Gemini MCP Server

# Enhanced Gemini MCP Server Features (v4.0.0) This enhanced version of the Gemini MCP server includes all the latest features from Google's Gemini API as of July 2025. ## Available Models ### Gemini 2.5 Series (Latest - With Thinking Capabilities) - **gemini-2.5-pro** - Most capable thinking model with 2M token context - **gemini-2.5-flash** - Fast thinking model with best price/performance (1M tokens) - **gemini-2.5-flash-lite** - Ultra-fast, cost-efficient thinking model ### Gemini 2.0 Series - **gemini-2.0-flash** - Fast, efficient model with 1M context window - **gemini-2.0-flash-lite** - Most cost-efficient model - **gemini-2.0-pro-experimental** - Experimental model with 2M context, excellent for coding ### Legacy Models - **gemini-1.5-pro** - Previous generation pro model (2M tokens) - **gemini-1.5-flash** - Previous generation fast model (1M tokens) ## Available Tools ### 1. `generate_text` - Advanced Text Generation Generate text with all the latest Gemini features: - **Model Selection**: Choose any available Gemini model - **System Instructions**: Guide model behavior with system prompts - **Temperature Control**: Fine-tune creativity (0-2) - **Advanced Sampling**: Control with topK and topP parameters - **JSON Mode**: Get structured JSON output with optional schema validation - **Google Search Grounding**: Get up-to-date information from the web - **Safety Settings**: Configure content filtering per category - **Conversation Memory**: Maintain context across multiple turns ### 2. `analyze_image` - Vision Analysis Analyze images using Gemini's vision capabilities: - Support for image URLs or base64-encoded images - Compatible with all vision-capable models - Natural language understanding of visual content ### 3. `count_tokens` - Token Counting Count tokens for any text with a specific model: - Accurate token counting for cost estimation - Model-specific tokenization ### 4. `list_models` - Model Discovery List all available models with filtering: - Filter by capabilities (thinking, vision, grounding, json_mode) - View model descriptions and context windows - Check feature availability ### 5. `embed_text` - Text Embeddings Generate embeddings for semantic search and similarity: - Latest embedding models (text-embedding-004) - Multilingual support - High-dimensional vectors for accuracy ## Available Resources - **gemini://models** - Detailed list of all available models - **gemini://capabilities** - Comprehensive API capabilities documentation ## Available Prompts ### 1. `code_review` - Comprehensive Code Review Use Gemini 2.5 Pro's thinking capabilities for in-depth code analysis ### 2. `explain_with_thinking` - Deep Explanations Leverage thinking models for thorough explanations of complex topics ### 3. `creative_writing` - Creative Content Generation Generate creative content with style and length control ## Advanced Features ### Thinking Models The Gemini 2.5 series includes "thinking" capabilities that allow models to reason through problems step-by-step before responding, resulting in more accurate and thoughtful outputs. ### JSON Mode with Schema Validation When `jsonMode` is enabled, you can provide a JSON schema to ensure the output matches your exact requirements. ### Google Search Grounding Enable real-time web search to ground responses in current information, perfect for: - Current events - Technical documentation - Fact-checking - Up-to-date information ### Multi-turn Conversations Maintain conversation context using `conversationId` to build more coherent, contextual interactions. ### Safety Configuration Fine-tune safety settings per request with granular control over: - Harassment - Hate speech - Sexually explicit content - Dangerous content ## Example Usage ### Advanced Text Generation with All Features ```json { "tool": "generate_text", "arguments": { "prompt": "Explain quantum computing", "model": "gemini-2.5-pro", "systemInstruction": "You are a physics professor explaining to undergraduate students", "temperature": 0.8, "maxTokens": 2048, "jsonMode": true, "jsonSchema": { "type": "object", "properties": { "explanation": { "type": "string" }, "key_concepts": { "type": "array", "items": { "type": "string" } }, "difficulty_level": { "type": "number", "minimum": 1, "maximum": 10 } } }, "grounding": true, "conversationId": "quantum-discussion-001" } } ``` ### Image Analysis ```json { "tool": "analyze_image", "arguments": { "prompt": "What's happening in this image? Describe any text, objects, and activities.", "imageBase64": "data:image/jpeg;base64,/9j/4AAQSkZJRg...", "model": "gemini-2.5-flash" } } ``` ## Best Practices 1. **Model Selection**: - Use `gemini-2.5-flash` for most tasks (best balance) - Use `gemini-2.5-pro` for complex reasoning and coding - Use `gemini-2.5-flash-lite` for high-volume, simple tasks 2. **Thinking Models**: - Enable for tasks requiring deep reasoning - Expect slightly longer response times but better quality 3. **Context Management**: - Use conversation IDs for multi-turn interactions - Monitor token usage with the count_tokens tool 4. **Safety and Grounding**: - Enable grounding for current information needs - Adjust safety settings based on your use case