> ## Documentation Index
> Fetch the complete documentation index at: https://docs.revyl.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Scripts

> Custom code blocks available to Script steps

Scripts are reusable code blocks stored in the Library. Reference them from Script steps in any test to run real code: set up state, generate credentials, call your APIs, or compute values your tests need at runtime. Python, JavaScript, TypeScript, and Bash are all supported.

## What Scripts Are For

* **Mutate backend state**: seed test data, create users, reset account balances, enqueue jobs, open/close feature toggles
* **Generate credentials**: mint a fresh test user, call your auth service for a short-lived token, rotate API keys per run
* **Compute dynamic values**: unique emails (`test+{{timestamp}}@example.com`), hash challenges, signed URLs, HMAC signatures
* **Call external services**: hit your APIs, a third-party sandbox, a CI system, or a notification channel mid-test
* **Parse or transform data**: pull a value out of a response, decode a JWT, compute a checksum for validation

Scripts bridge the gap between what the AI agent does on the device and the rest of your infrastructure.

## Creating a Script

1. Go to **Library** > **Scripts** tab
2. Click **New script**
3. Name it, pick a language (Python, JavaScript, TypeScript, or Bash), and write your code
4. Save. The script is now available org-wide.

Changes to a script propagate to every test that references it. The editor warns you before saving if other tests are using it.

## Using a Script in a Test

In the [test editor](/tests/creating-tests/test-editor), add a **Script** step and select a saved script from the dropdown. The script runs in a sandboxed environment when the test hits that step.

### Saving Output to a Variable

Set `variable_name` on the Script step and the script's stdout is stored in that variable. Reference it in later steps:

```
Script step:
  script: generate_test_user
  variable_name: user_email

Next instruction:
  "Type {{user_email}} into the email field"
```

Use this to chain steps that depend on values generated at runtime. Think signup flows, one-time codes, computed IDs.

## Inline vs Library Scripts

You can also write ad-hoc code directly in a Script step without saving it to the Library. Use inline scripts for test-specific logic, and save to the Library when you want to reuse the same code across multiple tests.