Module 2 (Gemini) - Lesson 2: Gemini Prompts Overview

Gemini Prompting: From Basic to Advanced

This lesson maps the Gemini lessons you will complete in this module and highlights the key API differences versus OpenAI and Anthropic.

Lesson Map

• Lesson 2a: Basic Prompt
• Lesson 2b: System Prompt
• Lesson 2c: Temperature Control
• Lesson 2d: Extended Prompts
• Lesson 2e: Streaming Responses
• Lesson 2f: Structured Output
• Lesson 2g: Tools and Function Calling

Gemini vs Other Providers

• Method: gemini.models.generateContent instead of anthropic.messages.create or openai.responses.create
• Request body: contents can be a string or { role, parts } object
• System behavior: config.systemInstruction replaces Anthropic system
• Streaming: generateContentStream returns an async iterator of chunks with text and usageMetadata
• Structured output: config.responseMimeType: "application/json" plus responseJsonSchema for light validation
• Tools: Declare tools with functionDeclarations; function calls arrive in candidates[0].content.parts

Quick Start Checklist

• ✅ Add GEMINI_API_KEY to your .env
• ✅ Install the SDK: pnpm add @google/genai
• ✅ Run the first example: pnpm tsx src/gemini/basic-prompt.ts

Continue to Lesson 2a to send your first Gemini prompt.