> ## Documentation Index
> Fetch the complete documentation index at: https://docs.rippit.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Databricks

> Let the Rippit agent query Unity Catalog through Databricks SQL.

Databricks plays the same role as Snowflake: it supplies the business data that conversation analysis needs to be joined against.

## What the agent gets

SQL execution against Unity Catalog, through Databricks SQL. Unity Catalog uses a three-level namespace, `catalog.schema.table`, so queries should be fully qualified.

## Setup

The MCP server lives in your own workspace, so there is no shared endpoint.

<Steps>
  <Step title="Note your workspace MCP URL">
    It follows this shape:

    ```text theme={null}
    https://<workspace-hostname>/api/2.0/mcp/sql
    ```

    Rippit accepts `.databricks.com` and `.azuredatabricks.net` hosts.
  </Step>

  <Step title="Create an OAuth application">
    Databricks does not let Rippit register itself. Create the OAuth application once at the **account** level, and add the redirect URI Rippit shows in the connect dialog, copied exactly.
  </Step>

  <Step title="Connect in Rippit">
    Paste the workspace MCP URL and the application's client ID and secret.
  </Step>
</Steps>

## Identity and scope

The connection uses the Databricks identity of the person who authorized it, so Unity Catalog grants apply as usual.

An organization-wide connection shares that identity with everyone in your Rippit organization. As with Snowflake, a purpose-built service principal or user whose grants match what the agent should read is a better choice than an individual's account.

## Tools

| Tool                    | Access      | What it does                                                            |
| ----------------------- | ----------- | ----------------------------------------------------------------------- |
| `execute_sql_read_only` | Read        | Run read-only SQL against Unity Catalog                                 |
| `execute_sql`           | Write       | Run SQL including `INSERT`, `UPDATE`, `CREATE TABLE`, and `ALTER TABLE` |
| `poll_sql_result`       | Unspecified | Fetch results of a long-running query by statement ID                   |

Only `execute_sql_read_only` is enabled by default.

<Warning>
  `execute_sql` runs any SQL, not just reads. Enabling it lets the agent modify and drop data in any catalog the connected identity can write to. If you only need analysis, leave it off.
</Warning>