Skip to main content

​
Overview

The QA Wolf AI can do more than create individual flows. You can use it to perform bulk operations:
  • creating multiple flows at once
  • refactoring shared logic
  • updating selectors across your test suite
  • etc.
Combined with SKILL.md files that customize the AI’s knowledge, this turns the agent into a powerful productivity tool for managing large test suites.

​
How to perform bulk edits

You can give the AI a list of tasks in a single message and it will execute them sequentially.

​
Create multiple flows at once

Instead of creating flows one at a time, describe them all in a single prompt: 
Create the following flows:
- Create Account
- Forgot Password
- Account Settings
The AI will create each flow along with the proper syntax.

​
More examples of bulk operations

Split a large flow

“Split the Checkout flow into three smaller flows: Add to Cart, Enter Shipping Info, and Complete Payment.”

Find all usages of a locator

“Find every flow that uses the data-testid='submit-btn' locator and replace them with data-testid='submit-button'.”

Copy setup steps between flows

“Copy the Arrange section from the Login flow into the Account Settings flow.”

Extract helper functions

“Refactor the login steps that appear in the Checkout, Profile, and Settings flows into a shared helper function.”

Add AAA comments

“Add Arrange, Act, Assert comments to every step in the Checkout flow.”

Standardize waits and timeouts

“Update all hardcoded waitForTimeout calls to use waitForSelector or toBeVisible assertions instead.”

Add error handling to flows

“Add try/catch blocks with meaningful failure messages to the payment step in every Checkout-related flow.”

Parameterize test data

“Refactor the Sign Up flow to accept email and password as environment variables instead of hardcoded values.”

Add assertions to existing flows

“Add a visibility assertion after every navigation step in the Onboarding group.”

Audit locator strategy

“List every flow that uses legacy playwright selectors and suggest replacements using getByRole or the proper Playwright locator syntax.”

Generate negative test cases

“For each form in the Registration flow, generate a companion flow that tests invalid inputs and expected error messages.”

Document flows with descriptions

“Add a one-line description comment at the top of every flow in the Payments group summarizing what it tests.”
For best results, be specific about which flows or groups you want the AI to modify. Naming the exact flows, groups, selectors, or patterns helps the AI apply your changes accurately.

​
How to customize AI behavior with SKILL.md files

SKILL.md files let you give the AI persistent context about your project, such as:
  • coding conventions
  • preferred selectors
  • shared utilities
  • domain knowledge
  • etc.

​
What is a SKILL.md file?

A SKILL.md file is a Markdown file with a structured header that the AI reads before responding. Use it to encode standards and patterns you’d otherwise repeat in every conversation.

​
SKILL.md format

Every SKILL.md file must start with the following header: 
---
name: the-name
description: the-description
---

The rest of the body goes here.
FieldPurpose
nameA short identifier for the skill (e.g., selector-conventions)
descriptionA brief summary the AI uses to decide when this skill is relevant
BodyThe full instructions, examples, and rules the AI should follow
Keep each SKILL.md file in its own folder for clarity and easy management.

​
Example: Flow and test guidelines

---
name: flow-guidelines
description: coding guidelines for writing flow files and tests
---

## Filename convention

- make sure file names don't contain spaces
- Good: \"clear-all-chat.flow.js\"
- Bad: \"Clear All Chats.flow.js\"

## Test content

- All empty tests should contain \"// Arrange\", \"// Act\", \"// Assert\" comments
- Only the first test needs the following code in the Arrange section:

  ```js
  const { context } = await launch();
  page = await context.newPage();
  ```

- The `let page` variable should be defined above, so it can be re-used across tests

​
Tips for effective bulk prompts

  • Be explicit about scope. Name the flows, groups, or patterns you want the AI to act on. “All flows in the Checkout group” is better than “all checkout flows.”
  • Combine operations when they’re related. “Refactor the login steps into a helper function and update all flows that use them” works well as a single request.
  • Review changes before publishing. Bulk edits are powerful—review the AI’s changes in Code Mode before promoting them.
  • Use SKILL.md files for recurring guidance. If you find yourself repeating the same instructions, encode them in a SKILL.md so the AI applies them automatically.
Last modified on February 19, 2026