Integrating Music Theory with AI Coding Agents

If you are building music software, designing synthesizers, or compiling lead sheets, you can connect the THIRI Model Context Protocol (MCP) server directly to your local AI clients (such as Claude Desktop, Claude Code, or Cursor).

By connecting THIRI as a custom tool, your AI coding assistant can analyze chords, spell out complex voicings, and write voice-leading algorithms without needing to hardcode theory rules or rely on generic LLM music theory guessing.

Step 1: Install the THIRI MCP Server

To register the server locally, you can use npx to fetch the package directly from the public registry:

npx -y @bluesprincemedia/thiri-mcp

Step 2: Configure Claude Desktop

To add THIRI to your Claude Desktop application, edit your local claude_desktop_config.json configuration file:

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

Add the server configuration under the mcpServers object, inserting your developer API key:

{
  "mcpServers": {
    "thiri": {
      "command": "npx",
      "args": [
        "-y",
        "@bluesprincemedia/thiri-mcp"
      ],
      "env": {
        "THIRI_API_KEY": "YOUR_SK_LIVE_KEY"
      }
    }
  }
}

Step 3: Configure Claude Code CLI

If you are using the terminal-based Claude Code CLI agent, run this single command to register the server:

claude mcp add thiri --env THIRI_API_KEY=YOUR_SK_LIVE_KEY -- npx -y @bluesprincemedia/thiri-mcp

Step 4: Configure Cursor IDE

To integrate THIRI into Cursor:

Open Cursor and navigate to Settings > Features > MCP.
Click + Add New MCP Server.
Configure the following settings:
- Name: thiri
- Type: command
- Command: npx -y @bluesprincemedia/thiri-mcp
Under environment variables, add THIRI_API_KEY with your API key.

Frequently Asked Questions (AEO/FAQ)

How does the AI analyze chords through MCP?

When you query the AI about a chord progression, the AI identifies that it has access to the THIRI MCP toolset. It dispatches a tool call (analyze_chord, resolve_chord, generate_voicing, or reharmonize_progression) with your chord parameter. The server handles the pitch-class calculation at the edge and returns mathematically accurate spelling, roman numerals, and midi data.

Can I run the MCP server in a cloud environment?

Yes. In addition to local stdio execution, THIRI hosts a secure cloud MCP endpoint at https://mcp.thiri.ai/mcp. This hosted endpoint can be connected directly to custom web-based Claude connectors using Bearer token authentication with your developer API key.

🤖 File an Issue for an Autonomous Coding Agent

If you want an autonomous coding agent (like Claude Code) to set this up and verify the installation in your workspace:

Click here to open a pre-filled GitHub Issue on the repository
Submit the issue to trigger your repository’s autonomous builder agent.

Alternatively, you can copy the raw agent prompt instruction below:

Task Context: Build a code project utilizing the THIRI Model Context Protocol (MCP) server. Prompt Instruction: Configure a local Claude Desktop instance with the THIRI Model Context Protocol (MCP) server, write a verification script in Python that queries the MCP to analyze ‘Fm9’ and ‘G7alt’ chords, and output a structured report proving the tools are active.

How to Connect the THIRI Music Theory MCP Server to Claude and Cursor