Skip to main content
This guide shows you how to configure the Databricks connector for WRITER Agent. After setting up this connector, WRITER Agent can perform operations like querying Databricks SQL, accessing Unity Catalog functions, and interacting with Genie spaces through Databricks’ managed MCP servers.
Databricks managed MCP servers are currently in Beta. Workspace administrators control access to this feature through the Databricks Previews page. If you don’t see MCP servers available in your workspace, navigate to Settings → Previews and search for “Managed MCP Servers” to enable the feature. This feature is not available for free-tier accounts.

​
Set up the Databricks connector

Configure the Databricks connector in AI Studio under Connectors & Tools. The Databricks connector uses OAuth to authenticate with Databricks-managed MCP servers.
The Databricks connector only supports organization-managed OAuth. WRITER-managed OAuth is not available for Databricks.

​
Obtain your Databricks MCP server URL

Before configuring the connector in AI Studio, locate your MCP server endpoint in your Databricks workspace:
  1. Log in to your Databricks workspace
  2. Navigate to AI/ML → Agents → MCP Servers
  3. Select the managed MCP server you want to connect (for example, SQL or Genie)
  4. Copy the MCP server endpoint URL for use in AI Studio The URL format depends on the type of MCP server:
    Server typeExample URL format
    Vector searchhttps://<workspace-hostname>/api/2.0/mcp/vector-search/{catalog}/{schema}
    DBSQLhttps://<workspace-hostname>/api/2.0/mcp/sql
    Genie spacehttps://<workspace-hostname>/api/2.0/mcp/genie/{genie_space_id}
    Unity Catalog functionshttps://<workspace-hostname>/api/2.0/mcp/functions/{catalog}/{schema}
For more details about Databricks managed MCP servers and how to get the IDs for specific server types, see Databricks MCP documentation.

​
Create a Databricks OAuth application

Create an OAuth 2.0 application in Databricks:
  1. Navigate to the Databricks Account Console
  2. Select Settings in the left navigation
  3. Select App connections
  4. Select Add connection
  5. Enter an application name (for example, “Writer MCP connector”)
  6. Add the Writer redirect URI: 
    https://app.writer.com/mcp/oauth/callback
  7. Select All APIs under Access Scopes
  8. Ensure Generate a client secret is selected
  9. Set the access token TTL (for example, 10 minutes) and refresh token TTL (for example, 129600 minutes or 90 days)
  10. Select Create
  11. Copy the client ID and client secret for use in AI Studio
For detailed instructions, see Databricks OAuth documentation.

​
Configure the connector in AI Studio

After creating your Databricks OAuth application:
  1. Navigate to Connectors & Tools in AI Studio
  2. Select the Databricks connector
  3. Select who has access by default (all users or specific teams)
  4. Enter your Databricks MCP server endpoint URL as the URL
  5. Enter your OAuth client ID and client secret

​
Next steps