Integration · Research and analyze

Google BigQuery and Meteor AI agents

Connect Google BigQuery to a Meteor Met: describe tables, read filtered rows and insert data after confirmation, with no free SQL. 4 real tools.

Book a demo See every integration

How does Google BigQuery connect to Meteor?

Meteor connects to Google BigQuery with the service account from your own project and gives your Met 4 tools with fixed operations: test the connection, describe a table's columns, read rows with column filters, ordering and a limit, and insert rows after confirmation. There is no free SQL: every query is built from the table and the columns that exist, and if you declare allowed tables, the connector refuses to touch any other. Each read returns up to 1,000 rows and each query is capped at 5 GB of scanned data.

What it automates

What a Meteor Met does with Google BigQuery

Answer with the figure that lives in your warehouse

Someone on the team asks about an order, a customer or a figure that sits in a BigQuery table. Your Met checks the table's columns first, builds the filter and answers only with the rows that came back. If none came back, it says so.

Store in BigQuery what the conversation leaves behind

When the conversation produces a figure worth keeping, your Met explains which row it will write, waits for confirmation and inserts it into the table you define, stamped with the current date and time.

Inspect a table before using it

The table description says which columns it has, their types and which ones are required, without reading a single row and with no data scanning cost.

Keep the Met inside the tables you choose

When you connect, you declare the list of allowed tables. If the Met or a flow asks for another one, the connector rejects it and the Met tells you exactly that, instead of looking for a workaround.

4 tools

Everything your Met can ask Google BigQuery for

Each one is a real action in the connector. You can switch off the ones you would rather it never used.

Tool	What it does
`bigquery_probar_conexion`	With allowed tables, checks it can see each one; without a list, that it can run a query.
`bigquery_describir_tabla`	Returns a table's columns with their type and mode, without reading a single row.
`bigquery_consultar_filas`	Reads rows with up to 10 filters on columns, the columns you choose, an order and a limit of up to 1,000 rows, and flags a truncated result.
`bigquery_insertar_filas`	Inserts up to 500 rows per call after explicit confirmation, optionally taking each value from another field or from the current date and time.

The limits

What the Google BigQuery integration does NOT do

We would rather say it here than have you find out during rollout.

It does not accept SQL. There is no way to write a free query: only a table, column filters, columns, an order and a limit.
It does not sum or group. It returns rows, up to 1,000 per read; it does not compute totals, averages or groupings inside BigQuery.
It does not update or delete rows, and it does not create tables or datasets. Its only write is inserting new rows into a table that already exists.
It does not filter or write nested or repeated columns. It works with single value columns such as text, numbers, dates, booleans or JSON.
It does not scan more than 5 GB per query, and one that would go beyond that stops. BigQuery bills by the columns it reads, not the rows it returns: on a large table, ask for fewer columns or filter on the partition column.

Before you start

What you need to connect Google BigQuery

1

A Google Cloud service account with its JSON key

In your project's Google Cloud console you create the service account and add a key in JSON format. In Meteor, under Connections, you paste the whole file: it is stored encrypted and never goes back to the browser or gets requested in the chat.
2

BigQuery permissions for that account

The account needs to run queries in its project and read the tables; to insert, it also needs edit permission on them and billing enabled on the project: the free BigQuery sandbox does not allow writes. You grant the permissions in Google Cloud IAM. If one is missing, the connector answers that the account has no permission for that operation.
3

Optional, the region and the allowed tables

When connecting you can set the dataset region and a list of tables written as project.dataset.table. Without a list, the connector reaches every table the service account can see.

Frequently asked questions

Questions about Google BigQuery and Meteor

Your Google Cloud project. Queries run in the project of the service account you connect, so they are billed there, on your Google account. The 5 GB scan cap per query keeps a read from sweeping an entire dataset without anyone deciding it.

No, and that is deliberate. The connector does not take SQL: it builds the query from the table, the columns that exist and the values as typed parameters. That way neither a Met nor a flow can read anything other than the table and columns they name.

Only the ones the service account can see, and if you declare allowed tables, only those. Whatever an account is not allowed to reach in Google Cloud, the connector cannot reach either.

A read returns at most 1,000 and the response flags that the result was truncated, so it is not read as the total. To get to what you need, you add filters or an order.

Step by step