Model Context Protocol moderate risk

Databricks Genie

This MCP server enables LLMs to interact with Databricks Genie for natural language querying and data exploration, requiring secure handling of Databricks credentials.

Connections & Capabilities

Connects To

GitHubDocker

Capabilities

readexec

Quickstart

Install

pip install -r requirements.txt

Exposed MCP Tools (4)

safe

get_genie_space_id()

Lists available Genie space IDs and titles.

Read-only operation, no side effects.

safe

get_space_info(space_id: str)

Retrieves title and description of a Genie space.

Read-only operation, no side effects.

moderate

ask_genie(space_id: str, question: str)

Starts a new Genie conversation and gets results based on a question.

Executes queries, potential for resource consumption.

moderate

follow_up(space_id: str, conversation_id: str, question: str)

Continues an existing Genie conversation with a follow-up question.

Executes queries, potential for resource consumption.

Safety Assessment

This server is relatively safe for read-only operations within Databricks Genie. However, the risk of token exposure and potential data leakage necessitates careful handling of credentials and input validation. Avoid using it with highly sensitive data without additional security measures.

Uses API keys for authentication, allowing for revocation.
Genie API likely has its own internal access controls.
Limited to querying and interacting within defined Genie spaces.
No direct filesystem access or shell execution.

Requires storing Databricks token, posing a risk if exposed.
Potential for data leakage if Genie space contains sensitive information.
Vulnerable to prompt injection if user input is not sanitized.
Lack of explicit rate limiting could lead to excessive API usage.