Snowflake Cube Server

What is Snowflake Cube Server?

This is an MCP server that enables AI agents to interact with a Cube semantic layer. It provides tools to query and describe the data exposed by a Cube deployment.

How to use Snowflake Cube Server?

Install via Smithery or configure in your MCP client. The server exposes two tools – read_data to query the Cube REST API and describe_data to describe available data – plus two resource URIs, context://data_description and data://{data_id}, for fetching descriptions and JSON-formatted query results.

Key features of Snowflake Cube Server

• Queries Cube REST API and returns data as YAML
• Returns a unique identifier for each query result
• Provides JSON representation via data://{data_id} resource
• Describes available data via describe_data tool
• Describes available data via context://data_description resource

Use cases of Snowflake Cube Server

• Asking an AI agent to generate a Cube query and return results
• Programmatically inspecting the schema and dimensions of a Cube deployment
• Chaining queries where results need to be reused or reformatted in JSON

FAQ from Snowflake Cube Server

Does this server require a Cube deployment to be running?

Yes. The server interacts with a Cube REST API, so an accessible Cube instance is required.

How can I get the query results in JSON instead of YAML?

After calling read_data, use the returned unique identifier to retrieve JSON data from the data://{data_id} resource.

What does describe_data return?

It returns a description of the data available in the Cube deployment. The same information is available from the context://data_description resource.

How do I install this server?

The server is available on Smithery at @isaacwasserman/mcp_cube_server.

—