How to Configure JetBrains AI Bigquery Sandbox

Published
Length
3 min read

Luis Felipe Dias

All posts

You set up a jetbrains ai bigquery sandbox by linking your code editor to a cloud project. Then you turn on the database helper inside your coding tool. Most teams start by giving the local database tool a broad cloud role. That shortcut lets developers test fast, but it opens up every table in the project. If the AI assistant reads schema details or runs queries, broad roles expose too much data. How the jetbrains ai bigquery sandbox setup works First, you need a Google Cloud project with th

Your coding agents run with your access.

Claude Code reaches prod with a developer's credentials, and approval prompts get clicked through. hoop.dev checks every query, masks sensitive data, and holds risky changes for a human.

You set up a jetbrains ai bigquery sandbox by linking your code editor to a cloud project. Then you turn on the database helper inside your coding tool. Most teams start by giving the local database tool a broad cloud role. That shortcut lets developers test fast, but it opens up every table in the project. If the AI assistant reads schema details or runs queries, broad roles expose too much data.

How the jetbrains ai bigquery sandbox setup works

First, you need a Google Cloud project with the sandbox mode turned on. The sandbox lets you run queries without adding a credit card. It puts strict limits on storage size and table life spans. Next, you open the database panel inside your JetBrains editor. You pick the BigQuery driver from the list of data sources. Then you sign in with your user account or a service account key.

Once the link works, the editor pulls down the project schema. It reads dataset names, table names, and column types. When you ask JetBrains AI for help, it uses this schema metadata. The assistant reads the table layout and writes SQL for you. You can then run that SQL straight from the editor query console. The database checks your cloud role and sends back the matching rows.

Teams often share one test project for all their local coding work. Each developer logs in and points their AI helper at the same datasets. To avoid permission errors, admins often attach project-wide reader or editor roles. This makes the initial setup quick for the whole team. Everyone can write prompts, generate SQL, and check the results right away.

What the native setup solves and where it falls short

The native setup solves basic login and cost control very well. Google Cloud checks who is logging in before any query runs. The sandbox mode also caps spending so a bad query cannot run up a huge bill. Your team gets a free place to test SQL generation with an AI helper.

Still, cloud roles only decide who can log in at the start. They do not inspect what each query does while it runs in flight. Once a developer has read rights on a table, every column is open. If the sandbox holds real customer samples, those rows come right back to the screen. Sensitive fields like emails, phone numbers, card numbers, or national IDs show up in plain text.

AI Sandbox Environments + BigQuery IAM: architecture patterns and best practices

Continue reading? Get the full guide.

Free. No spam. Unsubscribe anytime.

AI assistants also make fast mistakes when they write SQL for you. A prompt might ask for a simple check, but the model writes a table drop or update. If the developer has write permissions in the sandbox, that bad statement runs right away. Cloud roles cannot stop a single dangerous command inside an active session. They also cannot rewrite matched fields in results to hide private data on the fly.

How to tighten sandbox permissions before you start

You can lower your risk by shrinking the roles you hand out. Never use project owner or editor roles for local editor tools. Instead, give the user a role that only reads one test dataset. Keep real customer exports out of the sandbox project at all times.

You should also check what context the AI assistant sends out. The editor lets you pick which schemas attach to your chat prompts. Only attach the tables you need for the current task. Turn off auto-run features in your query console so you check every AI statement first. A human should read every SQL line before it hits the database.

Check database and API queries on the wire

Identity tools, roles, and network rules decide who opens a session. Setup alone cannot check what each query does or mask sensitive values in a response once a session starts. When workflows touch Postgres, MySQL, SQL Server, MongoDB, ClickHouse, Kubernetes APIs, SSH, or internal HTTP APIs, hoop.dev runs beside those services as an open-source proxy, a sidecar for your Postgres or API. It inspects requests coming in and results going back on the wire.

Your login details and client tools stay as they are while hoop.dev guardrails refuse bad deletes or drops before they run. On responses, hoop.dev data masking rewrites matched sensitive values in memory. Furthermore, hoop.dev holds SOC 2 Type II, is MIT licensed, and writes every query and decision to an audit trail that generates evidence for HIPAA, GDPR, and PCI DSS. See the source on GitHub.

Frequently asked questions

Does JetBrains AI need write permissions in a BigQuery sandbox?

No, the AI assistant only needs schema read permissions to write SQL queries. You only need write permissions if you plan to create or change tables yourself.

Can the AI assistant read raw table rows by itself?

By default, the assistant uses table and column names to build queries. Still, if a user pastes query results into the chat, the model reads those rows.

Why should we avoid service account keys on developer laptops?

Saved key files sit on local drives and never expire on their own. If a laptop is lost or copied, anyone with that file gets full database rights.

The query runs. The emails stay hidden.