The BigQuery Migration Service managed MCP extension allows users to perform tasks such as translating SQL queries into GoogleSQL syntax, generating DDL statements from SQL input queries, and getting explanations of SQL translations.

Why use the BigQuery Migration Service managed MCP server?

Google and Google Cloud managed MCP servers can be used in your AI applications with enterprise-ready governance, security, and access control.

Before you begin

In the Google Cloud console, on the project selector page, select or create a Google Cloud project. > Note: If you don't plan to keep the resources that you create in this > procedure, create a project instead of selecting an existing project. > After you finish these steps, you can delete the project, removing all > resources associated with the project.
Get your administrator to grant you the MCP Tool User role (roles/mcp.toolUser) on the Google Cloud project. If you created a new project, then you already have the required permissions.
Ensure your administrator has enabled the BigQuery Migration API on the Google Cloud project.

Configure authentication

This extension uses Google Application Default Credentials (ADC) to perform authentication. To login with ADC, run the following command in your terminal:

gcloud auth application-default login

For additional details, see the ADC documentation.

Install the extension

To install the extension, run the following command in your terminal:

gemini extensions install https://github.com/gemini-cli-extensions/bigquery-migration-service

Available tools

To see a complete list of available tools and their schemas, see the BigQuery Migration Service MCP reference.

Sample use cases

The following are sample use cases for the BigQuery Migration Service MCP server:

Using an MCP client with an IDE, translate a query file into GoogleSQL syntax.
Using an MCP client without an IDE, translate a specified query into GoogleSQL syntax.
Assess translation quality.
Get explanations of SQL translations.
Generate a DDL statement for a specified query.

Sample prompts

You can use the following sample prompts to create and manage BigQuery Migration Service resources:

Translate the DIALECT query in this FILENAME. Use PROJECT_ID and LOCATION. Persist the output and translation logs into separate directories.

When you use this prompt, the MCP client calls the translate_query tool to translate the query in the specified file. The MCP client periodically calls the get_translation tool to get the results. After the translation completes, the client writes the output to the output directory and the logs to the logs directory.
Translate this query from DIALECT: QUERY. Use PROJECT_ID and LOCATION.

When you use this prompt, the MCP client calls the translate_query tool to translate the specified query and displays the translation results.
Assess the translation quality.

When you use this prompt, the MCP client reads and examines the translation logs and displays a summary of the translation issues with suggested next steps.
Explain the translation.

When you use this prompt, the MCP client calls the explain_translation tool to get an explanation of the translation. If the translation logs contain RelationNotFound or AttributeNotFound errors, the MCP client should suggest that you create a metadata package. If you can't generate the metadata, you can send a prompt that requests the DDL statement.

A sample response looks like the following:

The translated code converts Teradata-specific features into their BigQuery equivalents. Here's a breakdown of the key changes: * MACRO to PROCEDURE: The YourMacroName macro was converted into a BigQuery stored procedure because BigQuery doesn't support macros. * SELECT INTO to SET: * For setting multiple OUT parameters in YourStoredProcedureName, the SELECT ... INTO is changed to SET (...) = (SELECT STRUCT(...)). * For single variable assignment in YourOtherProcedureName, SELECT ... INTO is replaced by SET variable = (SELECT ...) which is the standard in BigQuery. * Atomic Operations to MERGE: The BEGIN REQUEST ... END REQUEST blocks in the ProcedureA, ProcedureB, and ProcedureC procedures, which perform atomic "update or insert" operations, are translated into standard SQL MERGE statements. This is the correct and modern way to handle this logic in BigQuery.
Generate DDL for this input query.

The MCP client calls the generate_ddl_suggestion tool to start a suggestion job. The client gets the suggestion results by calling the fetch_ddl_suggestion tool. When the suggestion is available, the MCP client displays it.

If the DDL statements are correct, you can send a prompt to prepend the generated DDL statements to the query to improve the translation quality.
Prepend the generated DDL statements to the input query and retranslate.

When you use this prompt, the MCP client prepends the DDL statements to the original input query and calls the translate_query tool. The client calls the get_translation tool to get the translation. The new query translation and the logs persist when they're available.

If the generated DDL statements are correct, any RelationNotFound or AttributeNotFound errors should be resolved which results in improved translation quality.

In the prompts, replace the following:

DIALECT: The dialect of the SQL query you're translating.
QUERY: The query you're translating.
FILENAME: The file that contains the query you're translating.
PROJECT_NUMBER: Your Google Cloud project number.
LOCATION: The location of the SQL translator.

Optional security and safety configurations

MCP introduces new security risks and considerations due to the wide variety of actions that you can take with MCP tools. To minimize and manage these risks, Google Cloud offers defaults and customizable policies to control the use of MCP tools in your Google Cloud organization or project. For more information about MCP security and governance, see AI security and safety.

Quotas and limits

The BigQuery Migration Service MCP server doesn't have its own quotas. There is no limit on the number of calls that can be made to the MCP server. You are still subject to the quotas enforced by the APIs called by the MCP server tools.