DataContract MCP Server

An MCP server for working with Data Contracts - loading, querying, and analyzing contract definitions.

Installation

Prerequisites

• Python 3.13 or higher
• uv (recommended)

Using uv (recommended)

uv pip install -e .

Using pip

pip install -e .

Running the Server

python -m datacontract_mcp.server

Adding to Claude Desktop (for local development)

Add this to your Claude Desktop configuration:

{
  "mcpServers": {
    "datacontract": {
      "command": "uv",
      "args": ["run", "--directory", "<abs_repo_path>", "python", "-c", "from datacontract_mcp import main; main()"],
      "env": {
        "DATACONTRACTS_SOURCE": "<abs_repo_path>/datacontracts"
      }
    }
  }
}

Available Tools

Get Data Contract Schema

Retrieves the official Data Contract JSON schema:

datacontracts_get_schema

List Data Contracts

Lists all available Data Contracts in the datacontracts directory:

datacontracts_list_datacontracts

Get Data Contract

Retrieves the content of a specific Data Contract:

datacontracts_get_datacontract(filename="orders.datacontract.yaml")

Query Data Contract

Executes a read-only SQL query against data defined in a Data Contract:

datacontracts_query_datacontract(
    filename="orders.datacontract.yaml",
    query="SELECT * FROM \"orders\" LIMIT 10"
)

Available Resources

The server provides access to these resources:

• datacontract-ref://schema - The official Data Contract JSON schema
• datacontract-ref://example - A concrete example Data Contract from the retail domain

Included Data Contracts

The repository includes example data contracts for:

• orders.datacontract.yaml - Sample order data with CSV source
• video_history.datacontract.yaml - Sample video watching history with CSV source

Development

Environment Variables

• DATACONTRACTS_SOURCE - Directory containing Data Contract files (required)

Testing Locally

DATACONTRACTS_SOURCE=<abs_repo_path>/datacontracts mcp dev -m datacontract_mcp.server

License

MIT