Access

To learn more, please reach out directly to your Wunderkind team.

Prerequisites

Your server endpoint and website ID, provided by your Wunderkind representative during onboarding
An MCP-compatible client (Claude, Claude Code, Cursor, VS Code with GitHub Copilot, etc.)

The server uses MCP's Streamable HTTP transport. In the examples below, replace with the endpoint you were given.

Client setup

Claude.ai and Claude Desktop

Navigate to Settings > Connectors > Add custom connector.
Enter the URL: https:///mcp

Claude Code

Run the following command:

claude mcp add --transport http wunderkind-profiles https\://<server-endpoint>/mcp

If your integration was issued an access token, pass the matching header:

claude mcp add --transport http wunderkind-profiles https\://<server-endpoint>/mcp <br />--header "Authorization: Bearer <token>"

Cursor

Add the following to ~/.cursor/mcp.json:

{
  "mcpServers": {
    "wunderkind-profiles": {
      "url": "https://<server-endpoint>/mcp"
    }
  }
}

Visual Studio Code (GitHub Copilot)

Add the following to .vscode/mcp.json in your project root:

{
  "servers": {
    "wunderkind-profiles": {
      "url": "https://<server-endpoint>/mcp"
    }
  }
}

Other clients

Any MCP-compatible client can connect. In a client configured via JSON:

{
  "mcpServers": {
    "wunderkind-profiles": {
      "type": "http",
      "url": "https://<server-endpoint>/mcp"
    }
  }
}

For JSON-configured clients that support request headers, an access token goes in a headers entry:

{
  "mcpServers": {
    "wunderkind-profiles": {
      "type": "http",
      "url": "https://<server-endpoint>/mcp",
      "headers": { "Authorization": "Bearer <token>" }
    }
  }
}

Tools

`get_recommendations`

Personalized product recommendations for a shopper, ranked most relevant first.

Inputs

Input	Type	Required	Description
device_id	string	Yes	The Wunderkind device ID for the shopper (the bxdid value available to your on-site integration).

Output

{
  "items": [
    { "id": "SKU-12345" },
    { "id": "SKU-67890" }
  ]
}

Field	Type	Description
items	array of objects	Recommended products, ranked most relevant first. Always present; an empty array means no recommendations are available for this shopper.
items[].id	string	The product identifier, matching the product IDs in the catalog you share with Wunderkind. Additional per-item fields may be added in the future — treat items as objects and ignore fields you don't recognize.

Behavior notes

A shopper the system doesn't know yet (for example, a first-time visitor) returns an empty items list — a normal result, not an error. Your assistant should fall back to its default experience.
Responses contain only product IDs — no personal or profile data.Security and privacy

Verify your connection

Once connected, your assistant discovers the available tools automatically. Ask it:

"What products should I suggest to this shopper?"

The assistant should call get_recommendations and use the ranked product IDs in its reply.