pixelmator-pro-mcp

Production-ready Model Context Protocol (MCP) server for Pixelmator Pro on macOS — automated graphic design, vector typography, CoreML neural image enhancement, layer composition, and multi-format export.

✨ Features & Enhancements (over community version)

This implementation is thoroughly rebuilt against Pixelmator Pro's official AppleScript scripting definitions (.sdef):

• 🎨 16-bit Precision Color Engine: Seamlessly converts standard Hex codes (#1e1e2e, #00F0FF) and CSS named colors to AppleScript's native 16-bit {r, g, b} values (0–65535), completely eliminating AppleScript -1700 type mismatch crashes.
• ✍️ Native Vector Text Layers (pixelmator_add_text): Adds real, fully editable vector text layers with exact font selection, point size, 16-bit color, alignments, and canvas coordinates.
• 🛡️ Robust Layer Reference System (pixelmator_layers): Fixed the fatal -1728 object not found bug by using safe layer specifiers (first layer whose name is "..."). Full support for listing, selecting, toggling visibility, opacity adjustment, blend modes, and merging.
• 🧠 Apple CoreML Powerhouse:
  • pixelmator_auto_enhance: ML Enhance, Auto White Balance, Auto Light, Auto Color Balance, ML Denoise, and ML Deband.
  • pixelmator_super_resolution: 300% (3x) neural upscaling.
  • pixelmator_remove_background: Machine-learning foreground segmentation and background extraction.
  • pixelmator_select_subject: Smart subject detection with ML edge refinement.
• 📐 Clean Canvas Creation (pixelmator_create): Supports transparent canvases (automatically purges the default white solid layer) or custom background fills.
• 🚀 Universal App Identifier: Uses application id "com.apple.pixelmator", seamlessly working across both "Pixelmator Pro" and "Pixelmator Pro Creator Studio" editions.

📦 Installation & Setup

Requirements

• macOS with Pixelmator Pro installed
• Bun (or Node.js >= 18)

Build

cd ~/github/pixelmator-pro-mcp
bun install
bun run build

Run Tests

bun test

🛠️ MCP Configuration

Add this server to your MCP client configuration (e.g. claude_desktop_config.json or Pi settings):

Option A: via npx (Recommended for Claude Desktop & Node.js)

{
  "mcpServers": {
    "pixelmator-pro": {
      "command": "npx",
      "args": ["-y", "pixelmator-pro-mcp"]
    }
  }
}

Option B: via bunx (Blazing fast for Bun users)

{
  "mcpServers": {
    "pixelmator-pro": {
      "command": "bunx",
      "args": ["pixelmator-pro-mcp"]
    }
  }
}

Option C: Local Development

{
  "mcpServers": {
    "pixelmator-pro": {
      "command": "bun",
      "args": ["run", "/path/to/pixelmator-pro-mcp/dist/index.js"]
    }
  }
}

🛠️ Available Tools (22 Tools)

| Tool | Description | | :--- | :--- | | pixelmator_create | Create a new document with custom dimensions, transparent background, or fill color | | pixelmator_open | Open an existing image file (PNG, JPEG, WebP, TIFF, PSD, PXD) | | pixelmator_close | Cleanly close a document (or front document) with optional unsaved changes discard | | pixelmator_list_documents | List all open documents with dimensions, modified status, and active window state | | pixelmator_export | Export active document to PNG, JPEG, WebP, TIFF, PSD, or native PXD | | pixelmator_adjust_color | Comprehensive color grading engine: CoreML auto modes, precise sliders (temperature/tint, exposure, contrast, highlights, shadows, clarity, vibrance, saturation, sharpen, grain, vignette), presets, reset, and returns introspective snapshot | | pixelmator_selection | Comprehensive selection & masking engine: AI subject selection, rectangular/elliptical bounds, invert, extract selection to isolated layer, layer masks | | pixelmator_add_text | Add editable vector text layer with multiline auto-wrap (width), alignments (left, center, right, justify), vertical_alignment, rotation, opacity, font, size, and positioning | | pixelmator_add_shape | Add vector shape layers (rectangle, rounded rect, ellipse, star, line) with fills, strokes, shadows, rotation, and layer_opacity | | pixelmator_undo | Undo the most recent operation(s) with multi-step rollback (steps: N) in active or specified document | | pixelmator_redo | Redo the most recently undone operation(s) with multi-step replay (steps: N) | | pixelmator_overlay | Overlay an external image with blend mode (Screen, Multiply, Overlay, etc.) and opacity | | pixelmator_layers | Layer management: list, select, delete, rename, reorder (bring_to_front, send_to_back, move_before, move_after), toggle visibility, opacity, blend modes, merge | | pixelmator_fill | Fill current layer or selection with 16-bit precision color | | pixelmator_auto_enhance | CoreML enhancements (ML Enhance, Auto Light, White Balance, Denoise, Deband) | | pixelmator_super_resolution | 300% (3x) CoreML neural super resolution upscaling | | pixelmator_remove_background | CoreML neural background removal | | pixelmator_select_subject | CoreML subject detection with smart edge refinement | | pixelmator_doc_info | Inspect canvas dimensions, layer count, file path, and modified status | | pixelmator_resize | Resize image content or canvas dimensions | | pixelmator_transform | Rotate (90° CW/CCW, 180°) and flip (horizontally/vertically) | | pixelmator_crop | Crop to specific coordinate bounds or auto-trim border pixels |

📄 License

MIT