Claude Code + Gemini MCP Server

# Claude-Gemini MCP Usage Examples ## Basic Conversation ```bash # Start Claude Code claude # Ask Gemini a simple question mcp__gemini-collab__ask_gemini prompt: "What is the capital of France?" # Gemini responds directly in Claude's context! ``` ## Code Review Example ```bash # Have Gemini review your authentication code mcp__gemini-collab__gemini_code_review code: | def authenticate(username, password): if username == "admin" and password == "password123": return True return False focus: "security" # Gemini will point out security issues like: # - Hardcoded credentials # - Plain text password # - No hashing # - etc. ``` ## Brainstorming Session ```bash # Brainstorm startup ideas mcp__gemini-collab__gemini_brainstorm topic: "AI-powered tools for developers" context: "Looking for B2B SaaS ideas that solve real developer pain points" # Gemini provides creative suggestions ``` ## Advanced: Collaborative Problem Solving ```bash # Claude writes code > Write a Python function to calculate fibonacci numbers # Claude creates the function... # Then get Gemini's optimization suggestions mcp__gemini-collab__gemini_code_review code: "[paste Claude's code here]" focus: "performance" # Claude can then incorporate Gemini's feedback! ``` ## Temperature Control ```bash # Low temperature (0.2) for factual, consistent responses mcp__gemini-collab__ask_gemini prompt: "Explain the HTTP protocol" temperature: 0.2 # High temperature (0.8) for creative responses mcp__gemini-collab__ask_gemini prompt: "Write a haiku about debugging" temperature: 0.8 ``` ## Real-World Workflow 1. **Claude writes initial code** 2. **Gemini reviews for security/performance** 3. **Claude implements improvements** 4. **Both AIs brainstorm edge cases** 5. **Final optimized solution!** This creates a powerful AI pair programming experience where both models complement each other's strengths.