Cloudera AI MCP

🚀 Cloudera Machine Learning Model Control Protocol (MCP)

This MCP enables the integration of Python with Cloudera Machine Learning, allowing Claude to interact with CML services programmatically.

🚀 Quick Start

This MCP implementation allows seamless interaction between Python and Cloudera Machine Learning, enabling Claude to communicate with CML services programmatically.

✨ Features

1. Upload Folders: Upload entire folders to your CML project while preserving the directory structure.
2. Create Jobs: Create new CML jobs with custom settings.
3. List Jobs: View all jobs in a project and their current status.
4. Delete Jobs: Delete a single job or all jobs in a project.
5. Get Project ID: Retrieve the project ID by project name.
6. List Project Files: View files and directories in a project.
7. Model Management: Create, list, and manage ML models and deployments.
8. Experiment Tracking: Record and manage ML experiments and runs.
9. Application Management: Create, update, and manage CML applications.

📦 Installation

1. Clone this repository.
2. Install the dependencies:

   pip install -r requirements.txt
   

📚 Documentation

Configuration

MCP requires the following configuration:

1. host: The URL of your CML instance (e.g., "https://ml-xxxx.cloudera.site").
2. api_key: The API key for authentication.
3. project_id: Your CML project ID (optional - can now be obtained by project name).

You can provide this configuration via code when initializing MCP or use environment variables:

export CLOUDERA_ML_HOST="https://ml-xxxx.cloudera.site"
export CLOUDERA_ML_API_KEY="your-api-key"
# Optional: export CLOUDERA_ML_PROJECT_ID="your-project-id"

URL Configuration Notes

• The host URL should not contain a duplicate "https://" prefix.
• Trailing slashes are handled automatically.
• MCP will automatically format the URL correctly.

Running the MCP Server

This MCP can be run as a server, allowing Claude to interact with Cloudera ML.

Set up the Environment

Copy the example .env file and add your credentials:

cp .env.example .env
# Edit the .env file to provide your credentials

Start the Server

./server.py

The server uses stdio transport and can be directly connected to Claude by default.

Using with Claude Desktop

To use this server with the Claude Desktop application, add the following configuration to the "mcpServers" section of claude_desktop_config.json:

{
  // Other configurations...
  "mcpServers": {
    "my-mcp-server": {
      "type": "stdio",
      "command": ["python", "-m", "cloudera.mcp.server"],
      "args": []
    }
  }
}

Requirements

• Python 3.8+
• requests
• pathlib
• python-dotenv
• mcp[cli]

📄 License

MIT