Update support documentation to reflect current interface structure

- Add detailed three-tab interface description with clear workflow steps
- Add comprehensive Preview tab usage section with testing guidance
- Restructure tool settings to match current Configuration tab layout
- Update workflow descriptions to align with actual interface organization

support_docs.py

@@ -17,10 +17,18 @@ def create_support_docs():
             gr.Markdown("""
             ### Quick Start Guide
             
-            1. **Configure your Space** in the main Configuration tab
-            2. **Preview your assistant** using the Preview button
-            3. **Generate deployment package** with the Generate button
-            4. **Deploy to HuggingFace Spaces** following the README instructions
+            **Three-Tab Interface:**
+            1. **Configuration Tab**: Set up your space settings, assistant configuration, and tool integrations
+            2. **Preview Tab**: Test your assistant with real API integration before deployment  
+            3. **Support Tab**: Access comprehensive documentation and help (this tab)
+            
+            **Workflow Steps:**
+            1. **Configure your Space** in the Configuration tab (space title, description, model selection)
+            2. **Set up Assistant** with system prompt and optional research template
+            3. **Enable Tools** like web search, document RAG, or URL grounding as needed
+            4. **Preview & Test** using the Preview tab to validate your configuration
+            5. **Generate Package** with the "Generate Deployment Package" button
+            6. **Deploy to HuggingFace** following the included README instructions
             
             **Prerequisites:**
             - HuggingFace account (free at huggingface.co)
@@ -116,6 +124,39 @@ def create_support_docs():
             ```
             """)
         
+        with gr.Accordion("🔬 Preview Tab Usage", open=False):
+            gr.Markdown("""
+            ### Testing Your Assistant Before Deployment
+            
+            The Preview tab provides a sandbox environment to test your assistant with real API integration.
+            
+            **How to Use the Preview:**
+            1. **Configure First**: Complete your setup in the Configuration tab
+            2. **Click Preview Button**: Use "Preview Deployment Package" to activate the preview
+            3. **Test Chat Interface**: Interactive chat with your exact configuration
+            4. **Test URL Context**: Add URLs to test grounding functionality
+            5. **Export Conversations**: Save chat logs for analysis
+            
+            **Preview Features:**
+            - **Real API Integration**: Uses actual OpenRouter API when `OPENROUTER_API_KEY` is set
+            - **Configuration Display**: Shows your current assistant setup
+            - **URL Testing**: Add up to 4 URLs for context testing
+            - **Dynamic URL Management**: Add/remove URL fields as needed
+            - **Chat Export**: Download conversation logs as markdown files
+            - **Clear Function**: Reset chat history for new tests
+            
+            **Preview Requirements:**
+            - Set `OPENROUTER_API_KEY` environment variable for real API testing
+            - Without API key: Shows configuration but no actual chat responses
+            - All other features (URL fetching, configuration) work without API key
+            
+            **Testing Best Practices:**
+            - Test different types of queries to validate assistant behavior
+            - Try URL grounding with actual websites to test context integration
+            - Verify that system prompt and model settings work as expected
+            - Export successful conversations to use as examples for documentation
+            """)
+        
         with gr.Accordion("💬 Example Prompts", open=False):
             gr.Markdown("""
             ### Creating Effective Example Prompts
@@ -157,36 +198,57 @@ def create_support_docs():
             ```
             """)
         
-        with gr.Accordion("🔧 Tool Settings & Grounding", open=False):
+        with gr.Accordion("🔧 Tool Settings & Configuration", open=False):
             gr.Markdown("""
-            ### Tool Configuration Options
-            
-            **Dynamic URL Fetching**
-            - Allows assistant to fetch content from URLs mentioned in conversations
-            - Useful for research and fact-checking tasks
-            - Automatically extracts text content from web pages
-            - Limited to 3 URLs per message for performance
-            
-            **Document RAG (Retrieval-Augmented Generation)**
-            - Upload PDF, DOCX, TXT, or MD files as knowledge base
-            - Assistant can search and reference uploaded documents
-            - Perfect for course materials, policies, or reference documents
-            - Files are embedded in your deployment package
-            
-            **URL Grounding**
-            - Add up to 4 URLs that provide static context
-            - Content is fetched once and included in every response
-            - Great for course syllabi, institutional policies, or reference materials
-            - Content is cached to avoid repeated fetching
-            
-            **When to Use Each:**
-            - **Static URLs**: Course policies, syllabi, reference materials
-            - **Dynamic URLs**: Research tasks where students provide links
-            - **Document RAG**: Large document collections, textbooks, course readers
-            
-            **File Size Limits:**
-            - Individual files: 10MB maximum
-            - Total RAG package: Recommended under 50MB for deployment
+            ### Configuration Tab Structure
+            
+            **Main Configuration Fields** (always visible):
+            - **Space Title**: Display name for your HuggingFace Space
+            - **Space Description**: Brief explanation shown in space listing
+            - **Model Selection**: Choose from optimized OpenRouter models
+            - **API Key Variable**: Secret name for HuggingFace Space settings (default: `OPENROUTER_API_KEY`)
+            - **Access Code**: Optional password protection for student access
+            
+            **Assistant Configuration Accordion** (open by default):
+            - **System Prompt**: Main field defining assistant behavior and knowledge
+            - **Research Template**: Pre-configured academic research assistant checkbox
+            - **Web Search Integration**: Enable crawl4ai web search capabilities
+            - **Document RAG**: Upload documents for knowledge base (PDF/DOCX/TXT/MD support)
+            - **URL Grounding**: Add up to 4 static URLs for context (dynamic add/remove)
+            - **Example Prompts**: Clickable suggestions for users (one per line)
+            - **Dynamic URL Fetching**: Hidden field (always enabled) for runtime URL processing
+            
+            **Advanced Settings Accordion**:
+            - **Temperature**: Response creativity control (0.0-2.0)
+            - **Max Tokens**: Response length limit (50-4096)
+            
+            **Action Buttons**:
+            - **Preview Deployment Package**: Activate Preview tab testing
+            - **Generate Deployment Package**: Create downloadable zip file
+            
+            ### Tool Integration Options
+            
+            **Web Search (crawl4ai)**
+            - Real-time web searching using DuckDuckGo
+            - Advanced content extraction and crawling
+            - Automatically enabled with Research Template
+            
+            **Document RAG (Vector Search)**
+            - Upload files: PDF, DOCX, TXT, MD (10MB max each)
+            - Semantic chunking and FAISS vector search
+            - Embedded in deployment package for offline use
+            - Requires `sentence-transformers` and `faiss-cpu`
+            
+            **URL Grounding (Static Context)**
+            - Add 2-4 URLs for consistent context across all responses
+            - Content fetched once during generation and cached
+            - Dynamic add/remove buttons for scalable input
+            - Perfect for course syllabi, policies, reference materials
+            
+            **Dynamic URL Fetching (Runtime)**
+            - Always enabled - processes URLs mentioned in conversations
+            - Extracts content from user-provided links automatically
+            - Limited to reasonable content size for performance
             """)
         
         with gr.Accordion("🎛️ Advanced Settings", open=False):