BigQuery MCP server is a Model Context Protocol server that provides access to Google Cloud BigQuery. It enables LLMs to inspect database schemas and execute queries. The server exposes three MCP tools: execute-query, which executes a SQL query using the BigQuery dialect; list-tables, which lists all tables in the BigQuery database; and describe-table, which describes the schema of a specific table. Configuration is flexible: you can pass --project and --location as command line arguments (required), optionally limiting to specific datasets with --dataset (or BIGQUERY_DATASETS) and providing a service account key via --key-file (BIGQUERY_KEY_FILE). If a key file is not provided, the server uses default credentials. The README includes quickstart steps for installing via Smithery and for Claude Desktop usage, and includes development workflows for syncing dependencies, building distributions, and publishing to PyPI. Debugging guidance points to the MCP Inspector and running the server with uv. This server is intended for developers and AI applications that need schema awareness and query execution against BigQuery while supporting authentication and dataset restrictions.