Генерация изображений и текста через Gemini Web

★ 7.3 · ai-tooling

baoyu-danger-gemini-web is a Claude Code skill that generates images and text through a reverse-engineered Gemini Web API, requiring no official API key. It supports text generation, image creation from prompts, vision input via reference images, and multi-turn conversations tracked by `--sessionId`. The skill ships a TypeScript port of `gemini_webapi` with `scripts/main.ts` as the CLI entry point, runs via `bun` or `npx`, and exposes models including gemini-3-pro, gemini-3-flash, gemini-3-flash-thinking, and gemini-3.1-pro-preview. Authentication is handled through a local browser session (Chrome, Chromium, or Edge) with automatic cookie caching; explicit user consent for unofficial API usage is required before the first run. It is designed to act as an image-generation backend for other skills and for developers who need Gemini's vision capabilities outside the official API.