Mage MCP Server

🚀 Mage.ai Data Pipeline and Model Context Protocol (MCP) Integration

A comprehensive Model Context Protocol (MCP) integration solution designed for Mage.ai, enabling AI assistants like Anthropic's Claude to efficiently interact with Mage data pipelines.

🚀 Quick Start

This is a comprehensive Model Context Protocol (MCP) integration solution tailored for Mage.ai. It empowers AI assistants, such as Anthropic's Claude, to interact efficiently with Mage data pipelines.

✨ Features

• Browse Pipelines: Explore all Mage.ai pipelines and their components.
• View and Edit Code: Read and modify block code.
• Execute Pipelines: Run entire pipelines or individual blocks.
• Search Functionality: Find blocks by name, type, or content.
• Pipeline Creation: Create new pipelines and blocks from scratch.

📦 Installation

# Clone the repository
git clone https://github.com/yourusername/mage-mcp.git
cd mage-mcp

# Install dependencies
pip install -r requirements.txt

📚 Documentation

Configuration

The server can be configured via environment variables or command-line parameters:

• MAGE_API_URL: The URL of the Mage.ai API (default: http://localhost:6789/api/).
• MAGE_API_KEY: The API key for authentication (if required).

Create a .env file for persistent configuration:

MAGE_API_URL=http://localhost:6789/api/
MAGE_API_KEY=your_api_key_if_needed

Usage

Start the Server

Use the provided script to start the MCP server:

# Use standard input-output transport (default)
python run_server.py

# Use SSE transport
python run_server.py --transport sse --host localhost --port 3000

# Other options
python run_server.py --api-url http://your-mage-instance:6789/api/ --api-key your_key --debug

Integration Installation with Claude Desktop

For the best experience, install this server directly through Claude Desktop:

# Use the mcp CLI
mcp install run_server.py --name "Mage.ai Manager"

# With environment variable configuration
mcp install run_server.py -v MAGE_API_URL=http://localhost:6789/api/ -v MAGE_API_KEY=your_key

Development Mode

For testing and development:

mcp dev run_server.py

MCP Resources and Tools

Resource Addresses

• mage://pipelines - List all pipelines.
• mage://pipelines/{pipeline_uuid} - Get pipeline details.
• mage://pipelines/{pipeline_uuid}/blocks - Get blocks in a pipeline.
• mage://pipelines/{pipeline_uuid}/blocks/{block_id} - Get details of a specific block.
• mage://pipelines/{pipeline_uuid}/execute - Execute a pipeline.
• mage://pipelines/{pipeline_uuid}/execute/{execution_id} - Get execution status.

Utility Functions

• list_all_pipelines(): Return a list of all pipelines.
• get_pipeline_details(pipeline_uuid): Get details of a specified pipeline.
• execute_pipeline(pipeline_uuid, parameters): Execute a pipeline with optional parameters.
• search_blocks(query): Search for blocks by query content.

💻 Usage Examples

Basic Usage

# Create a new pipeline
pipeline = create_pipeline("Data Analysis Pipeline", "Data processing and analysis")

# Add a block to the pipeline
add_block_to_pipeline(pipeline.uuid, "Data Cleaning", "Data cleaning steps")

# Execute the pipeline
execution_id = execute_pipeline(pipeline.uuid, {"input_data": "raw_data.csv"})

# Get execution status
status = get_execution_status(pipeline.uuid, execution_id)
print(status)  # Print execution result

📄 License

This project is licensed under the MIT License. See the LICENSE file for details.