MCP Image Recognition Server

# Gemini API Key (Recommended for Gemini models) # GEMINI_API_KEY=your_gemini_key_here # ^ Uncomment this line and replace 'your_gemini_key_here' with your actual Gemini API Key. # To configure a custom Gemini API server, you would typically set a base URL # in the client or directly in the 'gemini' provider's initialization if it supports it. # However, for the `google-generativeai` library, the API endpoint is usually implicit # or configured via client options, not directly through an environment variable like BASE_URL. # OpenAI / Compatible API Key OPENAI_API_KEY=your_key_here # ^ Replace 'your_key_here' with your actual API Key for OpenAI-compatible services (e.g., OpenAI, Qwen, Doubao). # OpenAI Base URL (Optional - defaults to OpenAI's official API endpoint) # This is crucial for configuring custom API servers for OpenAI-compatible models. # For Qwen (Tongyi Qianwen - Aliyun DashScope): # OPENAI_BASE_URL=https://dashscope.aliyuncs.com/compatible-mode/v1 # For Doubao (Volcengine Ark): # OPENAI_BASE_URL=https://ark.cn-beijing.volces.com/api/v3 OPENAI_BASE_URL=https://api.openai.com/v1 # Default Model (Optional - will be used if 'model' parameter is not provided in a tool call) DEFAULT_MODEL=gemini-1.5-flash