img-mcp

Appearance

img-mcpDocumentation

Open-source MCP server for AI image generation and local image manipulation. Connect your agents to Google's Nano Banana models or process images locally with Node.js.

Get started
View on GitHub
Illustration of an AI agent workspace with image generation and local Node.js processing
🎨

Text-to-image

Generate images from natural-language prompts with controls for aspect ratio, resolution, and output format.

🖼️

Reference images

Guide generation by passing base64 images or data URIs to preserve style, subject, or composition.

✨

Transparent backgrounds

Produce clean PNG assets with transparent backgrounds ready for logos, overlays, and product shots.

📐

Resize & convert

Scale, crop, and convert formats locally with sharp. No API key or network call required.

🔑

Bring your own key

Your Google API key never leaves your machine. No shared credentials or proxy servers.

⚡

Zero-install usage

Run instantly with npx -y img-mcp. No global install or container setup required.

What is img-mcp? ​

img-mcp is an open-source Model Context Protocol (MCP) server that lets AI agents generate and manipulate images. It connects any MCP-compatible host to Google's Nano Banana image-generation models (Gemini Flash Image) through the Gemini API, and runs fast local operations with sharp and Node.js.

Because img-mcp speaks the standard MCP protocol, you can add image generation and manipulation to Claude, Cursor, Windsurf, Kimi, OpenCode, or the OpenAI Codex CLI with a single configuration block.

Quick start ​

For local image manipulation, no API key is needed:

bash
npx -y img-mcp

For AI image generation, set a Google API key first:

bash
export GOOGLE_API_KEY="your-google-api-key"
npx -y img-mcp

Then add the server to your MCP host and ask your agent to resize, convert, or generate an image.

Learn more ​

Released under the MIT License.

Copyright © 2026 Aboubakar Sidik Faha