Snowflake Labs Cortex Code Migrations Unit Test Migrator Setup

name: unit-test-migrator-setup description: One-time setup for the unit test migration framework. Configures source database (SQL Server, Redshift, or Oracle) and Snowflake connections, deploys the validation infrastructure to Snowflake, and loads existing baselines. parent_skill: migrate-objects license: Proprietary. See License-Skills for complete terms

Setup - Validation Framework

One-time setup to prepare the 2-sided testing infrastructure. Run this once per project before capturing baselines or migrating code units.

Prerequisites

• A Snowflake account with CREATE SCHEMA, CREATE TABLE, CREATE PROCEDURE privileges
• Network access to source database (SQL Server, Redshift, or Oracle) for baseline capture later
• scai CLI installed

Step 1: Confirm Snowflake Connection

Show the user the current Snowflake connection:

SELECT CURRENT_ACCOUNT(), CURRENT_USER(), CURRENT_ROLE(), CURRENT_DATABASE(), CURRENT_WAREHOUSE();

Ask the user:

"You are connected to Snowflake as:

• Account:
• User:
• Role:
• Database:
• Warehouse:

Is this the correct connection for the migration?"

If not, ask the user to set the correct connection before continuing.

Step 2: Verify Source Database Connection

Check if a source connection already exists:

# Check for SQL Server connections
scai connection list -l sqlserver --json 2>/dev/null

# Check for Redshift connections
scai connection list -l redshift --json 2>/dev/null

# Check for Oracle connections
scai connection list -l oracle --json 2>/dev/null

# Check for Teradata connections
scai connection list -l teradata --json 2>/dev/null

If a connection exists, test it.

For Oracle or Teradata, first verify the driver is cached before testing:

# Oracle
ls ~/.snowflake/scai/drivers/oracle/*.nupkg ~/.snowflake/scai/drivers/oracle/*.dll 2>/dev/null

# Teradata
ls ~/.snowflake/scai/drivers/teradata/*.nupkg ~/.snowflake/scai/drivers/teradata/*.dll 2>/dev/null

If no driver files are found, follow the connection skill (Step 1) to download the driver, then include --driver-path in the test command.

scai connection test -l <sqlserver|redshift|oracle|teradata> -s <CONNECTION_NAME> --json
# Oracle/Teradata without cached driver: add --driver-path <PATH_TO_NUPKG>

If no connection exists, follow the appropriate connection skill:

• SQL Server → ../../connection/sql-server-connection/SKILL.md
• Redshift → ../../connection/redshift-connection/SKILL.md
• Oracle → ../../connection/oracle-connection/SKILL.md
• Teradata → ../../connection/teradata-connection/SKILL.md

Once connected, return here to continue setup.

Step 2b: Configure Redshift Staging (Redshift only)

If the source platform is Redshift, load ./configure-redshift-source.md to set up the S3 bucket, Snowflake stage, and IAM role required for data migration staging.

If the source is SQL Server, Oracle, or Teradata, skip to Step 3.

Step 3: Verify scai Project

The validation framework assumes a scai migration project exists with converted code.

# Check for existing project
scai project info --json

# Verify converted code exists
ls snowflake/procedure/ 2>/dev/null
ls snowflake/function/ 2>/dev/null
ls artifacts/ 2>/dev/null

If no project exists, see the parent migration/SKILL.md to set one up.

Step 4: Create Validation Schema and Stage

The validation framework stores results in a VALIDATION schema. Confirm with the user and create it if it does not exist:

CREATE SCHEMA IF NOT EXISTS VALIDATION;
CREATE STAGE IF NOT EXISTS VALIDATION.VALIDATOR_STAGE;

Step 5: Deploy Validation Schema

Deploy the validation infrastructure used by scai test (uses configure database):

scai test validate --create-schema -c <CONNECTION_NAME>

This creates:

Step 6: Verify Setup

Run a quick health check:

SHOW TABLES IN SCHEMA VALIDATION;
SHOW VIEWS IN SCHEMA VALIDATION;

SELECT COUNT(*) AS total_results FROM VALIDATION.RESULTS;

CHECKPOINT

Confirm with the user:

• Snowflake connection tested successfully
• Source database connection configured (SQL Server, Redshift, or Oracle)
• Redshift staging configured (Redshift only — S3 bucket, stage, IAM role)
• scai project exists with converted code
• Test config generated (settings/test_config.yaml — auto-created by configure())
• VALIDATION schema created
• Validation schema deployed (scai test validate --create-schema)
• Baselines captured (or ready to capture)

Next Steps

Go back to the main skill to pick the next object to migrate ../SKILL.md

Source: Snowflake-Labs/cortex-code-migrations — distributed by TomeVault.