Redgate Test Data Manager

Drive Test Data Manager through your AI coding assistant. Paste a prompt, answer a few setup questions, and end up with a masked or subsetted database.

Important – In all cases, we recommend you use a separate, non-essential environment when testing Anonymize and Subset. We cannot accept responsibility for any issues resulting from setting up Anonymize and Subset. Any activities carried out on production or otherwise important systems are not advised and at your own risk.

What this is

Two prompts you paste into an AI coding tool (Claude Code, Cursor, GitHub Copilot, or similar). Each one walks the tool through the whole Test Data Manager workflow:

Anonymize: install the CLI, authenticate, classify your data, build a masking config, dry run, mask.
Subset: install the CLI, authenticate, prepare a target database, configure the subset, dry run, run.

The AI does the work. You confirm the database name when it matters, paste the occasional query result back, and approve before anything destructive runs.

When to use it

Choose this path if:

You already use Claude Code, Cursor, GitHub Copilot, or a similar coding assistant with shell access.
You want a guided walkthrough without reading the full CLI reference upfront.
You prefer to drive things from your editor or terminal.

Choose the GUI if you want a point-and-click setup.

Use the CLI directly if you're wiring Test Data Manager into CI/CD or want explicit control over every command.

What you'll need

A database to anonymize or subset (SQL Server, PostgreSQL, MySQL/MariaDB, or Oracle).
A test environment. Don't point this at production data.
An AI coding tool with shell access on your machine.
A Redgate account, or start a free trial during the auth step.
For subset: a separate target database with the same schema as your source. The prompt helps you create one if you don't have one yet.

How it works

Paste the prompt, tell it which database engine you're using, and the AI takes it from there. It will:

Check whether the Test Data Manager CLI is installed. If not, download and extract it for you.
Run the licence activation. A browser window opens; you log in.
Build a connection string from the non-secret details (server, port, database, username) and write it to a local redgate.env file with a placeholder for your password.
Ask you to add your password to that file directly, then take over from there.
Run the workflow, summarising results, and asking for explicit confirmation before anything that modifies data.
Offer to write a dated run record so you have an audit trail.

How your database password is kept out of the chat

The AI should never see your password. The prompt builds a redgate.env file with an <ADD_YOUR_PASSWORD> placeholder, then asks you to replace the placeholder in your text editor. The CLI reads the value from an environment variable that the AI loads at runtime and should not hold the value itself. If your working directory is a git repository, the AI also adds redgate.env to .gitignore for you.

Pick a workflow

Anonymize a database: mask PII while keeping referential integrity.
Create a subset: get a right-sized copy of your production data.