Tuzi Danger Gemini Web

This is a reverse-engineered Gemini Web API client that handles both text and image generation through Google's web interface instead of the official API. It supports vision input with reference images, multi-turn conversations with session management, and runs locally via TypeScript. You'll need to authenticate through Chrome on first run, and there's a consent flow before use since it relies on unofficial endpoints. Useful when you want Gemini's image generation capabilities without dealing with official API quotas or pricing, though the reverse-engineered nature means it could break if Google changes their web interface. Supports multiple Gemini 3.x models including the thinking variant.

Install to Claude Code

npx -y skills add tuziapi/tuzi-skills --skill tuzi-danger-gemini-web --agent claude-code