Skip to main content

​
BigQuery

Shanone’s BigQuery integration gives your agent 40 tools for Google’s data warehouse — running queries, managing datasets/tables/models, loading and exporting data, and controlling IAM access.

​
Getting Started

1

Create a service account

In Google Cloud Console, create a service account with the BigQuery permissions you want Shanone to have (e.g. BigQuery Data Editor, BigQuery Job User), then generate a JSON key for it.
2

Connect in Shanone

Ask your agent to query BigQuery, or paste the service account key JSON proactively into Integrations in the Shanone dashboard.
3

Retry your request

Once the key is stored, shanone_execute_tool calls for bigquery_* tools will succeed.
BigQuery uses a Service Account Key (JSON) rather than OAuth or a simple API key. Shanone exchanges this key for short-lived access tokens internally.

​
Available Tools

Shanone provides 40 tools for BigQuery, organized into these categories:
bigquery_query, bigquery_query_and_wait
bigquery_list_tables, bigquery_get_table, bigquery_list_partitions, bigquery_schema_to_json, bigquery_update_table, bigquery_create_table, bigquery_delete_table, bigquery_list_rows, bigquery_preview_table
bigquery_list_datasets, bigquery_get_dataset, bigquery_list_projects, bigquery_get_service_account_email, bigquery_update_dataset, bigquery_create_dataset, bigquery_delete_dataset
bigquery_list_jobs, bigquery_get_job, bigquery_cancel_job, bigquery_delete_job_metadata
bigquery_load_table_from_uri, bigquery_load_table_from_json, bigquery_copy_table, bigquery_extract_table, bigquery_insert_rows, bigquery_insert_rows_json
bigquery_list_models, bigquery_get_model, bigquery_update_model, bigquery_delete_model
bigquery_list_routines, bigquery_get_routine, bigquery_create_routine, bigquery_update_routine, bigquery_delete_routine
bigquery_get_iam_policy, bigquery_set_iam_policy, bigquery_test_iam_permissions

​
Common Use Cases

Ad-hoc data analysis

Run a SQL query against a dataset and summarize the results in plain language

Data pipeline ops

Load data from a GCS URI into a table, then monitor the load job to completion

Schema & access audits

Review table schemas and IAM policies before granting a new team access to a dataset

ML model management

List and inspect BigQuery ML models used by a reporting pipeline

​
Troubleshooting

Check the service account’s IAM roles with bigquery_test_iam_permissions — most read operations need at least bigquery.jobs.create and bigquery.tables.getData on the target dataset.
Long-running queries return a job reference rather than blocking. Poll bigquery_get_job until status.state is DONE, or use bigquery_query_and_wait for short queries where blocking is acceptable.
Check bigquery_get_job for the load job’s errorResult and errors fields — schema mismatches or malformed rows in the source file are the most common cause.