Daisys

Daisys MCP server provides an integration point to connect Daisys audio/audio-management capabilities with your MCP clients. It lets you run the Daisys MCP server locally or against a remote host, so you can manage audio storage and processing from your MCP-enabled tools while keeping credentials securely configured in your client.

How to use

To use the Daisys MCP server, you connect it through an MCP client (for example Claude Desktop, Cursor, mcp-cli, or mcp-vscode). You must have a Daisys account and set up your username and password to enable authentication. In your MCP client, register a server configuration named something like "daisys_mcp" and provide your credentials and the base storage path for audio files. You can run the server either locally via uvx or via uv after you install and build the source.

How to install

Prerequisites you need before installing:
- A Daisys account with an email and password
- A supported MCP client installed on your machine
- A working development environment with Python tooling managed by uv (uvx) if you build from source
- PortAudio dependencies for your platform (required for audio handling)
  - macOS: install PortAudio with brew
  - Linux: install PortAudio development headers and libjack development headers

# macOS
brew install portaudio

# Linux (Debian/Ubuntu)
sudo apt install portaudio19-dev libjack-dev

From prebuilt configuration (client integration)

Add a server configuration to your MCP client that points to the Daisys MCP server using the uvx runtime. The following snippet shows the required configuration structure and credentials placeholders.

{
  "mcpServers": {
    "daisys-mcp": {
      "command": "uvx",
      "args": ["daisys-mcp"],
      "env": {
        "DAISYS_EMAIL": "{Your Daisys Email}",
        "DAISYS_PASSWORD": "{Your Daisys Password}",
        "DAISYS_BASE_STORAGE_PATH": "{Path where you want to store your audio files}"
      }
    }
  }
}

From source code build and run

If you prefer building from source, follow the steps below to clone the project, set up the Python environment with uv, and run the server locally. This process uses uv to install dependencies and run the MCP server module.

# Clone the repository
git clone https://github.com/daisys-ai/daisys-mcp.git

cd daisys-mcp

# Install uv (Python package manager) and set up a virtual environment
# You can install uv with the provided install script or via its repository
curl -LsSf https://astral.sh/uv/install.sh | sh
uv venv
# Activate the virtual environment
# Windows
# .venv\Scripts\activate
# macOS/Linux
source .venv/bin/activate

uv pip install -e .

# Add your credentials to the MCP config as shown in the integration example above

Run the server from the built source

After building, use the final runtime command to start the MCP server with the required module. This ensures the server runs the Daisys MCP entry point.

{
  "mcpServers": {
    "daisys-mcp": {
      "command": "uv",
      "args": [
        "--directory",
        "{installation_path}/daisys-mcp",
        "run",
        "-m",
        "daisys_mcp.server"
      ],
      "env": {
        "DAISYS_EMAIL": "{Your Daisys Email}",
        "DAISYS_PASSWORD": "{Your Daisys Password}",
        "DAISYS_BASE_STORAGE_PATH": "{Path where you want to store your audio files}"
      }
    }
  }
}

Available tools

MCP Inspector

Debug and test locally with MCP Inspector to inspect server interactions and validate configuration.