Databricks FAQs

Question 1

Is authentication handled automatically for deployed apps?

Accepted Answer

Yes, for applications deployed on Databricks Apps, the integrated proxy automatically handles Databricks OAuth authentication. This ensures secure access to your workspace APIs using your existing Databricks CLI credentials without manual intervention.

Question 2

Which AI assistants are supported for integration?

Accepted Answer

This server is designed to integrate with AI assistants that support the Model Context Protocol (MCP), prominently featuring integration with Claude CLI. It acts as a bridge, defining what these AI assistants can 'see' and 'do' within your Databricks environment.

Question 3

How can I add custom prompts and tools?

Accepted Answer

You can add prompts by simply creating markdown files in a designated folder, which are dynamically loaded. For custom tools, you write Python functions that utilize the Databricks SDK and expose them as MCP tools using a dedicated decorator (@mcp_server.tool) within your server application.

Question 4

What is the Databricks MCP Server?

Accepted Answer

The Databricks MCP Server is a solution that enables you to host Model Context Protocol (MCP) prompts and custom AI tools directly on Databricks Apps. This allows AI assistants, such as Claude, to seamlessly interact with and execute operations within your Databricks workspace.

Question 5

What are the key benefits of deploying an MCP server on Databricks?

Accepted Answer

Deploying an MCP server allows for dynamic, AI-driven interactions with your Databricks workspace. Benefits include automating data science and ML tasks, creating dynamic AI prompts, executing custom operations via Python tools, and ensuring secure authentication through OAuth-enabled proxy, all accessible from your AI assistant.

Databricks

About

Key Features

Use Cases