diff --git a/src/interface.jl b/src/interface.jl
index 7c425e3..afa2438 100755
--- a/src/interface.jl
+++ b/src/interface.jl
@@ -205,7 +205,7 @@ function chat_mistral_openorca(a::agentReflex, usermsg::String)
       {context} = 
         "
         {earlierConversation}
-        {env status}
+        {env state}
         {shortterm memory}
         {longterm memory}
         "
@@ -268,7 +268,7 @@ function planner_mistral_openorca(a::agentReflex, usermsg::String)
       {context} = 
         "
         {earlierConversation}
-        {env status}
+        {env state}
         {shortterm memory}
         {longterm memory}
         "
@@ -317,6 +317,77 @@ function planner_mistral_openorca(a::agentReflex, usermsg::String)
   return prompt
 end
 
+function actor_mistral_openorca(a::agentReflex, usermsg::T) where {T<:AbstractString}
+  """
+    general prompt format:
+
+    "
+    <|im_start|>system
+    {role}
+    {tools}
+    {thinkingFormat}
+    <|im_end|>
+    {context}
+    <|im_start|>user
+    {usermsg}
+    <|im_end|>
+    <|im_start|>assistant
+
+    "
+
+    Note:
+      {context} = 
+        "
+        {earlierConversation}
+        {env state}
+        {shortterm memory}
+        {longterm memory}
+        "
+  """
+
+  prompt = 
+  """
+    <|im_start|>system
+    {role}
+    {tools}
+    {thinkingFormat}
+    <|im_end|>
+    {context}
+    <|im_start|>user
+    {usermsg}
+    <|im_end|>
+    <|im_start|>assistant
+
+  """
+  error("actor_mistral_openorca done")
+  prompt = replace(prompt, "{role}" => a.roles[a.role])
+  prompt = replace(prompt, "{thinkingFormat}" => a.thinkingFormat[:actor])
+  toolnames = ""
+  toollines = ""
+  for (toolname, v) in a.tools
+    toolline = "$toolname: $(v[:description]) $(v[:input]) $(v[:output])\n"
+    toollines *= toolline
+    toolnames *= "$toolname,"
+  end
+  prompt = replace(prompt, "{toolnames}" => toolnames)
+  prompt = replace(prompt, "{tools}" => "You have access to the following tools:\n$toollines")
+  
+  context = 
+  """
+  {earlierConversation}
+  {current status}
+  {longterm memory}
+  """
+  context = replace(context, "{earlierConversation}" => "My earlier talk with the user:\n$(a.earlierConversation)")
+  context = replace(context, "{current status}" => "")
+  context = replace(context, "{longterm memory}" => "")
+
+  prompt = replace(prompt, "{context}" => context)
+
+  prompt = replace(prompt, "{usermsg}" => "Assignment: $usermsg")
+
+  return prompt
+end
 
 """
   Chat with llm.
@@ -553,17 +624,21 @@ function work(a::agentReflex, usermsg::String)
     respond = sendReceivePrompt(a, prompt)
     plan = split(respond, "<|im_end|>")[1]
     @show respond
-    step = 0
-    while true
-      step += 1
-      isstep, stepdetail = extractStepFromPlan(a, plan, step)
+    a.step = 0
+    while true  # execute step by step
+      a.step += 1
+      isstep, task = extractStepFromPlan(a, plan)
       @show isstep
-      @show stepdetail
+      @show task
       if isstep
-        
+        respond = actor_mistral_openorca(a, task)
+        error("work done")
+      else  # finish all steps
+        break
+      end
 
 
-      error("work done")
+      
     end
     # evaluate
     
@@ -1033,12 +1108,12 @@ function chunktext(text::T, headers) where {T<:AbstractString}
 end
 
 
-function extractStepFromPlan(a::agent, plan::T, stepToExtract::Int) where {T<:AbstractString}
+function extractStepFromPlan(a::agent, plan::T) where {T<:AbstractString}
   prompt =
   """
   <|im_start|>system
   You are a helpful assistant.
-  Your job is to determine whether step {$stepToExtract} is in the user plan.
+  Your job is to determine whether step {$(a.step)} is in the user plan.
 
   Choose one of the following choices:
   If there isn't say, {no}.
diff --git a/src/type.jl b/src/type.jl
index 13ba5ad..6b4f493 100644
--- a/src/type.jl
+++ b/src/type.jl
@@ -35,7 +35,8 @@ abstract type agent end
   messages = Vector{Dict{Symbol, Any}}()
   tools::Union{Dict, Nothing} = nothing 
   thoughtlog::String = "nothing"  # logs unfinished thoughts
-  attempt::Int = 0 # no. of attempt
+  attempt::Int = 0 # attempted number
+  step::Int = 0 # step number
   thinkingroundlimit::Int = 5 # thinking round limit
   thinkingmode::Symbol = :no_thinking
   thinkingFormat::Union{Dict, Nothing} = nothing
@@ -86,9 +87,9 @@ function agentReflex(
     :planner=> 
       """Use the following format:
       Assignment: the input assignment your user is assigning and you must finish
-      Plan: first you should always think about the assignment and the info you have thoroughly then extract and devise a step by step plan to finish (pay attention to correct numeral calculation and commonsense).
+      Plan: first you should always think about the assignment and the info you have thoroughly then extract and devise a step by step plan to finish the assignment (pay attention to correct numeral calculation and commonsense).
       """,
-    :worker=> 
+    :actor=> 
       """Use the following format:
       Assignment: the input assignment your user is assigning and you must finish
       Thought: your should always think about what to do (pay attention to correct numeral calculation and commonsense).