Ezkielyna Image

Ezkielyna Image MCP

An MCP server that lets your AI assistant create and edit images from files on your computer: Claude Desktop, Claude Code, Cursor, OpenCode, or any other MCP client that runs local stdio servers. Images are generated through the Ezkielyna API.

Example: "My product photo is ~/Shop/mug.jpg and the logo is ~/Shop/logo.png. Put the logo on the mug and save it as mug-logo."

🇮🇩 Bahasa Indonesia: lihat bagian di bawah.

• One file, no dependencies: only Node.js 18 or newer. No npm install needed.
• Reference images come from local files (or https links); results are saved locally.
• Results can be edited again: the assistant just passes the previous result path.
• Paid calls are capped per session, the API key is never printed, and existing files are never overwritten.

npm: ezkielyna-image-mcp. Your app can start it with npx -y ezkielyna-image-mcp, no clone needed. Listed in the official MCP registry as io.github.EzkielynaGroup/ezkielyna-image-mcp.

Requirements

1. Node.js 18+ (node -v). Get the LTS from nodejs.org.
2. An Ezkielyna API key in group Media (the Auto group also includes image models). Buy API Credit at ezkielyna.store (from $0.5, QRIS/DANA/GoPay/USDT), redeem it under Wallet at api.ezkielyna.store, then create a key under API Keys. Standard/Premium keys cannot use image models.
3. Some balance. Prices per image are shown by the list_image_models tool and on ezkielyna.store/models.

Install

With npx (recommended): nothing to install. The configs below run npx -y ezkielyna-image-mcp, which downloads the package (one file, no dependencies) on first start.

From source (to read or change the code):

git clone https://github.com/EzkielynaGroup/ezkielyna-image-mcp.git
cd ezkielyna-image-mcp
node --test test/        # optional: runs against a local mock, no network, no cost
pwd                      # the full path of ezkielyna-image-mcp.mjs is <this folder>/ezkielyna-image-mcp.mjs

On Windows, if the app cannot start npx, use "command": "cmd" with "args": ["/c", "npx", "-y", "ezkielyna-image-mcp"]. For a source install, write paths in JSON with forward slashes (C:/Users/you/ezkielyna-image-mcp/ezkielyna-image-mcp.mjs) or double backslashes.

Configuration (environment)

Variable	Required	Default	Notes
EZKIELYNA_API_KEY	yes		key in group Media (or Auto)
EZKIELYNA_BASE_URL	no	https://api.ezkielyna.store/v1	API address
EZKIELYNA_IMAGE_DIR	no	~/Pictures/Ezkielyna	output folder (created if missing)
EZKIELYNA_MAX_IMAGES	no	30	paid images per session, a guard against runaway loops. Restart the app to reset.
EZKIELYNA_TIMEOUT_MS	no	300000	request timeout

Add it to your app

Replace sk-... with your Media key. Installed from source? Use "command": "node" with "args": ["/path/to/ezkielyna-image-mcp.mjs"] instead of npx.

Claude Desktop: Settings → Developer → Edit Config (claude_desktop_config.json), then fully quit and reopen Claude Desktop.

{
  "mcpServers": {
    "ezkielyna-image": {
      "command": "npx",
      "args": ["-y", "ezkielyna-image-mcp"],
      "env": { "EZKIELYNA_API_KEY": "sk-..." }
    }
  }
}

Claude Code

claude mcp add ezkielyna-image -e EZKIELYNA_API_KEY=sk-... -- npx -y ezkielyna-image-mcp

Add --scope user to make it available in every project. Check with claude mcp list.

Cursor: ~/.cursor/mcp.json (all projects) or .cursor/mcp.json (one project), same JSON as Claude Desktop.

OpenCode: opencode.json in the project, or ~/.config/opencode/opencode.json for all projects.

{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "ezkielyna-image": {
      "type": "local",
      "command": ["npx", "-y", "ezkielyna-image-mcp"],
      "enabled": true,
      "environment": { "EZKIELYNA_API_KEY": "sk-..." }
    }
  }
}

Tools

Tool	What it does	Cost
generate_image	New image from text. Options: model (default gpt-image-2.5), size (e.g. 1024x1024), filename.	per image
edit_image	Edit or combine 1–4 images (local paths or https links). Options: model, quality (gpt-image only), size, filename.	per image
list_image_models	Image models your key can use, price per image, and which ones can edit.	free

Edit models: gpt-image-2.5 (default, up to 4 references, good for logos and backgrounds) and nano-banana-pro (up to 4 references, high detail, more expensive). Keep the total upload under ~30 MB (~22 MB for nano-banana-pro, which is sent as base64); resize large phone photos to about 2000 px JPEG first.

Each result includes the file path, model and price, and the image itself is shown inline when it is 3.5 MB or smaller.

Security

• Reads only .png, .jpg, .jpeg, .webp files up to 20 MB whose content really is an image; links must be https.
• Writes only inside the output folder, sanitises file names and never overwrites (name-1.png, name-2.png, ...).
• The key is scrubbed from every answer and log line. Logs go to stderr; stdout carries only MCP messages.
• EZKIELYNA_MAX_IMAGES caps paid calls per session.

Other providers

The server is built for the Ezkielyna API. EZKIELYNA_BASE_URL can point to another OpenAI-compatible endpoint that serves the same image model names through /v1/images/generations and /v1/images/edits, but per-image prices will not be shown there.

Development

node --test test/

The tests spawn the server over stdio against a local mock gateway: no internet, no cost. Issues and pull requests are welcome.

---

Bahasa Indonesia

Server MCP untuk membuat dan mengedit gambar dari file di komputer lewat asisten AI (Claude Desktop, Claude Code, Cursor, OpenCode). Contoh: "Tempelkan logo ~/Toko/logo.png ke foto ~/Toko/kaos.jpg di dada kiri, simpan sebagai kaos-logo."

• Cukup Node.js 18+. Tersedia di npm: aplikasi menjalankannya dengan npx -y ezkielyna-image-mcp, tanpa clone. Bisa juga dari source (git clone, lihat di atas).
• Butuh API key grup Media (atau Auto) dari Ezkielyna API. Beli saldo mulai $0.5 lewat QRIS/DANA/GoPay: ezkielyna.store/beli-api-key.
• Konfigurasi untuk tiap aplikasi sama seperti contoh di atas; cukup ganti sk-....
• Panduan lengkap berbahasa Indonesia: ezkielyna.store/panduan/mcp-gambar.

License

MIT. Ezkielyna is an independent API gateway and is not affiliated with the model vendors or the MCP client makers.