Skip to main content

​
Prerequisites

Before you can configure and use Windsurf with Connect AI, you must first do the following:
  • Install the Windsurf IDE (MCP integrations do not support extensions or plugins).
  • Connect a data source to your Connect AI account. See Sources for more information.
  • Generate a Personal Access Token (PAT) on the Settings page. Copy this down, as it acts as your password during authentication.
  • Base64-encode your Connect AI credentials in the following format: Base64("email:PAT").

​
Connect Windsurf to the Connect AI MCP

1
Open Cascade Code Chat Assistant, either by pressing CTRL+L or by clicking the button highlighted below.
Windsurf Chat Assistant
2
In Windsurf, open the MCP configuration file, mcp_config.json, which is created during installation. The default path is C:\Users\<YourUsername>\.codeium\windsurf\mcpconfig.json. You can also open it directly from the IDE by clicking the MCPs button in the chat dialog, and then selecting the Config icon, as shown below:
Windsurf config file
3
Copy and paste the following into the mcp_config.json file. Replace the authorization header with the Base64-encoded credentials obtained in the prerequisites.
{
  "mcpServers": {
    "cdata-mcp": {
      "serverUrl": "https://mcp.cloud.cdata.com/mcp",
      "headers": {
        "Authorization": "Basic YOUR_BASE64_ENCODED_EMAIL:PAT" //Authorization obtained in the prerequisites
      }
    }
  }
}
4
Start the MCP server by clicking the MCPs button in the chat dialog and clicking the Refresh button.
Windsurf start MCP server
Once the MCP server is started, a green indicator appears.
5
You can now chat with the assistant. Windsurf lists the available tools, the connections, and a result set of a given table of a given connection.
Windsurf sample chat