Skip to main content

Configure Databricks for the AI Gateway

To allow your users to securely query Unity Catalog, run Databricks SQL, and use Genie and AI Search through their AI clients, you can configure Databricks managed Model Context Protocol (MCP) servers within the JumpCloud AI Gateway.

Prerequisites​

  • Before configuring specific enterprise applications, we recommend reviewing the Get Started: AI Gateway guide to familiarize yourself with the gateway architecture and core AI concepts (such as MCP servers and AI clients).

Completing Application Setup​

Databricks managed MCP servers authenticate with a Databricks personal access token (PAT) for the workspace. Create a PAT in your Databricks workspace settings, then use the managed MCP URL for the server you want to expose. See Databricks’ Managed MCP servers to learn more.

note

Each managed MCP server is workspace-specific. Replace {WORKSPACE-HOSTNAME} with your Databricks workspace hostname (for example, dbc-a1b2c3d4-e5f6.cloud.databricks.com). You can add multiple JumpCloud MCP server entries if you need more than one Databricks managed endpoint.

Configuring the Server in JumpCloud​

See Adding MCP Servers in Configure AI Gateway Integrations and use the following values to configure this server:

  • App: Select Databricks if the application already exists in JumpCloud. Otherwise, click + New App to open the Create New Application Integration wizard, select a prebuilt connector if available or Custom Application, then select the Manage MCP Servers checkbox. See Creating an Application from the AI Gateway to learn more.

If you created a new application, return to the Add Server modal and select it from the App drop-down before completing the remaining fields.

  • Name: Enter a name that matches the managed server you are adding, for example Databricks SQL
  • Prefix: Enter a value to prepend to MCP tool names
    • Note: Prefixes must be unique. Maximum 8 characters, letters and numbers only
  • URL: Enter the managed MCP URL for your workspace and server. Use the server names and URL patterns from Databricks’ Available managed servers:
    • Genie One: https://{WORKSPACE-HOSTNAME}/ai-gateway/mcp-services/system.ai.genie_one_mcp
    • Genie Agent: https://{WORKSPACE-HOSTNAME}/api/2.0/mcp/genie/{genie_space_id}
    • AI Search: https://{WORKSPACE-HOSTNAME}/api/2.0/mcp/ai-search/{catalog}/{schema}/{index_name}
    • Databricks SQL: https://{WORKSPACE-HOSTNAME}/api/2.0/mcp/sql
    • Unity Catalog functions: https://{WORKSPACE-HOSTNAME}/api/2.0/mcp/functions/{catalog}/{schema}/{function_name}
  • MCP authentication method: API Token
  • App API documentation link: Enter https://docs.databricks.com/aws/en/dev-tools/auth/pat

Your users will now see Databricks listed as an available MCP server when they authenticate their AI clients through the JumpCloud gateway. When prompted, users can click the documentation link you provided to learn how to create a Databricks personal access token.

Was this information helpful?