BigQuery MCP Data Agent

An AI-powered data analysis agent built with Google's Agent Development Kit (ADK), Gemini, BigQuery MCP Server, and Cloud Run.

Overview

This project demonstrates how an AI agent can connect to structured data in BigQuery through the Model Context Protocol (MCP).

The agent analyzes the NYC Citi Bike dataset and can answer analytical questions by inspecting the dataset, generating SQL queries, and retrieving data from BigQuery.

Related MCP server: BigQuery MCP Server

Technologies

• Google Agent Development Kit (ADK)

• Gemini

• BigQuery

• BigQuery MCP Server

• Model Context Protocol (MCP)

• Google Cloud Run

• Python

Dataset

The agent works with the public NYC Citi Bike dataset:

bigquery-public-data.new_york_citibike

The dataset contains information about Citi Bike trips and stations in New York City.

How It Works

1. The user asks a question about the Citi Bike data.

2. Gemini analyzes the question.

3. The ADK agent uses BigQuery MCP tools to inspect the dataset.

4. The agent generates and executes read-only SQL queries.

5. BigQuery returns the requested data.

6. Gemini analyzes the results and provides an answer.

MCP Tools Used

The agent uses the following BigQuery MCP tools:

• get_dataset_info

• list_table_ids

• get_table_info

• execute_sql_readonly

Example Question

We have budget for 3 coffee trucks.
We want to find the best city bike stations to place our coffee trucks.