Skip to content
DataMaker Docs

Clients setup

The DataMaker MCP server is a stdio process — you don’t run a separate web service. Each agent has its own config syntax for registering MCP servers. The instructions below are the working set as of mid-2026.

Prerequisites

  • A DataMaker API key with at least member role.

  • The datamaker-mcp binary on your PATH. Install with:

    Terminal window
    pipx install datamaker-mcp

    (or npx -y datamaker-mcp for a per-call install).

  • Optional: a DM_PROJECT_ID to scope the agent to one project. Without it, every tool call needs an explicit project_id argument.

Claude Desktop

~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%/Claude/claude_desktop_config.json (Windows):

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

Restart Claude. The toolbox icon at the bottom of any chat will show the DataMaker tools.

Claude Code

Terminal window
claude mcp add datamaker datamaker-mcp \
  --env DM_API_KEY=sk-... \
  --env DM_PROJECT_ID=proj_xxx

Then claude to start a session — /mcp shows the registered servers.

Cursor

~/.cursor/mcp.json:

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

Restart Cursor. Settings → MCP lists the server and lets you toggle it.

GitHub Copilot (VS Code)

In VS Code with the MCP preview enabled:

settings.json
{
  "github.copilot.mcp.servers": {
    "datamaker": {
      "command": "datamaker-mcp",
      "env": { "DM_API_KEY": "sk-...", "DM_PROJECT_ID": "proj_xxx" }
    }
  }
}

Continue.dev

~/.continue/config.json:

{
  "experimental": {
    "modelContextProtocolServers": [
      {
        "name": "datamaker",
        "command": "datamaker-mcp",
        "env": { "DM_API_KEY": "sk-...", "DM_PROJECT_ID": "proj_xxx" }
      }
    ]
  }
}

Cline / Roo Code

Both use a similar JSON config under their MCP settings. Point at datamaker-mcp, provide the env vars, restart.

Generic stdio client

If your agent supports the standard MCP stdio transport, this is the launch command:

Terminal window
DM_API_KEY=sk-... DM_PROJECT_ID=proj_xxx datamaker-mcp

The server speaks JSON-RPC 2.0 over stdin/stdout per the MCP spec.

Verifying the connection

After configuring, ask the agent: List my DataMaker templates. It should call list_templates and return a non-empty list. If you get an “I don’t have access” response, check:

  • The agent has reloaded the MCP server config.
  • DM_API_KEY is set and not expired.
  • The API key’s project membership includes DM_PROJECT_ID.

Troubleshooting

  • Tool not found → restart the agent. MCP servers are registered at startup; new tools after a server upgrade need a fresh connection.
  • Auth error on every tool → the API key is wrong or revoked. Generate a new one.
  • spawn datamaker-mcp ENOENT → the binary isn’t on the agent’s PATH. Provide an absolute path in command:, e.g. /Users/you/.local/bin/datamaker-mcp.

For more, see Troubleshooting → Common errors.