Riffmcp is a dedicated MCP server tool developed by Whitneyland, designed to integrate music composition, playback, and rendering capabilities with Large Language Models. Built primarily with Swift, this application runs natively on macOS, distributed conveniently via a .dmg file. It enables AI clients to interact with musical functions, transforming natural language requests into musical outputs and information. Riffmcp expands the utility of LLMs, allowing them to engage directly with creative audio tasks on the mcp platform.

What does Riffmcp connect to?

Riffmcp connects directly to Large Language Models, functioning as an MCP server specializing in music-related tasks. It supports the MCP stdio protocol, which is efficiently routed through its built-in HTTP server. This design allows compatibility with various AI clients, including Claude Desktop and Gemini CLI, alongside any other LLM that supports the MCP server protocol. Riffmcp acts as a bridge, enabling LLMs to execute musical commands and retrieve information within its local execution environment.

Core features and tools

Riffmcp offers capabilities inferred from natural language usage examples rather than formally named tools. These include playing various melodies, such as "Play a happy melody" or "Play a 12 bar jazz progression with walking bass." It also supports rendering sheet music, implied by commands like "Show me the sheet music for that." Additionally, users can query available instruments, as suggested by "What instruments are available?" These functions allow LLMs to interact creatively with music.

Who is it for?

Riffmcp is intended for users of Large Language Models, specifically those utilizing clients like Claude Desktop and Gemini CLI, who seek to incorporate music-related functionalities into their AI interactions. It serves individuals looking to expand their LLM's capabilities into areas of music composition, playback, and visual representation. Skip this if your LLM ecosystem does not support MCP servers or if music integration is not a required capability.

How to get started

No credentials are required for Riffmcp itself. The stdio protocol for communication is routed through a built-in HTTP server.

Use cases

• Enabling an LLM to compose and play music sequences.
• Allowing an LLM to request and view rendered sheet music for a musical piece.
• Providing the LLM with information about available instruments.

Riffmcp provides a valuable bridge between Large Language Models and musical functionality on macOS. Its native application design and direct integration with MCP-compatible clients make it an accessible tool for adding music composition and playback to AI workflows. This free utility expands creative possibilities for LLM users.