Skip to content
DataMaker Docs

MCP server

DataMaker ships a Model Context Protocol server that exposes the platform’s full surface as tools. Configure it in your AI coding agent — Claude, Cursor, Copilot, or any other MCP-aware client — and the agent can generate, fetch, push, mask, and run scenarios without leaving the editor.

What you get

  • 25+ tools covering templates, connections, generation, scenarios, and SAP OData.
  • Real auth — the MCP server runs against your real DataMaker workspace using your real API key. No sandboxes.
  • Streaming — long-running tools (like execute_scenario) stream their output back to the agent so it can react incrementally.
  • Workspace-aware — the server knows your project ID, so list_templates returns your templates, not a public catalogue.

Install

The MCP server runs as a stdio process. Install it once with pipx:

Terminal window
pipx install datamaker-mcp

Or npx (Node version):

Terminal window
npx -y datamaker-mcp --version

Configure your agent

Each agent has its own config file. Three of the common ones below.

Claude Desktop / Claude Code

Add to your config (~/Library/Application Support/Claude/claude_desktop_config.json on macOS, or in Claude Code via claude mcp add):

{
  "mcpServers": {
    "datamaker": {
      "command": "datamaker-mcp",
      "env": {
        "DM_API_KEY":     "${DM_API_KEY}",
        "DM_PROJECT_ID":  "proj_xxx"
      }
    }
  }
}

Cursor

~/.cursor/mcp.json:

{
  "mcpServers": {
    "datamaker": {
      "command": "datamaker-mcp",
      "env": {
        "DM_API_KEY":     "your-key",
        "DM_PROJECT_ID":  "proj_xxx"
      }
    }
  }
}

GitHub Copilot

In VS Code with the MCP preview enabled:

settings.json
{
  "github.copilot.mcp.servers": {
    "datamaker": {
      "command": "datamaker-mcp",
      "env": { "DM_API_KEY": "your-key" }
    }
  }
}

For other clients, see Clients setup.

What it looks like

Once configured, the agent has access to tools like list_templates, generate, fetch_endpoint_fields, execute_scenario, etc. You ask in plain English; the agent picks the tools.

A real example, abridged:

you:    Pull 25 German Business Partners from the SAP sandbox into a
        regression set called reg_bp_de_2026q2


agent:  I'll do that. Let me check what entity sets are available...
        [calls fetch_endpoint_fields on conn_s4_sandbox]
        Found 47 entity sets. Building a $filter for Country='DE'...
        [calls fetch_records on A_BusinessPartner with $filter]
        318 matches. Sampling 25...
        [calls save_set with name='reg_bp_de_2026q2']
        ✓ 25 records saved.

Pages

Source

datamaker-mcp is open source: github.com/automators-com/datamaker-mcp. File issues and PRs there.