WJZ-P/gemini-skill

gemini drawing MCP & skill through browser, can be used in openclaw or any agent that supports MCP. Gemini画图 MCP和sill，支持龙虾或任何agent使用٩(๑>◡<๑)۶

What it solves

Gemini Skill enables the automation of the Gemini web interface, allowing users to programmatically perform tasks such as AI image generation, text conversations, and image extraction without relying solely on an official API. It bridges the gap between AI agents and the Gemini web UI by providing a standardized way to control the browser.

How it works

The project uses the Chrome DevTools Protocol (CDP) via Puppeteer and a stealth plugin to control a browser instance. It operates through a "Daemon" architecture where a background process manages the browser's lifecycle, keeping it active for 30 minutes of inactivity before shutting down to save resources. It is implemented as an MCP (Model Context Protocol) server, meaning it can be plugged into AI clients (like Claude or CodeBuddy) to give them "skills" to interact with Gemini.

Who it’s for

  • AI Agent Developers: Those wanting to integrate Gemini's web capabilities (especially image generation) into their agent workflows.
  • Automation Enthusiasts: Users who want to automate repetitive tasks on the Gemini web platform.
  • Developers: Those looking for a reference on how to build MCP servers that control web interfaces.

Highlights

  • Full Image Workflow: Supports prompt-based image generation, uploading reference images for image-to-image tasks, and downloading high-resolution originals.
  • Automatic Watermark Removal: Includes a utility to automatically remove Gemini watermarks from downloaded images.
  • MCP Integration: Fully compatible with the Model Context Protocol, allowing any MCP-compliant client to trigger Gemini actions.
  • Smart Lifecycle Management: The browser daemon starts on demand and employs a lazy-destruction mechanism to avoid constant restarts.
  • Stealth Mode: Uses puppeteer-extra-plugin-stealth to bypass bot detection and mimic real user behavior.

Related

  • Project
    google-gemini/gemini-cli

    An open-source AI agent for the terminal that provides direct access to Gemini models for code analysis, automation, and system integration.

  • Project
    yeahhe365/Gemini-Nexus

    Gemini Nexus is a Chrome MV3 extension that adds a native AI assistant to the browser. It supports Google Gemini (both the web UI via reverse‑engineered tokens and the official API) and many OpenAI‑compatible services. The extension provides a side‑panel, a floating selection toolbar, image/screenshot input, and a Chrome‑DevTools‑based control loop so the LLM can navigate pages, click elements, fill forms, run scripts, and call external tool‑servers (MCP). It includes context‑management features, watermark removal for Gemini images, and a full build/publish workflow. The Gemini‑Web integration may violate Google’s Terms of Service, so users must accept the associated risk.

  • Project
    jamubc/gemini-mcp-tool

    An MCP server that connects AI assistants to the Gemini/Antigravity CLI, enabling large-scale codebase analysis and sandboxed code execution.

  • Project
    jacob-bd/gemini-notebook-mcp-cli

    A Python package that provides a `nlm` CLI and a local MCP server to programmatically manage Google Gemini Notebook (list/create notebooks, add sources, generate audio/video, etc.) and expose those actions to AI agents (Claude, Cursor, Gemini‑CLI, etc.). Install via `uv`/`pip`, log in with Google cookies (optionally encrypted), run `nlm setup` to connect your AI tools, then use simple commands or natural‑language prompts via the MCP server.

  • Project
    zhiyu1998/Gemi2Api-Server

    Gemi2Api‑Server is a FastAPI‑based bridge that lets you call Google Gemini via an OpenAI‑compatible REST API. It authenticates with Gemini using your browser cookies, forwards chat‑completion requests, optionally proxies generated images, and provides a simple admin UI. Deployable locally, via Docker, or on Render/HuggingFace, it’s useful for developers who want Gemini access without a formal public API.