Dorico MCP Server

🎵 Control Dorico via Claude Desktop or ChatGPT - A gift for composition majors

This MCP (Model Context Protocol) server enables natural language control of Steinberg Dorico music notation software through Claude Desktop or ChatGPT Desktop.

Features

🎹 Core Score Tools

• Create new scores with custom instruments

• Save and export scores (PDF, MusicXML)

• Navigate through the score

🎼 Note Input

• Add notes with pitch, duration, and articulation

• Create chords

• Add rests and ties

📝 Notation

• Set key signatures (all major and minor keys)

• Set time signatures

• Add dynamics (pp, p, mp, mf, f, ff, etc.)

• Add tempo markings

• Add slurs and articulations

🎵 Harmony Tools (화성학)

• Analyze chord quality and Roman numerals

• Suggest next chords based on context

• Generate chord progressions

• Check voice leading rules (parallel 5ths/8ves)

🎻 Orchestration Tools (오케스트레이션)

• Complete instrument database with ranges

• Check playability (instrument range validation)

• Transposition for transposing instruments

• Get detailed instrument information

🔍 Query Tools

• Get flows, layouts, and selection properties

• Access engraving, layout, and notation options

📋 Guided Workflows (Prompts)

• Harmonize a melody

• Orchestrate piano scores

• Species counterpoint exercises

Installation

Prerequisites

• Python 3.11+

• Steinberg Dorico (with Remote Control enabled)

• Claude Desktop or ChatGPT Desktop

Install from PyPI (Recommended)

pip install dorico-mcp-server

Install from Source (Development)

git clone https://github.com/happycastle114/dorico-mcp-server.git
cd dorico-mcp-server
pip install -e ".[dev]"

Enable Dorico Remote Control

1. Open Dorico

2. Go to Preferences → General

3. Enable Allow remote control

4. Note the port number (usually 4560)

Configuration

Claude Desktop

Add to your Claude Desktop configuration:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json Windows: %APPDATA%\Claude\claude_desktop_config.json Linux: ~/.config/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "dorico": {
      "command": "python",
      "args": ["-m", "dorico_mcp.server", "--stdio"]
    }
  }
}

ChatGPT Desktop

Add to your ChatGPT Desktop configuration:

macOS: ~/Library/Application Support/com.openai.chat/mcp.json Windows: %LOCALAPPDATA%\com.openai.chat\mcp.json Linux: ~/.config/com.openai.chat/mcp.json

{
  "mcpServers": {
    "dorico": {
      "command": "python",
      "args": ["-m", "dorico_mcp.server", "--stdio"]
    }
  }
}

Note: Restart the app after updating the configuration.

Usage Examples

Creating a Score

"Create a new score for string quartet in G major, 3/4 time, tempo 120"

Adding Notes

"Add a C major chord (C4, E4, G4) as quarter notes"
"Add a melody: C4, D4, E4, F4, G4 as eighth notes"

Harmony Analysis

"What chord would work well after I-IV-V?"
"Generate a 8-bar chord progression in A minor with an authentic cadence"

Orchestration Help

"Is F2 playable on the violin?"
"What's the comfortable range for French horn?"
"What transposition does the Bb clarinet use?"

Counterpoint

"Check if this counterpoint follows first species rules: 
  Cantus: C4, D4, E4, F4, E4, D4, C4
  Counterpoint: G4, A4, B4, C5, B4, A4, G4"
"Generate a counterpoint melody above this cantus firmus: D4, E4, F4, G4, F4, E4, D4"

Voice Leading Validation

"Check this four-part harmony for parallel fifths:
  Soprano: C5, D5, E5
  Alto: E4, F4, G4
  Tenor: G3, A3, B3
  Bass: C3, D3, E3"

Available Tools (51)

Development

Run tests

pytest

Type checking

mypy src/dorico_mcp

Linting

ruff check src/dorico_mcp

Architecture

dorico-mcp-server/
├── src/dorico_mcp/
│   ├── __init__.py         # Package exports
│   ├── server.py           # FastMCP server with tools/resources/prompts
│   ├── client.py           # Dorico WebSocket client
│   ├── commands.py         # Dorico command builders (pure functions)
│   ├── models.py           # Pydantic models
│   └── tools/
│       ├── __init__.py     # Harmony analysis
│       └── instruments.py  # Instrument database
├── tests/                  # Pytest tests
├── docs/                   # Documentation
└── examples/               # Usage examples

Dorico Remote Control API

This server communicates with Dorico via its WebSocket-based Remote Control API:

• Protocol: WebSocket (JSON messages)

• Port: Dynamic (usually 4560-4565)

• Authentication: Session token (stored locally after first approval)

Key Limitations

• Read access is limited: Can only read currently selected items

• Write-focused: Best for sending commands/input

• Full score reading: Requires MusicXML export

Contributing

Contributions are welcome! Please:

1. Fork the repository

2. Create a feature branch

3. Add tests for new functionality

4. Submit a pull request

License

MIT License

Acknowledgments

• AbletonMCP - Inspiration for MCP structure

• MuseScore-MCP - Notation MCP reference

• Dorico.Net - Dorico API documentation

• music21 - Music theory analysis

Made with ❤️ for composition majors everywhere