BigQuery MCP server

Last verified: 2026-09

Query Google BigQuery datasets.

Community Databases & data Needs a secret Claude Desktop, Cursor, Windsurf, Claude Code, VS Code

What it does

Run queries and inspect datasets in a GCP project. Needs Application Default Credentials or a service-account JSON. Treat the service account as the model's identity.

It sits in Databases & data: A database MCP is SQL or an API with your credentials.

Good for

Run a dry SELECT on one dataset the service account can see.
List tables in a project without opening the console.
Let the model draft SQL you then run in the console yourself.
Keep the job user + dataViewer role, never Owner.

Tools

query / run — Execute SQL in the project.
list datasets / tables — Discover what the account can see.

Config (Claude Desktop / Cursor / Windsurf)

Paste the JSON below. Same mcpServers shape. Replace placeholder paths and secrets. Windows paths look like C:\\Users\\you\\project, not /path/to. Host-side steps →

Claude Desktop

OS	Config file
macOS	`~/Library/Application Support/Claude/claude_desktop_config.json`
Windows	`%APPDATA%\Claude\claude_desktop_config.json` (usually `C:\Users\<you>\AppData\Roaming\Claude\`)
Linux	`~/.config/Claude/claude_desktop_config.json`

Cursor

Scope	macOS / Linux	Windows
This project	`.cursor/mcp.json` in the repo root
This user	`~/.cursor/mcp.json`	`%USERPROFILE%\.cursor\mcp.json`

Windsurf

OS	Config file
macOS / Linux	`~/.codeium/windsurf/mcp_config.json`
Windows	`%USERPROFILE%\.codeium\windsurf\mcp_config.json`

{
  "mcpServers": {
    "bigquery": {
      "command": "npx",
      "args": [
        "-y",
        "@ergut/mcp-bigquery-server"
      ],
      "env": {
        "GOOGLE_APPLICATION_CREDENTIALS": "/path/to/sa.json"
      }
    }
  }
}

Windsurf remote MCP: use serverUrl instead of url if the block below is HTTP.

One-liner: npx -y @ergut/mcp-bigquery-server

Secrets it wants: GOOGLE_APPLICATION_CREDENTIALS

Source repo →

How to get started

Create a service account with bigquery.jobUser + dataViewer on one dataset.
Point GOOGLE_APPLICATION_CREDENTIALS at that JSON. Never use an Owner key.
Ask it to list tables in that dataset, then a dry SELECT.

Access risk

BigQuery jobs cost money. A key with edit access can overwrite tables.

When to skip it

Skip it for Snowflake/Postgres, or if you only need a CSV (Filesystem + SQLite).

Instead: Snowflake, PostgreSQL, Google Cloud.

Databases & dataSnowflakeRun SQL on Snowflake. OfficialPostgreSQLRun SQL against a Postgres database. Cloud & infraGoogle CloudCall selected GCP APIs.

Vs alternatives

Server	Official?	Needs a secret?	Best for
BigQuery	No	Yes	Query Google BigQuery datasets.
Snowflake	No	Yes	Run SQL on Snowflake.
PostgreSQL	Yes	Yes	Run SQL against a Postgres database.
Google Cloud	No	Yes	Call selected GCP APIs.

More in Databases & data

OfficialPostgreSQLRun SQL against a Postgres database. OfficialRedisGet and set keys in Redis. OfficialSQLiteQuery a local SQLite file. Databases & dataAirtableRead and write Airtable bases.

All 11 in this category →

FAQ

What identity does it use?

GOOGLE_APPLICATION_CREDENTIALS or ADC. That service account is the model. dataViewer + jobUser on one dataset.

Do queries cost money?

Yes. On-demand bytes scanned. Limit the tables you expose.

BigQuery or the GCP Toolbox listing?

This page is BigQuery SQL. The gcp listing on this site is Toolbox for databases — different package.

Can it create tables?

If the SA can. Do not give it edit access while you are evaluating.