Overview
This MCP server acts as a translator between AI models and Google BigQuery, allowing LLMs to access and query your data securely through the Model Context Protocol. It enables natural language interactions with BigQuery, returning results in plain English and avoiding manual SQL writing for common questions. The server exposes read-only access to both tables and materialized views within your datasets, and it can explore dataset schemas with clear labeling of resource types (tables vs views). It enforces a safe processing limit (1 GB by default) to prevent heavy queries, and relies on Google Cloud authentication (either gcloud application-default login or a service account key) to control access. The project can be configured via Claude Desktop's config file, and the current integration targets Claude Desktop (developer preview). Prerequisites include Node.js 14+, a Google Cloud project with BigQuery enabled, and either the Cloud CLI or a service account key. Quick install via Smithery automates the setup and updates Claude Desktop configuration; a manual setup path guides authentication, config, and start commands.
Features
Natural language SQL querying
Run SQL queries by asking questions in plain English, eliminating manual SQL writing.
Access to tables and materialized views
Access both tables and materialized views in your datasets.
Schema exploration with labeled resources
Explore dataset schemas with clear labeling of resource types (tables vs views).
Safe data processing limits
Analyze data within safe limits (1GB query limit by default).
Read-only data security
Enforces read-only access to keep data secure.
Who Is This For?
- AI developers:Integrate MCP-enabled BigQuery access into Claude Desktop to query data naturally.
