Add Google Gemini MCP server with 2M context, multimodal AI, and function calling

[adampiispanen]

Add Google Gemini MCP Server

Google's multimodal AI with 2M token context window supporting text, images, video, audio, and PDF analysis.

Features

• ✅ Text Generation - Gemini 1.5 Pro, Flash, and 2.0 experimental models
• ✅ Multi-Turn Chat - Conversational AI with context retention
• ✅ Vision Analysis - Image understanding and description
• ✅ Video Analysis - Frame-by-frame video content analysis
• ✅ PDF Processing - Extract and analyze PDF documents
• ✅ Function Calling - Tool use and structured outputs
• ✅ Text Embeddings - text-embedding-004 for semantic search
• ✅ Streaming - Real-time response streaming
• ✅ Token Counting - Estimate costs before generation
• ✅ Batch Generation - Parallel processing for efficiency
• ✅ JSON Mode - Structured output generation

Configuration

• Authentication: Simple API key (no OAuth)
• Transport: streamable-http
• Resources: 512Mi memory, 500m CPU
• Free Tier: 15 RPM, 1M tokens/minute, 1500 requests/day

Use Cases

• Multimodal AI applications
• Document analysis and extraction
• Video content understanding
• Image captioning and analysis
• Long-context document processing (2M tokens!)
• Function calling for agentic workflows
• Semantic search with embeddings

Models

• gemini-1.5-pro - Highest quality, 2M context
• gemini-1.5-flash - Fast and efficient
• gemini-2.0-flash-exp - Latest experimental features
• text-embedding-004 - Text embeddings

Key Advantages

• 2 million token context window (largest available)
• True multimodal support (text, image, video, audio, PDF)
• Built-in function calling
• Competitive with GPT-4 and Claude

Validation

✅ All validations pass: npm run validate-servers