Skip to main content

​
Databricks

Shanone’s Databricks integration gives your agent 155 tools spanning SQL execution, cluster/job/pipeline management, Unity Catalog administration, MLflow experiments, model serving, secrets, and workspace files.

​
Getting Started

1

Connect via OAuth

Ask your agent to do something with Databricks (e.g. “list my SQL warehouses”), or connect proactively from Integrations in the Shanone dashboard. You’ll be asked for your workspace URL (e.g. https://adb-1234567890123456.7.azuredatabricks.net) and then get a connect_link to authorize Shanone.
2

Authorize the workspace

Approve Databricks’ OAuth consent screen for the specific workspace you entered — Databricks credentials are workspace-scoped, so reconnect separately for each workspace you want Shanone to act on.
3

Retry your request

Once connected, shanone_execute_tool calls for databricks_* tools will succeed.

​
Available Tools

Shanone provides 155 tools for Databricks, organized into these categories:
databricks_execute_sql, databricks_get_statement_status, databricks_get_statement_result, databricks_cancel_statement, databricks_query, databricks_execute_async, databricks_list_query_history, databricks_get_query
databricks_list_warehouses, databricks_get_warehouse, databricks_create_warehouse, databricks_start_warehouse, databricks_stop_warehouse, databricks_delete_warehouse
databricks_list_clusters, databricks_get_cluster, databricks_create_cluster, databricks_start_cluster, databricks_terminate_cluster, databricks_delete_cluster, databricks_list_instance_pools, databricks_get_instance_pool, databricks_create_instance_pool, databricks_edit_instance_pool, databricks_delete_instance_pool, databricks_get_instance_pool_permissions, databricks_set_instance_pool_permissions
databricks_list_jobs, databricks_get_job, databricks_create_job, databricks_run_job_now, databricks_get_run, databricks_list_runs, databricks_cancel_run, databricks_get_run_output
databricks_list_pipelines, databricks_get_pipeline, databricks_create_pipeline, databricks_update_pipeline, databricks_delete_pipeline, databricks_clone_pipeline, databricks_start_pipeline, databricks_stop_pipeline, databricks_get_pipeline_update, databricks_list_pipeline_updates, databricks_list_pipeline_events, databricks_get_pipeline_permissions
databricks_list_catalogs, databricks_get_catalog, databricks_create_catalog, databricks_update_catalog, databricks_delete_catalog, databricks_list_schemas, databricks_get_schema, databricks_create_schema, databricks_update_schema, databricks_delete_schema, databricks_list_tables, databricks_get_table, databricks_list_table_summaries, databricks_table_exists, databricks_delete_table
databricks_list_volumes, databricks_get_volume, databricks_create_volume, databricks_update_volume, databricks_delete_volume, databricks_dbfs_list, databricks_dbfs_get_status, databricks_dbfs_mkdirs, databricks_dbfs_put, databricks_dbfs_read, databricks_dbfs_delete, databricks_dbfs_move, databricks_dbfs_create, databricks_dbfs_add_block, databricks_dbfs_close
databricks_list_experiments, databricks_get_experiment, databricks_get_experiment_by_name, databricks_create_experiment, databricks_update_experiment, databricks_delete_experiment, databricks_restore_experiment, databricks_search_experiments, databricks_set_experiment_tag, databricks_get_experiment_permissions, databricks_set_experiment_permissions, databricks_update_experiment_permissions, databricks_create_mlflow_run, databricks_get_mlflow_run, databricks_update_mlflow_run, databricks_delete_mlflow_run, databricks_restore_mlflow_run, databricks_search_mlflow_runs, databricks_log_metric, databricks_log_param, databricks_log_batch, databricks_set_mlflow_tag, databricks_delete_mlflow_tag, databricks_list_artifacts, databricks_get_metric_history
databricks_list_serving_endpoints, databricks_get_serving_endpoint, databricks_create_serving_endpoint, databricks_update_serving_endpoint_config, databricks_delete_serving_endpoint, databricks_query_serving_endpoint, databricks_get_serving_endpoint_logs, databricks_export_serving_endpoint_metrics, databricks_get_serving_endpoint_permissions, databricks_set_serving_endpoint_permissions, databricks_put_serving_endpoint_ai_gateway, databricks_patch_serving_endpoint_tags
databricks_create_context, databricks_get_context_status, databricks_destroy_context, databricks_execute_command, databricks_get_command_status, databricks_cancel_command, databricks_list_cluster_libraries, databricks_list_all_cluster_libraries, databricks_install_cluster_libraries, databricks_uninstall_cluster_libraries
databricks_list_secret_scopes, databricks_create_secret_scope, databricks_delete_secret_scope, databricks_list_secrets, databricks_get_secret, databricks_put_secret, databricks_delete_secret, databricks_list_secret_acls, databricks_get_secret_acl, databricks_put_secret_acl, databricks_delete_secret_acl, databricks_list_git_credentials, databricks_get_git_credential, databricks_create_git_credential, databricks_update_git_credential, databricks_delete_git_credential
databricks_list_repos, databricks_get_repo, databricks_create_repo, databricks_update_repo, databricks_delete_repo, databricks_get_repo_permissions, databricks_set_repo_permissions, databricks_update_repo_permissions, databricks_get_repo_permission_levels, databricks_list_workspace, databricks_get_workspace_status, databricks_export_workspace, databricks_import_workspace, databricks_mkdirs, databricks_delete_workspace

​
Common Use Cases

Ad-hoc data analysis

Run SQL against a warehouse and get results back without opening the Databricks UI

Pipeline & job monitoring

Check job run status, pull failure logs, and trigger reruns from a chat conversation

Unity Catalog governance

Audit catalogs/schemas/tables and manage permissions across the data estate

ML experiment tracking

Log metrics/params to an MLflow run and compare experiment results

​
Troubleshooting

Databricks OAuth tokens are workspace-scoped and expire; if a long-idle connection stops working, reconnect from Integrations to refresh the token for that specific workspace_url.
Use databricks_execute_async instead, then poll databricks_get_statement_status / databricks_get_statement_result — synchronous execution isn’t suited to long-running queries.
Unity Catalog operations (create/update/delete on catalogs, schemas, tables) require the connected identity to have the appropriate USE CATALOG/CREATE/OWNER privileges — check with a workspace admin if a tool that should work returns 403.
Call databricks_get_run_output on the resulting run ID to see the actual task error — the “now” call only confirms the run was triggered, not that it completed successfully.