This is a reverse-engineered Gemini Web API client that lets you generate both text and images without going through official channels. It handles multi-turn conversations, accepts reference images for vision tasks, and supports the full Gemini 3.0 model lineup including the thinking variant. The consent flow is thorough, maybe overly so, but makes sense given you're using an unofficial API that could break anytime. Useful when you need Gemini's image generation capabilities in your Claude workflow or want to compare outputs across models. Authentication happens through browser cookie extraction, and it caches everything locally. If Google's official API pricing bothers you or you need features not yet in the official release, this fills the gap.
npx skills add https://github.com/xy121718/baoyu-skills --skill baoyu-danger-gemini-web