MCP Server, connect AI to bank & card data validation

What is MCP and why does it matter?

The Model Context Protocol (MCP) lets AI assistants call external tools in real time. Instead of guessing at bank details from training data, your AI can validate real routing numbers, BINs, IBANs, and SWIFT codes — just by asking a question.

Example: Ask Claude "Is routing number 021000021 valid?" and it will call our validate_routing_number tool automatically.

Available MCP tools

16 tools mapped to BankDataStack REST endpoints, plus self-service registration.

Getting started

Same URL for every client. No OAuth screen to click through — call register for an instant sandbox key, or use a real key from your dashboard.

Connect to Claude

Works the same on claude.ai and in the Claude Desktop app — the connector is saved on your Claude account, not in a local file.

Connect with Claude Code

From any terminal (outside a claude session):

claude mcp add --transport http mcp-bankdatastack https://mcp.bankdatastack.com/mcp

Then open a Claude Code session and ask it to register a sandbox key and validate something — no separate auth step is needed for the sandbox tier. Remove the server later with:

claude mcp remove mcp-bankdatastack

Connect to Cursor

Paste this into ~/.cursor/mcp.json on a Mac, or %USERPROFILE%\.cursor\mcp.json on Windows (or .cursor/mcp.json in a project folder for a project-only setup). Save, quit Cursor fully, and reopen it.

{
  "mcpServers": {
    "bankdatastack": {
      "url": "https://mcp.bankdatastack.com/mcp"
    }
  }
}

No auth block needed to get started — ask Cursor to call register for a sandbox key the first time you use it. For a production key instead, add ?apikey=YOUR_KEY to the url above.

After setup, try: "Look up SWIFT code DEUTDEFF"

Connect to ChatGPT

In ChatGPT, open Settings → Connectors and add a custom MCP connector.

Python: validate a routing number

A complete example that registers a sandbox key, then validates a routing number over MCP.

import httpx
import json
import asyncio

MCP_SERVER_URL = "https://mcp.bankdatastack.com/mcp"

async def call_mcp_tool(tool_name: str, arguments: dict, api_key: str | None = None):
    headers = {"Authorization": f"Bearer {api_key}"} if api_key else {}
    async with httpx.AsyncClient(timeout=30) as client:
        response = await client.post(
            MCP_SERVER_URL,
            headers=headers,
            json={
                "jsonrpc": "2.0",
                "id": "1",
                "method": "tools/call",
                "params": {"name": tool_name, "arguments": arguments}
            }
        )
        return response.json()

async def main():
    # 1. No key yet? Register a free sandbox credential — no signup, no card.
    reg = await call_mcp_tool("register", {"client_name": "python-example"})
    token = json.loads(reg["result"]["content"][0]["text"])["access_token"]

    # 2. Use it to validate a routing number.
    result = await call_mcp_tool(
        "validate_routing_number",
        {"routing_number": "021000021"},
        api_key=token,
    )
    print(json.dumps(result, indent=2))

asyncio.run(main())

Frequently asked questions