MCP BigQuery Server

What is MCP BigQuery Server?

MCP BigQuery Server is a lightweight server written in Go that allows you to query BigQuery tables using the Model Context Protocol (MCP). It is designed for developers who need to interact with BigQuery datasets through MCP-compatible clients like Cursor.

How to use MCP BigQuery Server?

Install via Homebrew with brew tap monochromegane/tap && brew install monochromegane/tap/mcp-bigquery-server. Then configure it in Cursor by adding a JSON entry to .cursor/mcp.json with the command mcp-bigquery-server start and flags --project and --dataset (repeatable).

Key features of MCP BigQuery Server

• List all allowed datasets
• List tables in a specified dataset
• Get the schema of a specified table
• Dry run a query to estimate cost and time

Use cases of MCP BigQuery Server

• Discover which BigQuery datasets are available for querying
• Explore table listings and schemas within a dataset
• Estimate query costs before running a full query
• Perform ad‑hoc SQL queries via an MCP‑enabled editor

FAQ from MCP BigQuery Server

What tools does MCP BigQuery Server provide?

It provides four tools: list_allowed_datasets, list_tables, get_table_schema, and dry_run_query.

How do I install MCP BigQuery Server?

Install using Homebrew with brew tap monochromegane/tap && brew install monochromegane/tap/mcp-bigquery-server.

How can I specify multiple datasets?

Pass the --dataset flag multiple times in the configuration, for example --dataset test1 --dataset test2.

What license is MCP BigQuery Server under?

It is licensed under the MIT License.