Model Context Protocol moderate risk

Bigquery

This MCP server provides read and write access to BigQuery, enabling LLMs to query data, inspect schemas, and manage tables within a specified GCP project and location.

Connections & Capabilities

Connects To

GCP

Capabilities

readwriteexec

Quickstart

Install

npx -y @smithery/cli install mcp-server-bigquery --client claude

Config

{
  "mcpServers": {
    "bigquery": {
      "command": "uvx",
      "args": ["mcp-server-bigquery"],
      "env": {
        "BIGQUERY_PROJECT": "{{GCP_PROJECT_ID}}",
        "BIGQUERY_LOCATION": "{{GCP_LOCATION}}"
      }
    }
  }
}

Exposed MCP Tools (3)

high

execute-query

Executes a SQL query using the BigQuery dialect.

Can modify or delete data, depending on the query.

safe

list-tables

Lists all tables in the BigQuery database.

Read-only operation that does not modify data.

safe

describe-table

Describes the schema of a specific table.

Read-only operation that does not modify data.

Safety Assessment

This server allows both read and write operations on BigQuery. It's safe when used with appropriate IAM permissions and when queries are carefully vetted. It's risky if the service account has excessive privileges or if the LLM generates malicious queries.

Read-only access can be enforced by limiting tool usage.
Uses GCP's IAM for access control.
Queries are executed within the BigQuery environment.
Optional timeout to prevent long-running queries.

Can execute arbitrary SQL queries, potentially leading to data modification or deletion.
Requires GCP project credentials, which if compromised, could lead to unauthorized access.
No built-in query sanitization or validation.
Lack of sandboxing for query execution.