Update app.py

app.py

@@ -1,24 +1,20 @@
+from smolagents import (
+    load_tool,
+    CodeAgent,
+    ToolCallingAgent,
+    InferenceClientModel,
+    LiteLLMModel,
+    OpenAIModel,
+    GradioUI,
+    MCPClient
+)
+from mcp import StdioServerParameters
 import os
 from dotenv import load_dotenv
-from smolagents import CodeAgent, ToolCallingAgent, LiteLLMModel, MCPClient
-from mcp import StdioServerParameters
-import base64
-from PIL import Image
-import io
+import gradio as gr
 
-# --- 1. Environment and Model Setup ---
-# Load environment variables from a .env file (for API keys)
 load_dotenv()
 
-# Initialize the language model that our agents will use.
-# Ensure your GEMINI_API_KEY is set in your .env file.
-model = LiteLLMModel(
-    model_id="gemini/gemini-2.0-flash-exp",
-    api_key=os.getenv("GEMINI_API_KEY")
-)
-
-# --- 2. MCP Server Configuration ---
-# Define the connection parameters for your MCP servers.
 kgb_server_parameters = StdioServerParameters(
     command="npx",
     args=[        
@@ -28,7 +24,7 @@ kgb_server_parameters = StdioServerParameters(
         "sse-only"],
 )
 
-t2i_server_parameters = StdioServerParameters(
+T2I_server_parameters = StdioServerParameters(
     command="npx",
     args=[        
         "mcp-remote",
@@ -37,89 +33,441 @@ t2i_server_parameters = StdioServerParameters(
         "sse-only"],
 )
 
-server_parameters = [kgb_server_parameters, t2i_server_parameters]
+# Model will be initialized dynamically based on user selection
+
 
-# --- 3. Main Application Logic ---
-def run_storycrafter():
-    # Instantiate the MCPClient *before* the 'with' block, as per the working example.
+
+# Load tools from all MCP servers using MCPClient
+server_parameters = [kgb_server_parameters, T2I_server_parameters]
+
+def initialize_model(model_provider, api_key_or_token):
+    """Initialize the selected model with user credentials"""
+    
+    if not api_key_or_token.strip():
+        raise ValueError("Please provide your API key or token")
+    
+    if model_provider == "Gemini (Google)":
+        return LiteLLMModel(
+            model_id="gemini/gemini-2.0-flash-exp",
+            api_key=api_key_or_token
+        )
+    elif model_provider == "Hugging Face (DeepSeek via Together)":
+        return InferenceClientModel(
+            model_id="deepseek-ai/DeepSeek-R1-0528",
+            provider="together",
+            token=api_key_or_token,
+            max_tokens=5000
+        )
+    elif model_provider == "DeepSeek (Direct API)":
+        return OpenAIModel(
+            model_id="deepseek-chat",
+            api_key=api_key_or_token,
+            base_url="https://api.deepseek.com"
+        )
+    else:
+        raise ValueError(f"Unsupported model provider: {model_provider}")
+
+def create_business_content_ui():
+    """Create specialized UI for business content creation"""
+    
+    # Initialize MCP client and keep it alive for the entire UI session
+    print("🔗 Initializing MCP client for the session...")
     mcp = MCPClient(server_parameters)
     
-    # Use the created MCPClient instance as a context manager.
     with mcp:
-        print("Connecting to MCP servers and fetching tools...")
-        # Get all available tools from all connected MCP servers.
+        # Load all MCP tools once and keep client alive
         all_tools = mcp.get_tools()
-        print(f"Found {len(all_tools)} tools.")
-        if not all_tools:
-            print("Warning: No tools were loaded from the MCP servers. Agents will have limited capabilities.")
+        print(f"✅ Loaded {len(all_tools)} MCP tools successfully")
         
-        # --- 4. Agent Definitions ---
-        # The Writer Agent is given all available tools. 
-        writer_agent = ToolCallingAgent(
-            tools=all_tools,
-            model=model,
-            name="writer",
-            description="A creative agent that writes short stories. It can use a knowledge graph tool to research topics for inspiration."
-        )
-
-        # The Illustrator Agent is also given all available tools.
-        illustrator_agent = ToolCallingAgent(
-            tools=all_tools,
-            model=model,
-            name="illustrator",
-            description="An artist agent that creates illustrations based on a descriptive prompt using a text-to-image tool."
-        )
-
-        # The Director Agent orchestrates the other two agents.
-        director_agent = CodeAgent(
-            tools=[],
-            model=model,
-            managed_agents=[writer_agent, illustrator_agent],
-            system_prompt="""
-            You are the Director of Agentic Storycrafter, a creative team. Your job is to manage the writer and illustrator agents to create a story with an illustration.
-
-            Here is your workflow:
-            1.  Receive a user's prompt for a story.
-            2.  Call the `writer` agent to write a story based on the user's prompt.
-            3.  After the story is written, create a short, descriptive prompt for an illustration that captures the essence of the story.
-            4.  Call the `illustrator` agent with this new prompt to generate an image. The result will be a dictionary containing image data.
-            5.  Return a dictionary containing both the final 'story' and the 'image_data' from the illustrator.
-            """
-        )
+        def initialize_agents(model):
+            """Initialize specialized agents using loaded MCP tools and user-selected model"""
+            
+            # 1. Business Research Agent - Uses all MCP tools (Knowledge Graph Builder focus)
+            research_agent = CodeAgent(
+                tools=all_tools,
+                model=model,
+                add_base_tools=True,
+                name="business_researcher",
+                description="""Expert business researcher specializing in market analysis, competitive intelligence, and tech industry trends. 
+                Uses knowledge graph tools to extract entities, relationships, and key business insights from topics. 
+                Focuses on: market size, key players, business relationships, competitive landscape, and strategic context."""
+            )
+            
+            # 2. Content Strategy Agent - Creates structured business content
+            content_strategy_agent = ToolCallingAgent(
+                tools=[],
+                model=model,
+                max_steps=3,
+                name="content_strategist", 
+                description="""Professional business writer specializing in executive-level content creation.
+                Creates structured, strategic business content including market analyses, competitive briefs, and strategic recommendations.
+                Writes for C-level executives, investors, and business stakeholders with focus on actionable insights."""
+            )
+            
+            # 3. Content Formatter Agent - Professional document formatting
+            content_formatter_agent = ToolCallingAgent(
+                tools=[],
+                model=model,
+                max_steps=2,
+                name="content_formatter",
+                description="""Document formatting specialist focused on professional business document structure.
+                Converts content into well-structured markdown with proper headers, tables, bullet points, and professional formatting.
+                Ensures consistency, readability, and professional presentation standards."""
+            )
+            
+            # 4. Visual Creation Agent - Uses all MCP tools (Text-to-Image focus)
+            visual_agent = CodeAgent(
+                tools=all_tools,
+                model=model,
+                add_base_tools=True,
+                name="visual_designer",
+                description="""Business visualization specialist creating professional infographics, charts, and presentation visuals.
+                Uses text-to-image tools to convert content into compelling visual formats suitable for executive presentations.
+                Focuses on clean, professional designs that enhance business storytelling."""
+            )
+            
+            # Business Content Manager - Coordinates all agents
+            business_content_manager = CodeAgent(
+                tools=all_tools,
+                model=model,
+                managed_agents=[research_agent, content_strategy_agent, content_formatter_agent, visual_agent],
+                additional_authorized_imports=["json", "re", "datetime"],
+                add_base_tools=True,
+                name="agentic_inkwell_manager",
+                description="""Agentic Inkwell Manager - Coordinates multi-agent business content creation workflow. 
+                Manages the complete pipeline from research to final formatted output with optional visual conversion.
+                Where specialized agents gather around the digital inkwell to craft intelligence together."""
+            )
+            
+            return business_content_manager
+    
+    # Content type options
+    content_types = [
+        "Market Analysis Report",
+        "Competitive Intelligence Brief", 
+        "Technology Trend Analysis",
+        "Product Launch Strategy",
+        "Investment Research Report",
+        "Strategic Planning Document"
+    ]
+    
+    # Writing style options
+    writing_styles = [
+        "Executive Summary (C-level audience)",
+        "Technical Brief (Developer/Engineer audience)", 
+        "Investor Pitch (VC/Stakeholder audience)",
+        "Market Research (Analyst audience)",
+        "Internal Memo (Team communication)"
+    ]
+    
+    # Visual style options
+    visual_styles = [
+        "Professional Infographics",
+        "Corporate Presentation Style",
+        "Minimalist Charts", 
+        "Executive Dashboard",
+        "Technical Diagrams",
+        "No Visuals (Text Only)"
+    ]
+    
+    with gr.Blocks(title="Agentic Inkwell", theme=gr.themes.Soft()) as demo:
+        gr.Markdown("# ✒️ Agentic Inkwell")
+        gr.Markdown("*Where Agents Craft Intelligence* - Generate comprehensive business reports with collaborative agentic writing")
         
-        # --- 5. The Creative Workflow ---
-        user_prompt = "a story about a wise old owl living in a library of forgotten books"
-
-        print(f"\n--- Director's Task ---")
-        print(f"Prompt: {user_prompt}\n")
-
-        final_output = director_agent.run(f"Create a story and illustration for the following prompt: {user_prompt}")
+        # Model Configuration Section
+        with gr.Row():
+            with gr.Column():
+                gr.Markdown("### 🤖 Model Configuration")
+                model_provider = gr.Dropdown(
+                    choices=[
+                        "Gemini (Google)",
+                        "Hugging Face (DeepSeek via Together)", 
+                        "DeepSeek (Direct API)"
+                    ],
+                    label="Select AI Model Provider",
+                    value="Gemini (Google)",
+                    info="Choose your preferred AI model provider"
+                )
+                
+                api_key_input = gr.Textbox(
+                    label="API Key / Token",
+                    placeholder="Enter your API key or token here...",
+                    type="password",
+                    info="Your API key will be used securely and not stored"
+                )
+                
+                gr.Markdown("""
+                **🔑 Where to get your API keys:**
+                - **Gemini**: Get free API key at [Google AI Studio](https://aistudio.google.com/app/apikey)
+                - **Hugging Face**: Get free token at [HF Settings](https://huggingface.co/settings/tokens)  
+                - **DeepSeek**: Get API key at [DeepSeek Platform](https://platform.deepseek.com/api_keys)
+                """, elem_classes=["api-info"])
         
-        print("\n--- Agentic Storycrafter Result ---")
+        with gr.Row():
+            with gr.Column(scale=2):
+                # Main input
+                topic_input = gr.Textbox(
+                    label="📋 Business Topic or Research Question",
+                    placeholder="e.g., 'AI Agent frameworks market analysis 2025' or 'NVIDIA vs AMD in AI chip market'",
+                    lines=3
+                )
+                
+                # Content controls
+                with gr.Row():
+                    content_type = gr.Dropdown(
+                        choices=content_types,
+                        label="📊 Content Type",
+                        value="Market Analysis Report"
+                    )
+                    
+                    writing_style = gr.Dropdown(
+                        choices=writing_styles,
+                        label="✍️ Writing Style", 
+                        value="Executive Summary (C-level audience)"
+                    )
+                
+                with gr.Row():
+                    visual_style = gr.Dropdown(
+                        choices=visual_styles,
+                        label="🎨 Visual Output",
+                        value="No Visuals (Text Only)"
+                    )
+                    
+                    include_sources = gr.Checkbox(
+                        label="📚 Include Source References",
+                        value=True
+                    )
+                
+                # Generate button
+                generate_btn = gr.Button("✒️ Craft with Agentic Inkwell", variant="primary", size="lg")
+                
+            with gr.Column(scale=1):
+                # Status and progress
+                status_box = gr.Textbox(
+                    label="📊 Generation Status",
+                    value="Ready to generate content...",
+                    interactive=False,
+                    lines=8
+                )
+        
+        # Output section
+        with gr.Row():
+            with gr.Column():
+                # Main content output
+                content_output = gr.Textbox(
+                    label="📄 Generated Business Content (Markdown)",
+                    lines=20,
+                    max_lines=30,
+                    show_copy_button=True
+                )
+                
+                # Action buttons row
+                with gr.Row():
+                    # Download button
+                    download_btn = gr.DownloadButton(
+                        label="💾 Download Markdown",
+                        visible=False
+                    )
+                    
+                    # Visual conversion button (appears after content generation)
+                    convert_visual_btn = gr.Button(
+                        "🎨 Convert to Images",
+                        variant="secondary",
+                        visible=False
+                    )
+        
+        # Visual output section (shown when visuals are generated)
+        with gr.Row(visible=False) as visual_output_row:
+            with gr.Column():
+                visual_status = gr.Textbox(
+                    label="🎨 Image Conversion Status",
+                    interactive=False,
+                    lines=3
+                )
+                
+                visual_output = gr.Gallery(
+                    label="🖼️ Generated Images",
+                    columns=2,
+                    height=500,
+                    show_label=True
+                )
+        
+        def generate_business_content(topic, content_type, writing_style, visual_style, include_sources, model_provider, api_key):
+            """Main function to coordinate business content generation"""
+            
+            if not topic.strip():
+                return "Please enter a business topic or research question.", "", None, gr.update(visible=False), gr.update(visible=False)
+            
+            if not api_key.strip():
+                return "Please provide your API key or token.", "", None, gr.update(visible=False), gr.update(visible=False)
+            
+            try:
+                # Initialize the user-selected model
+                yield "🤖 Initializing your selected AI model...", "", None, gr.update(visible=False), gr.update(visible=False)
+                
+                try:
+                    model = initialize_model(model_provider, api_key)
+                except Exception as e:
+                    error_msg = f"❌ Failed to initialize {model_provider}: {str(e)}"
+                    yield error_msg, "", None, gr.update(visible=False), gr.update(visible=False)
+                    return
+                
+                # Initialize agents with the user-selected model
+                yield "🔧 Setting up specialized agents...", "", None, gr.update(visible=False), gr.update(visible=False)
+                manager = initialize_agents(model)
+                
+                # Update status
+                status = "🔍 Starting business research..."
+                yield status, "", None, gr.update(visible=False), gr.update(visible=False)
+                
+                # Create detailed prompt for the manager
+                prompt = f"""
+                Create a comprehensive {content_type.lower()} about: {topic}
+                
+                Requirements:
+                - Writing Style: {writing_style}
+                - Include source references: {include_sources}
+                - Visual conversion needed: {visual_style != 'No Visuals (Text Only)'}
+                - Visual style: {visual_style if visual_style != 'No Visuals (Text Only)' else 'None'}
+                
+                Follow this workflow:
+                1. Use business_researcher to gather comprehensive market intelligence and competitive data
+                2. Use content_strategist to create structured business content with strategic insights  
+                3. Use content_formatter to format into professional markdown document
+                4. {"Use visual_designer to create professional visuals if requested" if visual_style != 'No Visuals (Text Only)' else "Skip visual generation"}
+                
+                Deliver a complete, professional business document ready for executive presentation.
+                """
+                
+                # Update status
+                status = "🤖 Coordinating multi-agent content creation..."
+                yield status, "", None, gr.update(visible=False), gr.update(visible=False)
+                
+                # Generate content using the manager
+                result = manager.run(prompt)
+                
+                # Update status
+                status = "✅ Content generation completed successfully!"
+                
+                # Prepare download file
+                import tempfile
+                import os
+                
+                temp_file = tempfile.NamedTemporaryFile(mode='w', suffix='.md', delete=False, encoding='utf-8')
+                temp_file.write(result)
+                temp_file.close()
+                
+                # Show/hide visual gallery based on whether visuals were generated
+                show_visuals = visual_style != 'No Visuals (Text Only)' and "![" in result
+                
+                yield (
+                    status, 
+                    result, 
+                    temp_file.name,
+                    gr.update(visible=True),  # download button
+                    gr.update(visible=True)   # convert to images button
+                )
+                
+            except Exception as e:
+                error_msg = f"❌ Error generating content: {str(e)}"
+                yield error_msg, "", None, gr.update(visible=False), gr.update(visible=False)
         
-        result_dict = eval(final_output)
-
-        story = result_dict.get("story")
-        image_data = result_dict.get("image_data")
-
-        print("\n--- STORY ---")
-        print(story)
 
-        if image_data and 'b64_json' in image_data:
-            print("\n--- ILLUSTRATION ---")
-            print("Illustration created. Saving to 'story_illustration.png'")
+        
+        def create_business_visuals(content_text, model_provider, api_key):
+            """Convert text content to images using T2l_text_to_images_and_base64_generator tool"""
+            
+            if not content_text.strip():
+                return "❌ No content available to convert", []
+            
+            if not api_key.strip():
+                return "❌ Please provide your API key or token", []
+            
             try:
-                img_bytes = base64.b64decode(image_data['b64_json'])
-                img = Image.open(io.BytesIO(img_bytes))
-                img.save("story_illustration.png")
-                print("Image saved successfully.")
+                # Initialize model for visual conversion
+                yield "🤖 Initializing model for image conversion...", []
+                
+                try:
+                    model = initialize_model(model_provider, api_key)
+                except Exception as e:
+                    yield f"❌ Failed to initialize model: {str(e)}", []
+                    return
+                
+                # Use already loaded MCP tools
+                yield "🎨 Converting text to images...", []
+                
+                # Create visual agent with already loaded tools
+                visual_agent = CodeAgent(
+                    tools=all_tools,
+                    model=model,
+                    add_base_tools=True,
+                    name="t2i_visual_converter"
+                )
+                
+                # Directly convert the content text to images
+                result = visual_agent.run(f"""
+                Use the T2l_text_to_images_and_base64_generator tool to convert this text to images:
+                
+                text_content: "{content_text[:1000]}"
+                aspect_ratio_str: "16:9 (Widescreen)"
+                font_size: 36
+                style: "plain"
+                bg_color_name: "White"
+                font_choice: "Arial"
+                
+                Convert the business content to professional images.
+                """)
+                
+                if result:
+                    yield "✅ Successfully converted text to images!", [("Business Content", result)]
+                else:
+                    yield "⚠️ No images were generated", []
+                
             except Exception as e:
-                print(f"Error saving image: {e}")
-        else:
-            print("\n--- ILLUSTRATION ---")
-            print("No illustration was generated.")
+                error_msg = f"❌ Image conversion failed: {str(e)}"
+                yield error_msg, []
+        
+        # Connect the generate button
+        generate_btn.click(
+            fn=generate_business_content,
+            inputs=[topic_input, content_type, writing_style, visual_style, include_sources, model_provider, api_key_input],
+            outputs=[status_box, content_output, download_btn, download_btn, convert_visual_btn],
+            show_progress=True
+        )
+        
+        # Connect the convert to images button - directly convert text to images
+        convert_visual_btn.click(
+            fn=create_business_visuals,
+            inputs=[content_output, model_provider, api_key_input],
+            outputs=[visual_status, visual_output],
+            show_progress=True
+        ).then(
+            fn=lambda: gr.update(visible=True),
+            outputs=visual_output_row
+        )
+        
+        # Example buttons for quick testing
+        with gr.Row():
+            gr.Markdown("### 🎯 Quick Examples:")
+            
+            example1_btn = gr.Button("📱 AI Smartphone Market", size="sm")
+            example2_btn = gr.Button("🚗 EV Battery Tech", size="sm") 
+            example3_btn = gr.Button("☁️ Cloud AI Services", size="sm")
+            
+            def set_example1():
+                return "AI-powered smartphone features market analysis 2025"
+            def set_example2(): 
+                return "Electric vehicle battery technology competitive landscape"
+            def set_example3():
+                return "Cloud-based AI services market opportunities and threats"
+                
+            example1_btn.click(fn=set_example1, outputs=topic_input)
+            example2_btn.click(fn=set_example2, outputs=topic_input)
+            example3_btn.click(fn=set_example3, outputs=topic_input)
+    
+    return demo
 
-# --- 6. Execution Start ---
 if __name__ == "__main__":
-    run_storycrafter()
-
+    # Launch the business content creation interface
+    demo = create_business_content_ui()
+    demo.launch()