Which SDK should I use for Google Gemini in 2025?

You must use the @google/genai SDK. The older @google/generative-ai SDK is deprecated and scheduled for sunset in late 2025.

What is the context window for Gemini 2.5 models?

Gemini 2.5 Pro and Flash models support a 1,048,576 (1M) token context window, providing massive capacity for large document or codebase analysis.

Does this skill support multimodal file inputs?

Yes, it includes specific implementation patterns and MIME type configurations for images, video, audio, and PDF files.

What is 'Thinking Mode' in the Gemini 2.5 models?

Thinking Mode allows the model to use an internal reasoning budget to solve complex math, logic, or coding problems before providing a final answer.

Is Gemini 2.5 Flash-Lite supported for function calling?

No, the Flash-Lite model is optimized for cost and latency but does not support function calling or code execution.

Google Gemini API Integration

Name: Google Gemini API Integration
Author: secondsky

bysecondsky

•

データサイエンスとML

Integrates the 2025 Google Gemini 2.5 API using the modern @google/genai SDK for multimodal reasoning and long-context AI workflows.

概要

This skill provides a production-ready implementation of the Google Gemini API, specifically designed to facilitate the migration from the deprecated @google/generative-ai SDK to the modern @google/genai v1.27+ standard. It equips Claude Code with optimized patterns for the Gemini 2.5 model series—including Pro, Flash, and Flash-Lite—enabling high-performance features such as 1M token context windows, built-in 'Thinking Mode' for complex reasoning, parallel function calling, and native multimodal support for images, video, and PDFs.

主な機能

Advanced multimodal processing for images, video, audio, and PDF documents
Optimized configurations for Gemini 2.5 Pro, Flash, and Flash-Lite
Production patterns for parallel and compositional function calling
Context caching and thinking mode optimization for complex reasoning tasks
Comprehensive migration paths from deprecated @google/generative-ai SDK
21 GitHub stars

ユースケース

Building agentic workflows that require 1M+ token context windows for large codebase analysis
Implementing multimodal document and media analysis within serverless environments like Cloudflare Workers
Migrating legacy Google AI integrations to the 2025 GenAI SDK standard before sunset dates

概要

主な機能

Advanced multimodal processing for images, video, audio, and PDF documents
Optimized configurations for Gemini 2.5 Pro, Flash, and Flash-Lite
Production patterns for parallel and compositional function calling
Context caching and thinking mode optimization for complex reasoning tasks
Comprehensive migration paths from deprecated @google/generative-ai SDK
21 GitHub stars

ユースケース

Building agentic workflows that require 1M+ token context windows for large codebase analysis
Implementing multimodal document and media analysis within serverless environments like Cloudflare Workers
Migrating legacy Google AI integrations to the 2025 GenAI SDK standard before sunset dates