Log inSubmit server
BigQuery Data Platform logo

BigQuery Data Platform

Read-only BigQuery discovery and query tools with query-cost controls, freshness checks, and scheduled-work monitoring.

InstallRepositoryWebsite
Full README Compare alternatives Embed widget Claim this listing

Install BigQuery Data Platform

Pick your client, then add it in one click or paste the config.

STDIOPython
~/Library/Application Support/Claude/claude_desktop_config.json
claude_desktop_config.json
{
  "mcpServers": {
    "bigquery-data-platform": {
      "command": "uvx",
      "args": [
        "data-platform-mcp"
      ],
      "env": {
        "BQ_CODE_ASSET_LOCATION": "YOUR_VALUE_HERE"
      }
    }
  }
}

Settings → Developer → Edit Config, merge this into mcpServers, then fully quit and reopen Claude Desktop.

Values are written into the snippet above so you can paste it as is. They never leave your browser.

Automated check passed— started and listed 14 tools correctly (3d ago).

Overview

BigQuery Data Platform MCP server exposes read-only BigQuery discovery, lineage, scheduling, code-asset, and query tools to MCP clients. It helps an AI client inspect datasets and schemas, identify stale or dependent assets, review scheduled work, and execute validated SELECT statements using the operator’s Google credentials. Reach for it when plain-language questions need answers grounded in BigQuery metadata and data without permitting mutations. Query execution can scan data, while most discovery operations are free.

Use cases

  • Discover datasets and table schemas before answering a data question
  • Check whether a source table is stale before trusting its results
  • Review scheduled queries and notebook runs for failed data jobs
  • Trace code assets that read a table before changing its schema
  • Run validated read-only SELECT queries through an MCP client

Key features

  • Read-only BigQuery metadata and query access
  • Partitioning, freshness, size, and row-count inspection
  • Scheduled-query and notebook-run monitoring
  • BigQuery Studio code-asset discovery
  • Multiple environments with a selectable default
  • Validated SELECT execution with query-cost awareness

Tools14

What BigQuery Data Platform lets your agent do. Expand a tool for its parameters.

Verified live~3.4k tokens of context

Security & health

What it can reach and how reliably it runs — from our own checks, not the maintainer’s claims.

ActiveLast checked Today

Recent health check succeeded.

4/4 checks healthy over the last 40d

What it can do

From the 14 tools this server returned when we ran it; 14 declare their own safety hints.

Makes network requests · 13Read-only tools · 14
Network at startup2 connections
Files written at startup30

Observed 2026-10-19 in an isolated sandbox. A description of reach, not a security audit.

Supply-chain signal

No high-severity advisories surfaced by our automated scan.

Critical 0High 0Medium 0Low 0

Scanned 15d ago via OSV.dev · data-platform-mcp (PyPI)

Adoption & maintenance

Factual signals from GitHub, npm, and our automated checks — not a rating.

Last commit1mo agoMost recent push to the default branch.
Availability100%Our rolling endpoint + install checks that succeeded.
Install checkPassedOur sandbox started it and listed its tools.
Tools exposed14Callable tools this server registers over MCP.
Directory activity2 viewsConfig copies, upvotes, and views on AllMCPs.

How BigQuery Data Platform works

What the BigQuery Data Platform MCP server does

The BigQuery Data Platform MCP server gives an MCP client read-only access to BigQuery information and selected BigQuery Studio assets. It supports metadata discovery, table freshness checks, scheduled-query inspection, code-asset lookup, and validated SELECT queries. The server does not translate natural language into SQL; the connected AI client performs that translation, while the server executes the resulting query under the configured Google credentials.

Metadata tools can list datasets and tables, return expanded nested-column paths, report partitioning, size, and row counts, and identify when tables were last written. These checks help distinguish an absent table from a table that still exists but has stopped receiving data.

Tools and capabilities

Available capabilities include:

  • Listing datasets, tables, and views.
  • Inspecting schema, partition metadata, storage size, and row counts.
  • Checking table write freshness.
  • Listing configured environments and identifying the default.
  • Reviewing scheduled queries, their SQL, destinations, recent runs, disabled state, or failures.
  • Listing BigQuery Studio notebooks, saved queries, and data canvases, then retrieving individual asset bodies.
  • Finding code assets that read a specified table.
  • Listing scheduled notebook executions, inspecting schedules, and showing recent failures.
  • Running a validated, read-only SELECT and returning rows.

Only run_query scans data and may incur query charges. find_code_assets_using_table consumes Dataform read quota rather than query billing, and its results cover only the assets it could inspect. Unreadable assets are reported separately, so an incomplete scan is not evidence that no other asset uses the table.

How it works

Each tool can target an optional environment; omitting that argument selects the configured default. The server uses BigQuery APIs for warehouse metadata and queries. Scheduled-query information comes from the BigQuery Data Transfer Service. BigQuery Studio code assets are stored as Dataform repositories, so those tools use the Dataform API. Notebook schedules and executions are read through Vertex AI REST endpoints.

The BigQuery Data Platform MCP server strips notebook outputs before returning notebook bodies and reports the reduction. Code-asset locations are regional and may differ from the dataset region. When a multi-region is configured, the server can probe regions within it; an explicit code-asset location disables that probing.

Notebook execution results are filtered locally because Vertex AI does not support the relevant server-side outcome filter. Results are read newest-first and bounded by a lookback period, which defaults to 30 days. Runs from deleted schedules remain visible and are labelled accordingly.

Setup and configuration

Configuration uses a TOML file such as ~/.config/data-platform-mcp/config.toml. It can define a default environment and multiple named environments, each with a project, optional service-account impersonation, and an optional dataset allowlist. The supplied example uses a read-only service account for the project.

Code-asset location can be pinned with the code_asset_location setting or the BQ_CODE_ASSET_LOCATION environment variable. No location setting is needed when the server should probe regions associated with a configured multi-region.

Additional permissions are needed for optional resource groups. Scheduled-query tools require roles/bigquerydatatransfer.viewer; code-asset tools require roles/dataform.viewer; notebook scheduling and execution tools require roles/aiplatform.viewer. If one of these roles is absent, the affected tools report the missing role while other functions continue to work.

Limitations and notes

The server is read-only: its query tool accepts SELECT statements and does not provide mutation operations. Queries can still scan substantial amounts of data, so metadata discovery should precede execution.

Partitioning is read from table metadata rather than inferred from column names. A date-like column does not reduce scan cost unless the table is actually partitioned. Scheduled-query destination detection may extract a target from SQL when the query does not declare one; that result is labelled as a heuristic.

Dataform reads are quota-limited and may retry with backoff. Persistent exhaustion is reported as something to retry later. Notebook schedule status alone does not prove that notebook code succeeded; execution-job outcomes are used to identify failures.

Read the full README View source on GitHub

Reviews

No reviews yet — be the first to share how it worked for you.

Frequently Asked Questions about BigQuery Data Platform

BigQuery Data Platform is an MCP server that connects AI clients to Google BigQuery using configured Google credentials. Its main tools list datasets and tables, inspect schemas and freshness, review scheduled queries and code assets, monitor notebook runs, and execute validated read-only SELECT queries. It does not perform mutations.

Install details are provided through the project's documentation and its PyPI package, `data-platform-mcp`. Configure the server with a TOML file defining a default environment and project, then provide any required Google authentication and permissions for the resources you want to inspect.

Yes. The README names Claude Desktop as a supported MCP client. Claude Desktop can use the server to discover BigQuery metadata, inspect supported scheduled and code assets, and submit read-only SELECT queries through the server.

No. The server is read-only and its query operation is limited to validated SELECT statements. It exposes discovery and inspection tools rather than insert, update, delete, schema mutation, or other write operations.

It uses `check_table_freshness` to report when tables were last written. This can reveal sources that still exist but have stopped updating. Scheduled-query and notebook-run tools can then help investigate disabled schedules or failed executions.

Related MCP servers

All in Data Platforms All alternatives
Add a badge to your READMEShow it’s listed — live SVG, Markdown or HTML.

AllMCPs Directory Badge

Full Badge Customizer

Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.

Badge Style:
Live Dynamic SVG PreviewBigQuery Data Platform AllMCPs Directory Badge
Markdown (GitHub README)
[![AllMCPs](https://allmcps.com/api/badge/bigquery-data-platform?style=directory)](https://allmcps.com/mcp/bigquery-data-platform)
HTML Embed
<a href="https://allmcps.com/mcp/bigquery-data-platform"><img src="https://allmcps.com/api/badge/bigquery-data-platform?style=directory" alt="BigQuery Data Platform on AllMCPs" /></a>