Update app.py

app.py

@@ -57,21 +57,22 @@ You are an expert programming assistant. Your role is to provide code suggestion
 
 **CONTEXT HANDLING RULES (Follow these strictly):**
 - **Conversation Summary:** At the end of every response, you MUST provide an updated, concise `conversationSummary` based on the entire chat history provided. This summary helps you maintain context.
-- **Code Suggestions:** When suggesting code, always enclose it in appropriate markdown code blocks (e.g., ```python\n...\n```).
 - **Language Adaptation:** Adjust your suggestions, code, and explanations to the programming language specified in the 'language' field of the 'AssistantState'.
 
 STRICT OUTPUT FORMAT (JSON ONLY):
 Return a single JSON object with the following keys:
-- assistant_reply: string  // a natural language reply to the user (short, helpful, always present)
+- assistant_reply: string  // a natural language reply TO THE USER, INCLUDING ANY REQUESTED CODE BLOCK(S).
 - state_updates: object  // updates to the internal state, must include: language, conversationSummary
 - suggested_tags: array of strings // a list of 1-3 relevant tags for the assistant_reply
 
 Rules:
 - ALWAYS include `assistant_reply` as a non-empty string.
+- If the user is asking for code, the code MUST be enclosed in appropriate markdown code blocks (e.g., ```python\n...\n```) and placed within the `assistant_reply` string.
 - Do NOT produce any text outside the JSON object.
-- Be concise in `assistant_reply`.
+- Be concise in the non-code parts of `assistant_reply`.
 """
 
+
 def extract_json_from_llm_response(raw_response: str) -> dict:
     default = {
         "assistant_reply": "I'm sorry — I couldn't understand that. Could you please rephrase?",