DataProduct MCP Server

A server that helps you work with Data Products - making it easy to load, validate, query, and analyze your data assets, including their linked data contracts.

What It Does

This MCP server provides a simple interface for working with Data Products:

• Loads and validates data products
• Analyzes linked data contracts within data products
• Provides tools to query and analyze your data
• Recognizes files automatically by their extensions (.dataproduct.yaml and .datacontract.yaml)
• Handles data assets in a consistent way

Installation

Prerequisites

• Python 3.10 or higher
• uv (recommended) or pip

Quick Install

With uv (recommended):

uv pip install -e .

With pip:

pip install -e .

Running the Server

Set up where your data files are stored:

# Point to your data files
export DATAASSET_SOURCE=/path/to/assets/directory

# Run the server
python -m datacontract_mcp.server

# For development with MCP CLI
mcp dev -m datacontract_mcp.server

Using with Claude Desktop

Add this configuration for local development:

{
  "mcpServers": {
    "datacontract": {
      "command": "uv",
      "args": ["run", "--directory", "<abs_repo_path>", "python", "-m", "datacontract_mcp.server"],
      "env": {
        "DATAASSET_SOURCE": "<abs_repo_path>/dataassets"
      }
    }
  }
}

Available Tools

Data Product Tools

List Data Products

See all available Data Products:

dataproducts_list

Get Data Product

Get a specific Data Product:

dataproducts_get(filename="shelf_warmers.dataproduct.yaml")

Validate Data Product

Check if a Data Product is valid:

dataproducts_validate(filename="shelf_warmers.dataproduct.yaml")

Get Data Product Outputs

Get all output ports from a Data Product, along with their linked contracts:

dataproducts_get_outputs(filename="shelf_warmers.dataproduct.yaml")

Get Output Schema

Get the schema for a specific output port using its linked data contract:

dataproducts_get_output_schema(
    filename="shelf_warmers.dataproduct.yaml",
    port_id="shelf_warmers"
)

Query Data Product

Query data from a Data Product's output port:

dataproducts_query(
    filename="shelf_warmers.dataproduct.yaml",
    query="SELECT * FROM \"shelf_warmers\" LIMIT 10"
)

With additional options:

dataproducts_query(
    filename="shelf_warmers.dataproduct.yaml", 
    query="SELECT * FROM \"shelf_warmers\" LIMIT 10",
    port_id="shelf_warmers",   # Optional output port ID
    server="local",            # Optional server key
    model="shelf_warmers",     # Optional model key
    include_metadata=True
)

Included Examples

The examples directory contains sample files for both data products and their supporting data contracts:

Data Products

• examples/shelf_warmers.dataproduct.yaml - Shows which products haven't sold in 3 months
• examples/orders.dataproduct.yaml - Order data with order details
• examples/video_history.dataproduct.yaml - Video watching history

Supporting Data Contracts

• examples/shelf_warmers.datacontract.yaml - Contract for shelf warmers data
• examples/orders.datacontract.yaml - Sample order data contract (CSV format)
• examples/video_history.datacontract.yaml - Video history data contract (CSV format)

Available Resources

Access these built-in resources:

• dataproduct-ref://schema - The official Data Product JSON schema
• dataproduct-ref://example - A sample Data Product

Configuration

Environment Variables

• DATAASSET_SOURCE - Where to find your data assets (both products and contracts)

AWS S3 Configuration

To access data from S3:

• AWS_REGION or AWS_DEFAULT_REGION - AWS region (default: us-east-1)
• S3_BUCKETS - List of allowed S3 buckets (comma-separated)
• S3_MAX_BUCKETS - Maximum number of buckets (default: 10)

Authentication options:

1. AWS Profile (for development): AWS_PROFILE or S3_PROFILE
2. Direct credentials: AWS_ACCESS_KEY_ID/AWS_SECRET_ACCESS_KEY
3. Default chain (environment variables, ~/.aws/credentials, IAM roles)

Testing Locally

DATAASSET_SOURCE=<abs_repo_path>/dataassets mcp dev -m datacontract_mcp.server

License

MIT