MCP Server Replicate

# MCP Server Workflow Patterns ## Text-to-Image Generation The text-to-image workflow is designed to help users find and use the right model for their specific needs. The workflow follows these steps: 1. **Initial User Input** - User provides their requirements through the `text_to_image` prompt - System guides them to specify: - Subject matter - Style preferences - Quality requirements - Technical requirements (size, etc.) 2. **Model Discovery** - System uses `search_available_models` to find suitable models - Models are scored based on: - Popularity (run count) - Featured status - Version stability - Tag matching - Task relevance - Results are sorted but presented to user for selection 3. **Model Selection** - User can view model details using `get_model_details` - System provides guidance on model capabilities - User makes final model selection - System confirms version to use 4. **Parameter Configuration** - System applies quality presets based on user requirements - Style presets are applied if specified - Size and other parameters are configured - User can override any parameters 5. **Image Generation** - System uses `generate_image` with selected model and parameters - Progress is tracked - Results are validated - Retry logic handles failures ### Example Flow ```python # 1. User provides requirements through text_to_image prompt response = await text_to_image() # 2. Search for models based on requirements models = await search_available_models( query="cat portrait", style="photorealistic" ) # 3. Get details for user's chosen model model = await get_model_details("stability-ai/sdxl") # 4. Generate image with chosen model result = await generate_image( model_version=model.latest_version.id, prompt="a photorealistic cat portrait", style="photorealistic", quality="balanced" ) ``` ### Key Principles 1. **User Agency** - System suggests but doesn't decide - User makes final model selection - Parameters can be overridden 2. **Transparency** - Model scoring is visible - Capabilities are clearly communicated - Limitations are disclosed 3. **Flexibility** - Multiple models can be used - Parameters are customizable - Quality/style presets are optional 4. **Reliability** - Version stability is considered - Error handling is robust - Progress is tracked ## Template Usage Templates provide consistent parameter sets but should be used flexibly: 1. **Quality Presets** - Provide baseline parameters - Can be overridden - Match user's speed/quality needs 2. **Style Presets** - Enhance prompts - Add style-specific parameters - Are optional and customizable 3. **Aspect Ratio Presets** - Match common use cases - Ensure valid dimensions - Can be customized ## Error Handling 1. **Model Selection** - Handle no matches gracefully - Provide alternatives - Explain limitations 2. **Parameter Validation** - Validate before submission - Provide clear error messages - Suggest corrections 3. **Generation Failures** - Implement retry logic - Track progress - Provide status updates ## Best Practices 1. **Model Selection** - Always let user choose model - Provide scoring context - Explain trade-offs 2. **Parameter Configuration** - Use presets as starting points - Allow customization - Validate combinations 3. **Error Handling** - Be proactive about potential issues - Provide clear error messages - Implement proper retry logic 4. **User Interaction** - Guide don't decide - Explain options - Respect user choices ``` ```