Metatag Genie

Name: Metatag Genie
Author: terryso

@terryso

Repository

About Metatag Genie

一个基于 Node.js 和 TypeScript 的 MCP 服务，用于读写图片元数据

#metatag-genie #image-metadata #mcp-service

Basic information

Config

Add this server to your MCP-compatible client using the configuration below.

{
  "mcpServers": {
    "MetaTagGenie": {
      "command": "npx",
      "args": [
        "-y",
        "metatag-genie"
      ]
    }
  }
}

Tools

Writes metadata (tags, description, people, location) to an image file.

Overview

What is Metatag Genie?

Metatag Genie is an MCP (Model Context Protocol) server for macOS that writes image metadata to enhance Spotlight search. It communicates over Stdio and can be called by AI agents or other applications needing local image metadata management.

How to use Metatag Genie?

Install via npm install -g metatag-genie and run metatag-genie, or run directly with npx metatag-genie. For integration in Cursor, create .cursor/mcp.json with the command npx -y metatag-genie. The server listens on Stdio and expects JSON-RPC 2.0 messages.

Key features of Metatag Genie

Supports writing metadata to JPG, PNG, and HEIC images
Writes tags, description, people (as keywords), and location (text)
Written metadata is indexed by macOS Spotlight
Uses ExifTool (managed automatically via exiftool-vendored)
Communicates over Stdio with MCP-compliant JSON-RPC 2.0

Use cases of Metatag Genie

Add descriptive tags to a photo collection for better Spotlight search
Enrich vacation pictures with location and people names automatically via an AI agent
Embed metadata in images before sharing or archiving them
Integrate with Cursor or other AI editors to tag screenshots or design assets

FAQ from Metatag Genie

Which image formats are supported?

JPG, PNG, and HEIC files are supported.

What runtime is required?

Node.js 22.x LTS or higher is required.

Are the written metadata actually searchable by macOS Spotlight?

Yes, the metadata written (tags, description, people, location) can be found by macOS Spotlight.

Does the server listen on a network port?

No, the server only communicates via standard input/output (Stdio). It does not bind a network port.

How do I integrate Metatag Genie with Cursor?

Create a .cursor/mcp.json file in your project root with the command npx -y metatag-genie. The AI assistant in Cursor can then use the writeImageMetadata tool.