diff --git a/src/interface.jl b/src/interface.jl
index 7d0f10f..ee435f3 100755
--- a/src/interface.jl
+++ b/src/interface.jl
@@ -234,7 +234,7 @@ function chat_mistral_openorca(a::agentReflex)
   return response
 end
 
-#WORKING
+
 function planner_mistral_openorca(a::agentReflex)
   """
     general prompt format:
@@ -511,6 +511,7 @@ end
 """
 function conversation(a::agentReflex, usermsg::String; attemptlimit::Int=3)
   a.attemptlimit = attemptlimit
+  workstate = nothing
   response = nothing
 
   # a.earlierConversation = conversationSummary(a)
@@ -521,10 +522,16 @@ function conversation(a::agentReflex, usermsg::String; attemptlimit::Int=3)
   @show isusetools
 
   if isusetools # use tools before responseing
-    response = work(a)
+    workstate, response = work(a)
   end
 
-  response = chat_mistral_openorca(a) #TODO if chatbox is used, skip this 
+  # if LLM using chatbox, use returning msg form chatbox as conversation response
+  if workstate == "chatbox" ?
+    #TODO paraphrase msg so that it is human friendlier word.
+  else
+    response = chat_mistral_openorca(a)
+  end
+  
   response = removeTrailingCharacters(response)
   _ = addNewMessage(a, "assistant", response)
 
@@ -536,51 +543,46 @@ end
   There are many work() depend on thinking mode.
 """ 
 function work(a::agentReflex)
-  actorstate = nothing
+  workstate = nothing
   response = nothing
 
   while true  # Work loop
-    #WORKING
+    
     a.memory[:shortterm] = OrderedDict{String, Any}()
     a.memory[:log] = OrderedDict{String, Any}()
 
-
-
-
     # plan
     if a.attempt <= a.attemptlimit
       toolname = nothing
       toolinput = nothing
-      if a.newplan == true
-        a.attempt += 1
-        a.step = 0
-        objective, plan = planner_mistral_openorca(a)
-        println("")
-        @show objective
-        @show plan
+      
+      a.step = 0
+      objective, plan = planner_mistral_openorca(a)
+      println("")
+      @show objective
+      @show plan
 
-        # sometimes LLM add not-need word I don't want
-        # plan = splittext(response, ["Step 1", "<|im_end|>", "Response", "Execution", 
-        #                           "Result", "Recommendation", "My response"])
-        # plan = replace(plan, "Plan:"=>"")
-        # println("")
-        # @show plan
-        
-        a.newplan = false
-        a.memory[:shortterm]["Plan $(a.attempt):"] = plan
-        a.memory[:log]["Plan $(a.attempt):"] = plan
-      end
+      # sometimes LLM add not-need word I don't want
+      # plan = splittext(response, ["Step 1", "<|im_end|>", "Response", "Execution", 
+      #                           "Result", "Recommendation", "My response"])
+      # plan = replace(plan, "Plan:"=>"")
+      # println("")
+      # @show plan
+      
+      a.memory[:shortterm]["Plan $(a.attempt):"] = plan
+      a.memory[:log]["Plan $(a.attempt):"] = plan
       
       println("")
       @show a.attempt
 
       # enter actor loop
-      actorstate, msgToUser = actor(a)
+      actorstate, msgToUser = actor(a)  #WORKING
 
       if actorstate == "chatbox"
         response = msgToUser
+        workstate = actorstate
         break
-      elseif actorstate == "all steps done" ||  actorstate == "formulateUserresponse"
+      elseif actorstate == "all steps done"
         println("all steps done")
 
         response = formulateUserresponse(a)
@@ -592,7 +594,7 @@ function work(a::agentReflex)
         a.memory[:shortterm]["response $(a.attempt):"] = response
         a.memory[:log]["response $(a.attempt):"] = response
 
-        # evaluate. if score < 8/10 try again.
+        # evaluate. if score > 6/10 good enough.
         guideline = writeEvaluationGuideline(a, a.memory[:shortterm]["user:"])
 
         println("")
@@ -600,7 +602,7 @@ function work(a::agentReflex)
 
         score = grading(a, guideline, response)
         @show score
-        if score >= 6 # good enough answer
+        if score > 6 # good enough answer
           break
         else  # self evaluate and reflect then try again
           analysis = analyze(a)
@@ -621,11 +623,10 @@ function work(a::agentReflex)
           chunkedtext = chunktext(lessonwithcontext, headers)
 
           a.memory[:longterm][chunkedtext["Context:"]] = chunkedtext["Lesson:"]
-          a.newplan = true
+          a.attempt += 1
           println("")
           println("RETRY $(a.attempt +1)")
           println("")
-          
         end
       else
         error("undefied condition, actorstate $actorstate $(@__LINE__)")
@@ -638,11 +639,7 @@ function work(a::agentReflex)
   end
 
   # good enough answer
-
-
-  # communicates with user
-  _ = addNewMessage(a, "assistant", response)
-  return  actorstate, response
+  return  workstate, response
 end
 
 
@@ -902,6 +899,8 @@ function actor(a::agentReflex)
   totalsteps = checkTotalStepInPlan(a)
 
   while true  # Actor loop
+
+    # decide whether to repeat step or do the next step
     if a.step == 0
       a.step = 1
     else
@@ -917,10 +916,8 @@ function actor(a::agentReflex)
         a.memory[:shortterm] = removeHeaders(a.memory[:shortterm], a.step, ["Plan"])
         a.memory[:log] = removeHeaders(a.memory[:log], a.step, ["Plan"])
         println("repeating step $(a.step)")
-      elseif decision == "formulateUserresponse"
-        actorState = "formulateUserresponse"
-        msgToUser = nothing
-        break
+
+      #WORKING
       else
         error("undefined condition decision = $decision $(@__LINE__)")
       end
@@ -928,7 +925,7 @@ function actor(a::agentReflex)
     
 
     @show a.step
-    if a.step < totalsteps
+    if a.step < totalsteps -1   # the last step of the plan is responding, let work() do this part
 
       prompt_actor = actor_mistral_openorca(a)
 
@@ -978,14 +975,10 @@ function actor(a::agentReflex)
         msgToUser = toolinput
         actorState = toolname
         break
-      elseif toolname == "formulateUserresponse"
-        msgToUser = toolinput
-        actorState = toolname
-        break
       elseif toolname == "skipstep"
         # skip
       else  # function call
-        f = a.tools[Symbol(toolname)][:func]
+        f = a.tools[toolname][:func]
         toolresult = f(a, toolinput)
         @show toolresult
         a.memory[:shortterm]["Obs $(a.step):"] = toolresult
@@ -1365,8 +1358,6 @@ function goNogo(a)
     decision = "Yes"
   elseif occursin("No", response)
     decision = "No"
-  elseif occursin("formulateUserresponse", response)
-    decision = "formulateUserresponse"
   else
     error("undefied condition, decision $decision $(@__LINE__)")
   end