Rockfish MCP Server

Rockfish MCP Server

A Model Context Protocol (MCP) server that provides access to the Rockfish API, enabling AI assistants to interact with Rockfish's machine learning platform.

Features

This MCP server provides tools for the following Rockfish resources:

• Databases: Create, list, update, and delete databases
• Worker Sets: Manage worker sets for distributed processing
• Workflows: Create and manage ML workflows
• Models: Upload, list, and manage ML models
• Projects: Organize and manage projects
• Datasets: Create and manage datasets

Installation

1. Clone the repository:

2. Install dependencies:

3. Set up environment variables:

Configuration

Create a .env file with your Rockfish API credentials:

Usage

Run the MCP server:

Or use the console script:

Claude Desktop Setup

To use this MCP server with Claude Desktop:

1. Complete the installation steps above (clone, install dependencies, set up .env file)
2. Find your Claude Desktop configuration directory:
   • macOS: ~/Library/Application Support/Claude/
   • Windows: %APPDATA%\Claude\
   • Linux: ~/.config/Claude/
3. Create or edit the claude_desktop_config.json file in that directory:

4. Update the paths in the configuration:
   • Replace /path/to/your/project/.venv/bin/python with the actual path to your Python executable
   • Replace your_api_key_here with your actual Rockfish API key
   • Adjust ROCKFISH_BASE_URL if you're using a different endpoint
5. Get the correct Python path by running this command in your project directory:

6. Example configuration (replace with your actual paths and API key):

7. Restart Claude Desktop after making these changes
8. Test the connection by asking Claude to list your Rockfish databases or projects

MCP Inspector Setup

The MCP Inspector is a debugging tool that helps you test your MCP server before connecting it to Claude Desktop.

Installation

Usage

1. Start the MCP Inspector:

2. Or create a test script for easier repeated testing:

Make it executable and run:

3. The Inspector will open in your browser and show:
   • Available tools (should show all 22 Rockfish tools)
   • Tool schemas and descriptions
   • Interactive tool testing interface
4. Test your tools by:
   • Selecting a tool from the list (e.g., list_databases)
   • Filling in required parameters
   • Clicking "Call Tool" to test the API call
   • Viewing the response

Useful Tools to Test First

• list_databases - Simple GET request with no parameters
• list_projects - Another simple list operation
• get_database - Test with a database ID from the list
• create_database - Test creating a new resource

Troubleshooting

• MCP server not appearing: Check that the Python path is correct and the virtual environment is activated
• Authentication errors: Verify your ROCKFISH_API_KEY is correct
• Connection issues: Confirm your ROCKFISH_BASE_URL is accessible
• Path issues on Windows: Use forward slashes or escaped backslashes in JSON paths

Available Tools

Database Tools

• list_databases: List all databases
• create_database: Create a new database
• get_database: Get a specific database by ID
• update_database: Update a database
• delete_database: Delete a database

Worker Set Tools

• list_worker_sets: List all worker sets
• create_worker_set: Create a new worker set
• get_worker_set: Get a specific worker set by ID
• delete_worker_set: Delete a worker set

Workflow Tools

• list_workflows: List all workflows
• create_workflow: Create and run a new workflow
• get_workflow: Get a specific workflow by ID
• update_workflow: Update a workflow

Model Tools

• list_models: List all models
• upload_model: Upload a new model
• get_model: Get a specific model by ID
• delete_model: Delete a model

Project Tools

• list_projects: List all projects
• create_project: Create a new project
• get_project: Get a specific project by ID
• update_project: Update a project

Dataset Tools

• list_datasets: List all datasets
• create_dataset: Create a new dataset
• get_dataset: Get a specific dataset by ID
• update_dataset: Update a dataset
• delete_dataset: Delete a dataset

Development

To contribute to this project:

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests if applicable
5. Submit a pull request

License

MIT License