversation(a::agentReflex, usermsg::String)
+  response = nothing
+
+  _ = addNewMessage(a, "user", usermsg)
+  if isusetools # use tools before responseing
+    workstate, response = work(a)
+  end
+
+  response = chat_mistral_openorca(a)
+  response = removeTrailingCharacters(response)
+  _ = addNewMessage(a, "assistant", response)
+  return response
+end
+
 """
   Continuously run llm functions except when llm is getting Answer: or chatbox.
   There are many work() depend on thinking mode.