Update app/services/chat_service.py

app/services/chat_service.py

@@ -32,69 +32,55 @@ class ChatService:
         try:
             context = self.prepare_context(code_chunks)
             
-            prompt = f"""You are QODEX AI, an expert code assistant analyzing the {repository_name} repository.
+            prompt = f"""You are an expert code assistant analyzing the {repository_name} repository.
 
 User Question: {query}
 
-Relevant Code Context:
+Code Context:
 {context}
 
-RESPONSE FORMAT WITH KEYWORDS:
-Use these EXACT keywords for formatting:
+RESPONSE FORMAT:
+Use clean, professional markdown formatting similar to GitHub README files:
 
-[ANSWER] - Direct answer to question
-[SECTION] - Main section title  
-[SUBSECTION] - Subsection title
-[POINT] - Bullet point
-[CODE] - Code elements (functions, variables, classes)
-[FILE] - File references
-[HIGHLIGHT] - Important concepts
-[FLOW] - Step-by-step process
-[INSIGHT] - Key insights or explanations
-[NOTE] - Additional important information
-[DIVIDER] - Section separator
+## Main Answer
+[Direct answer to the user's question]
 
-EXAMPLE RESPONSE FORMAT:
+## Implementation Overview  
+[High-level explanation of how the feature/system works]
 
-[ANSWER]The authentication system uses JWT tokens with refresh token rotation for enhanced security.
+### Key Components
+- **Component Name**: Brief description (`filename.py`, lines X-Y)
+- **Component Name**: Brief description (`filename.py`, lines X-Y)
 
-[SECTION]How Authentication Works
+### Technical Details
+[Detailed explanation with code references]
 
-The system implements a comprehensive authentication flow that balances security with user experience. When a user logs in, the application creates two types of tokens that work together to maintain secure sessions.
+When referencing code:
+- Use **bold** for important file names and concepts
+- Use `backticks` for functions, variables, classes, and code snippets
+- Reference specific files and line numbers: `filename.py` (lines X-Y)
+- Use > blockquotes for important insights or warnings
 
-[SUBSECTION]Token Generation Process
+### How It Works
+1. **Step 1**: Description of first step
+2. **Step 2**: Description of second step  
+3. **Step 3**: Description of third step
 
-[FLOW]User Login Flow:
-[POINT]User submits credentials through the login form
-[POINT]Server validates credentials against the database  
-[POINT]If valid, server generates both access and refresh tokens
-[POINT]Tokens are stored in httpOnly cookies for security
-[POINT]User is redirected to the dashboard
+> **Key Insight**: Important observations about the implementation
 
-The [CODE]generateTokens[/CODE] function in [FILE]auth.service.ts[/FILE] handles this entire process. It uses the [HIGHLIGHT]jsonwebtoken library[/HIGHLIGHT] to create tokens with specific expiration times.
+## Additional Notes
+[Any limitations, missing information, or recommendations]
 
-[SUBSECTION]Token Security Features
+REQUIREMENTS:
+- NO emojis - use clean text only
+- Be comprehensive and detailed
+- Reference specific files and line numbers
+- Explain both WHAT the code does and HOW it works
+- Use proper markdown hierarchy (##, ###, -, >, etc.)
+- Focus on explanation rather than just code listing
+- Professional documentation style
 
-[INSIGHT]The system uses short-lived access tokens (15 minutes) combined with longer-lived refresh tokens (7 days). This approach minimizes the risk window if an access token is compromised, while the refresh token allows seamless re-authentication.
-
-[POINT][HIGHLIGHT]Access Tokens[/HIGHLIGHT]: Contain user claims and expire quickly
-[POINT][HIGHLIGHT]Refresh Tokens[/HIGHLIGHT]: Used to generate new access tokens  
-[POINT][HIGHLIGHT]Automatic Rotation[/HIGHLIGHT]: Refresh tokens are replaced on each use
-
-[DIVIDER]
-
-[NOTE]The refresh token rotation prevents token replay attacks by invalidating old refresh tokens immediately after use.
-
-INSTRUCTIONS:
-1. Use the EXACT keyword format shown above
-2. Provide detailed explanations, not just code references
-3. Explain WHY things work the way they do
-4. Focus on concepts and implementation logic
-5. Use sources for code proof, but explain in your own words
-6. If context is insufficient, clearly state what's missing
-7. Be comprehensive but well-structured
-
-Your Expert Analysis:"""
+Your detailed markdown analysis:"""
             
             response = self.model.generate_content(prompt)