ergut/mcp-bigquery-server

Overview

Gives your AI assistant read-only access to BigQuery, so it can answer questions about your warehouse data without any risk of it changing or deleting anything. A safer starting point than a full read-write connector if you want to let an assistant explore business data but keep it away from write access. Requires your own BigQuery project and credentials.

Best for

Best for
Businesses that want safe, read-only AI access to their BigQuery data

Use cases

• Ask a question about warehouse data with no risk of accidental changes
• Pull a read-only summary of sales or usage figures for a report
• Let a broader team query business data safely through an assistant
• Explore a new dataset without write permissions to worry about

How to use

Install

npx -y @ergut/mcp-bigquery-server

Tools exposed

• npx

Tested with

Claude Desktop, Claude Code

Example client config

{\n  "mcpServers": {\n    "bigquery": {\n      "command": "npx",\n      "args": [\n        "-y",\n        "@ergut/mcp-bigquery-server",\n        "--project-id",\n        "your-project-id",\n        "--location",\n        "us-central1",\n        "--config-file",\n        "/path/to/config.json"\n      ]\n    }\n  }\n}

Notes

A Model Context Protocol (MCP) server that provides secure, read-only access to BigQuery datasets. It enables Large Language Models (LLMs) to safely query and analyze data through a standardized interface.

140 stars on GitHub. Last updated 2026-05-22. Licensed MIT.

Use cases

• Let an LLM run read-only SQL queries on BigQuery tables
• Give an AI assistant access to business data for analysis
• Integrate BigQuery with MCP-compatible tools and agents

Pros

• Read-only design prevents accidental data modification
• Standard MCP interface works with many LLM frameworks
• Written in TypeScript for type safety and easy integration

Cons

• Limited to read-only queries, no write or DDL operations
• Requires BigQuery credentials and network access to Google Cloud
• MCP ecosystem still maturing, fewer clients than REST APIs

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.