mcp-openvision

# System Patterns ## Simplified Architecture MCP OpenVision follows a straightforward architecture: ```mermaid flowchart TD Client[Client Applications] MCP[MCP Server Layer] OpenRouter[OpenRouter API] Client --> MCP MCP --> OpenRouter subgraph "MCP OpenVision" MCP end ``` ## Core Components ### MCP Server Module - **Purpose**: Provide MCP-compliant tools for image analysis - **Pattern**: Follows standard MCP tool definitions - **Key File**: `server.py` ### Essential Utilities - **Error Handling**: Simple exception hierarchy - **Configuration**: Basic environment variable support - **Image Processing**: Minimal preprocessing helpers ## Data Flow The system follows a simple request-response flow: ```mermaid sequenceDiagram Client->>MCP Server: Image Analysis Request MCP Server->>OpenRouter: Vision API Request OpenRouter->>MCP Server: Analysis Results MCP Server->>Client: Formatted Results ``` ## Error Handling Approach Basic error handling pattern: ```mermaid flowchart TD Error[Error Occurs] --> Handled{Can be handled?} Handled -->|Yes| RetryOrContinue[Retry or Continue] Handled -->|No| TransformError[Transform to User-Friendly Error] TransformError --> ReturnError[Return Error to Client] ``` ## Configuration Simple configuration from environment variables: ```mermaid flowchart TD Config[Configuration] Param[Function Parameters] Env[Environment Variables] Default[Default Values] Config --> Param Param -->|Not Set| Env Env -->|Not Set| Default ``` ## Testing Strategy Focus on core functionality testing: ```mermaid flowchart TD Unit[Unit Tests] Mock[Mocked API Tests] Unit --> Mock ``` ## MCP Compliance The project strictly adheres to MCP patterns: 1. **Tool Definitions**: Using `@mcp.tool()` decorators 2. **Type Annotations**: Following MCP type definitions 3. **Async Patterns**: Using proper async/await patterns 4. **Error Handling**: Returning informative errors