Update app.py

app.py

@@ -416,165 +416,164 @@ def screenshot_quick(history: List[List[str]]) -> List[List[str]]:
     """Quick screenshot function"""
     return process_message("Take a screenshot of the current page", history)
 
-def create_interface():
-    """Create the Gradio interface"""
+
     
-    with gr.Blocks(
-        title="MCP Browser Agent - Token Optimized",
-        theme=gr.themes.Soft()
-    ) as interface:
-        
-        gr.HTML("""
-        <div class="header">
-            <h1>🌐 MCP Browser Agent - Token Optimized</h1>
-            <p>AI-powered web browsing with persistent sessions and optimized token usage</p>
-        </div>
-        """)
-        
-        with gr.Row():
-            with gr.Column(scale=1):
-                gr.Markdown("### 🔧 Configuration")
-                api_key_input = gr.Textbox(
-                    label="Mistral API Key",
-                    placeholder="Enter your Mistral API key...",
-                    type="password",
-                    lines=1
-                )
-                
-                init_button = gr.Button("Initialize Agent", variant="primary")
-                status_output = gr.Textbox(
-                    label="Status & Available Tools",
-                    interactive=False,
-                    lines=6
-                )
-                
-                gr.Markdown("### 💰 Token Optimization")
-                token_stats_button = gr.Button("Show Token Stats", variant="secondary")
-                token_stats_output = gr.Textbox(
-                    label="Token Usage Statistics",
-                    interactive=False,
-                    lines=8
-                )
-                
-                gr.Markdown("""
-                ### 📝 Optimized Usage Tips
-                **Token Savings Features:**
-                - Only last 3 conversation pairs sent to API
-                - Session context maintained separately
-                - Reduced max tokens per response
-                - Smart context summarization
-                
-                **Best Practices:**
-                - Be specific in your requests
-                - Use "take screenshot" to check current state
-                - Ask for "browser status" if you need context
-                - Long conversations automatically optimized
-                """)
-                
-            with gr.Column(scale=2):
-                gr.Markdown("### 💬 Chat with Browser Agent")
-                
-                chatbot = gr.Chatbot(
-                    label="Conversation",
-                    height=500,
-                    show_copy_button=True
+with gr.Blocks(
+    title="MCP Browser Agent - Token Optimized",
+    theme=gr.themes.Soft()
+) as interface:
+    
+    gr.HTML("""
+    <div class="header">
+        <h1>🌐 MCP Browser Agent - Token Optimized</h1>
+        <p>AI-powered web browsing with persistent sessions and optimized token usage</p>
+    </div>
+    """)
+    
+    with gr.Row():
+        with gr.Column(scale=1):
+            gr.Markdown("### 🔧 Configuration")
+            api_key_input = gr.Textbox(
+                label="Mistral API Key",
+                placeholder="Enter your Mistral API key...",
+                type="password",
+                lines=1
+            )
+            
+            init_button = gr.Button("Initialize Agent", variant="primary")
+            status_output = gr.Textbox(
+                label="Status & Available Tools",
+                interactive=False,
+                lines=6
+            )
+            
+            gr.Markdown("### 💰 Token Optimization")
+            token_stats_button = gr.Button("Show Token Stats", variant="secondary")
+            token_stats_output = gr.Textbox(
+                label="Token Usage Statistics",
+                interactive=False,
+                lines=8
+            )
+            
+            gr.Markdown("""
+            ### 📝 Optimized Usage Tips
+            **Token Savings Features:**
+            - Only last 3 conversation pairs sent to API
+            - Session context maintained separately
+            - Reduced max tokens per response
+            - Smart context summarization
+            
+            **Best Practices:**
+            - Be specific in your requests
+            - Use "take screenshot" to check current state
+            - Ask for "browser status" if you need context
+            - Long conversations automatically optimized
+            """)
+            
+        with gr.Column(scale=2):
+            gr.Markdown("### 💬 Chat with Browser Agent")
+            
+            chatbot = gr.Chatbot(
+                label="Conversation",
+                height=500,
+                show_copy_button=True
+            )
+            
+            with gr.Row():
+                message_input = gr.Textbox(
+                    label="Message",
+                    placeholder="Enter your browsing request...",
+                    lines=2,
+                    scale=4
                 )
-                
-                with gr.Row():
-                    message_input = gr.Textbox(
-                        label="Message",
-                        placeholder="Enter your browsing request...",
-                        lines=2,
-                        scale=4
-                    )
-                    send_button = gr.Button("Send", variant="primary", scale=1)
-                
-                with gr.Row():
-                    clear_button = gr.Button("Clear Chat", variant="secondary")
-                    screenshot_button = gr.Button("Quick Screenshot", variant="secondary")
-        
-        # Event handlers
-        init_button.click(
-            fn=initialize_agent,
-            inputs=[api_key_input],
-            outputs=[status_output]
-        )
+                send_button = gr.Button("Send", variant="primary", scale=1)
+            
+            with gr.Row():
+                clear_button = gr.Button("Clear Chat", variant="secondary")
+                screenshot_button = gr.Button("Quick Screenshot", variant="secondary")
+    
+    # Event handlers
+    init_button.click(
+        fn=initialize_agent,
+        inputs=[api_key_input],
+        outputs=[status_output]
+    )
+    
+    send_button.click(
+        fn=process_message,
+        inputs=[message_input, chatbot],
+        outputs=[chatbot]
+    ).then(
+        fn=lambda: "",
+        outputs=[message_input]
+    )
+    
+    message_input.submit(
+        fn=process_message,
+        inputs=[message_input, chatbot],
+        outputs=[chatbot]
+    ).then(
+        fn=lambda: "",
+        outputs=[message_input]
+    )
+    
+    clear_button.click(
+        fn=lambda: [],
+        outputs=[chatbot]
+    )
+    
+    screenshot_button.click(
+        fn=screenshot_quick,
+        inputs=[chatbot],
+        outputs=[chatbot]
+    )
+    
+    token_stats_button.click(
+        fn=get_token_stats,
+        inputs=[chatbot],
+        outputs=[token_stats_output]
+    )
+    
+    # Add helpful information
+    with gr.Accordion("ℹ️ Token Optimization Guide", open=False):
+        gr.Markdown("""
+        ## 💰 How Token Optimization Works
         
-        send_button.click(
-            fn=process_message,
-            inputs=[message_input, chatbot],
-            outputs=[chatbot]
-        ).then(
-            fn=lambda: "",
-            outputs=[message_input]
-        )
+        **The Problem with Original Code:**
+        - Every API call sent complete conversation history
+        - Token usage grew exponentially with conversation length
+        - Costs could explode for long sessions
         
-        message_input.submit(
-            fn=process_message,
-            inputs=[message_input, chatbot],
-            outputs=[chatbot]
-        ).then(
-            fn=lambda: "",
-            outputs=[message_input]
-        )
+        **Our Optimization Solutions:**
         
-        clear_button.click(
-            fn=lambda: [],
-            outputs=[chatbot]
-        )
+        1. **Limited History Window**: Only last 3 conversation pairs sent to API
+        2. **Session Context**: Browser state maintained separately from chat history
+        3. **Smart Summarization**: Key session info added to each request
+        4. **Reduced Limits**: Lower max_tokens and max_iterations
+        5. **Token Tracking**: Real-time savings statistics
         
-        screenshot_button.click(
-            fn=screenshot_quick,
-            inputs=[chatbot],
-            outputs=[chatbot]
-        )
+        **Token Savings Example:**
+        ```
+        Original: 10 messages = 5,000 tokens per API call
+        Optimized: 10 messages = 500 tokens per API call
+        Savings: 90% reduction in token usage!
+        ```
         
-        token_stats_button.click(
-            fn=get_token_stats,
-            inputs=[chatbot],
-            outputs=[token_stats_output]
-        )
+        **What This Means:**
+        - ✅ Persistent browser sessions still work
+        - ✅ 90%+ reduction in API costs
+        - ✅ Faster response times
+        - ✅ Better performance for long conversations
+        - ⚠️ Agent has limited memory of old messages
         
-        # Add helpful information
-        with gr.Accordion("ℹ️ Token Optimization Guide", open=False):
-            gr.Markdown("""
-            ## 💰 How Token Optimization Works
-            
-            **The Problem with Original Code:**
-            - Every API call sent complete conversation history
-            - Token usage grew exponentially with conversation length
-            - Costs could explode for long sessions
-            
-            **Our Optimization Solutions:**
-            
-            1. **Limited History Window**: Only last 3 conversation pairs sent to API
-            2. **Session Context**: Browser state maintained separately from chat history
-            3. **Smart Summarization**: Key session info added to each request
-            4. **Reduced Limits**: Lower max_tokens and max_iterations
-            5. **Token Tracking**: Real-time savings statistics
-            
-            **Token Savings Example:**
-            ```
-            Original: 10 messages = 5,000 tokens per API call
-            Optimized: 10 messages = 500 tokens per API call
-            Savings: 90% reduction in token usage!
-            ```
-            
-            **What This Means:**
-            - ✅ Persistent browser sessions still work
-            - ✅ 90%+ reduction in API costs
-            - ✅ Faster response times
-            - ✅ Better performance for long conversations
-            - ⚠️ Agent has limited memory of old messages
-            
-            **If Agent Needs Earlier Context:**
-            - Use "browser status" to check current state
-            - Take screenshots to show current page
-            - Re-explain context if needed
-            - Clear chat periodically for fresh start
-            """)
+        **If Agent Needs Earlier Context:**
+        - Use "browser status" to check current state
+        - Take screenshots to show current page
+        - Re-explain context if needed
+        - Clear chat periodically for fresh start
+        """)
+    
     
-    return interface
 
 def cleanup_agent():
     """Cleanup agent resources"""
@@ -599,16 +598,11 @@ if __name__ == "__main__":
         try:
             logger.info("🚀 Starting MCP Browser Agent Application with Token Optimization...")
             
-            # Create and launch interface
-            interface = create_interface()
-            
-            # Launch interface (this will block)
             interface.launch(
                 server_name="0.0.0.0",
                 server_port=7860,
                 share=False,
-                show_error=True,
-                quiet=False
+                show_error=True
             )
         except Exception as e:
             logger.error(f"Application error: {e}")