Connect the Conversations server

Connecting the Conversations server to your AI assistant is a single step: add the shared prototype server URL to your AI platform's MCP configuration. There's no API key to request and no per-customer setup — all prototype users connect to the same endpoint.

Prerequisites

An AI platform that supports MCP, such as:

Server URL

Use the following URL in your MCP configuration:

https://uqa.api.exponea.dev/cocoaas/public/api/clarity-search/v1/mcp/019d4917-3c76-7479-9f00-06c620b231bb

This is a shared prototype endpoint connected to the Pacific Apparel sample catalog.

Add the server to your AI assistant

How you add the connection depends on which AI platform you use.

Run the following command to add the server:

claude mcp add conversations --transport http https://uqa.api.exponea.dev/cocoaas/public/api/clarity-search/v1/mcp/019d4917-3c76-7479-9f00-06c620b231bb

Or add the configuration manually to ~/.claude/mcp.json:

{
  "mcpServers": {
    "conversations": {
      "type": "http",
      "url": "https://uqa.api.exponea.dev/cocoaas/public/api/clarity-search/v1/mcp/019d4917-3c76-7479-9f00-06c620b231bb"
    }
  }
}

To verify the server was added, run:

claude mcp list

Verify the connection

Once configured, ask your AI assistant a basic catalog question:

What products are in this catalog?

If the connection works, your AI assistant describes the Pacific Apparel catalog or returns example products. If it responds with a generic answer, see Test the Conversations server for more verification prompts.

Next steps

Tools: review the tools your AI assistant can call.
Best practices: tips for building prototypes on the sample catalog.