Skip to main content

MCP Server for IterableData

Run IterableData operations as Model Context Protocol tools in Cursor, Claude Desktop, and other MCP clients.

Install​

pip install iterabledata[mcp]

Run (stdio)​

iterable-mcp

The server registers read-first tools: detect_format, describe_capabilities, read_sample, infer_schema, analyze_dataset, generate_documentation, and gated convert_file (writes require confirm=True).

Registry manifest: server.json at the repository root (io.github.datenoio/iterabledata). See Agent discovery to submit it to the MCP Registry.

Cursor configuration​

Add to MCP settings (example):

{
  "mcpServers": {
    "iterabledata": {
      "command": "iterable-mcp",
      "args": []
    }
  }
}

Use the full path to iterable-mcp if it is not on PATH.

Claude Desktop​

Add a server entry in Claude Desktop MCP config pointing to iterable-mcp with stdio transport (same as above).

Programmatic use​

from iterable.mcp import create_mcp_server

mcp = create_mcp_server()
mcp.run()

Security​

  • Prefer read_sample with redact=True for sensitive data
  • convert_file will not write unless confirm=True
  • generate_documentation may call external LLM APIs when used

See also​