Add action explanation and follow-up functionality to MachaChatSession

- Introduced `explain_action` method to provide detailed explanations for pending actions in the approval queue.
- Added `answer_action_followup` method to handle user follow-up questions regarding proposed actions.
- Updated `main` function to support discussion mode with action explanations and follow-ups.
- Refactored `conversation.py` to utilize the unified chat implementation from `chat.py`, enhancing compatibility and functionality.
- Enhanced error handling for file operations and user input validation in both new methods.

chat.py

@@ -348,10 +348,147 @@ class MachaChatSession:
         """
         response = self.process_message(question, verbose=verbose)
         return response
+    
+    def explain_action(self, action_index: int) -> str:
+        """Explain a pending action from the approval queue"""
+        # Get action from approval queue
+        approval_queue_file = self.agent.state_dir / "approval_queue.json"
+        
+        if not approval_queue_file.exists():
+            return "Error: No approval queue found."
+        
+        try:
+            with open(approval_queue_file, 'r') as f:
+                queue = json.load(f)
+            
+            if not (0 <= action_index < len(queue)):
+                return f"Error: Action #{action_index} not found in approval queue (queue has {len(queue)} items)."
+            
+            action_item = queue[action_index]
+        except Exception as e:
+            return f"Error reading approval queue: {e}"
+        
+        action = action_item.get("action", {})
+        context = action_item.get("context", {})
+        timestamp = action_item.get("timestamp", "unknown")
+        
+        # Build explanation prompt
+        prompt = f"""You are Macha, explaining a proposed system action to the user.
+
+ACTION DETAILS:
+- Proposed Action: {action.get('proposed_action', 'N/A')}
+- Action Type: {action.get('action_type', 'N/A')}
+- Risk Level: {action.get('risk_level', 'N/A')}
+- Diagnosis: {action.get('diagnosis', 'N/A')}
+- Commands to execute: {', '.join(action.get('commands', []))}
+- Timestamp: {timestamp}
+
+SYSTEM CONTEXT:
+{json.dumps(context, indent=2)}
+
+Please provide a clear, concise explanation of:
+1. What problem was detected
+2. What this action will do to fix it
+3. Why this approach was chosen
+4. Any potential risks or side effects
+5. Expected outcome
+
+Be conversational and helpful. Use plain language, not technical jargon unless necessary."""
+
+        try:
+            response = self.agent._query_ollama(prompt, temperature=0.7)
+            return response
+        except Exception as e:
+            return f"Error generating explanation: {e}"
+    
+    def answer_action_followup(self, action_index: int, user_question: str) -> str:
+        """Answer a follow-up question about a pending action"""
+        # Get action from approval queue
+        approval_queue_file = self.agent.state_dir / "approval_queue.json"
+        
+        if not approval_queue_file.exists():
+            return "Error: No approval queue found."
+        
+        try:
+            with open(approval_queue_file, 'r') as f:
+                queue = json.load(f)
+            
+            if not (0 <= action_index < len(queue)):
+                return f"Error: Action #{action_index} not found."
+            
+            action_item = queue[action_index]
+        except Exception as e:
+            return f"Error reading approval queue: {e}"
+        
+        action = action_item.get("action", {})
+        context = action_item.get("context", {})
+        
+        # Build follow-up prompt
+        prompt = f"""You are Macha, answering a follow-up question about a proposed action.
+
+ACTION SUMMARY:
+- Proposed: {action.get('proposed_action', 'N/A')}
+- Type: {action.get('action_type', 'N/A')}
+- Risk: {action.get('risk_level', 'N/A')}
+- Diagnosis: {action.get('diagnosis', 'N/A')}
+- Commands: {', '.join(action.get('commands', []))}
+
+SYSTEM CONTEXT:
+{json.dumps(context, indent=2)[:2000]}
+
+USER'S QUESTION:
+{user_question}
+
+Please answer the user's question clearly and honestly. If you're uncertain about something, say so. Focus on helping them make an informed decision about whether to approve this action."""
+
+        try:
+            response = self.agent._query_ollama(prompt, temperature=0.7)
+            return response
+        except Exception as e:
+            return f"Error: {e}"
 
 
 def main():
-    """Main entry point for macha-chat"""
+    """Main entry point for macha-chat and conversation.py"""
+    
+    # Check for --discuss flag (used by macha-approve discuss)
+    if "--discuss" in sys.argv:
+        try:
+            discuss_index = sys.argv.index("--discuss")
+            if discuss_index + 1 >= len(sys.argv):
+                print("Error: --discuss requires an action number", file=sys.stderr)
+                sys.exit(1)
+            
+            action_number = int(sys.argv[discuss_index + 1])
+            
+            session = MachaChatSession()
+            
+            # Check if this is a follow-up question or initial explanation
+            if "--follow-up" in sys.argv:
+                followup_index = sys.argv.index("--follow-up")
+                if followup_index + 1 >= len(sys.argv):
+                    print("Error: --follow-up requires a question", file=sys.stderr)
+                    sys.exit(1)
+                
+                # Get the rest of the arguments as the question
+                question = " ".join(sys.argv[followup_index + 1:])
+                response = session.answer_action_followup(action_number, question)
+                print(response)
+            else:
+                # Initial explanation
+                explanation = session.explain_action(action_number)
+                print(explanation)
+            
+            return
+            
+        except (ValueError, IndexError) as e:
+            print(f"Error: Invalid action number: {e}", file=sys.stderr)
+            sys.exit(1)
+        except Exception as e:
+            print(f"Error: {e}", file=sys.stderr)
+            sys.exit(1)
+    
+    # Normal interactive chat mode
     session = MachaChatSession()
     session.run_interactive()