Skip to main content
Tuist’s Automated QA uses AI to test your app previews, finding bugs, UI issues, and unexpected behavior before your users do.

Overview

Automated QA leverages Claude and GPT models to interact with your app like a real user, following test prompts and reporting issues automatically.
Automated QA provides:
  • Natural language test prompts - Describe tests in plain English
  • AI-powered execution - Claude/GPT controls the app
  • Visual documentation - Screenshots and screen recordings of every test
  • Issue detection - Automatic identification of crashes, errors, and UI problems
  • VCS integration - Post test summaries to pull requests

How It Works

  1. Create a preview of your app
  2. Write a test prompt describing what to test
  3. AI agent launches the app and follows your instructions
  4. Issues are detected and documented with screenshots
  5. Test report is generated and posted to your pull request

Creating QA Tests

Basic Test

1

Create an app preview

First, create a preview of your app:
2

Run a QA test

Define what to test in natural language:
3

View results

Watch the test execute in real-time or check the results in Tuist Cloud:

Test Prompts

Write effective test prompts by:

Be Specific

“Tap the blue Login button in the top right”Not: “Log in”

Include Context

“Navigate to Settings > Account > Privacy”Not: “Check privacy settings”

Define Success

“Verify the success message appears”Not: “Submit the form”

Mention Credentials

“Use email: test@example.com, password: Test123”Not: “Sign in”

Example Test Prompts

Test Configuration

Launch Arguments

Pass launch arguments to configure your app for testing:

App Description

Provide context about your app to improve AI understanding:

Test Credentials

Configure default test credentials:
Never commit real user credentials. Use test accounts specific to QA testing.

Understanding Results

Test Report

Each QA run generates a detailed report:
  • Status - Completed, Failed, or In Progress
  • Duration - Total test execution time
  • Steps - Breakdown of actions taken
  • Issues - Bugs and problems found
  • Screenshots - Visual documentation of each step
  • Video recording - Complete screen recording of the test

Issue Detection

The AI automatically identifies:
  • App crashes or force quits
  • Unhandled exceptions
  • System errors
  • Missing or broken UI elements
  • Incorrect labels or text
  • Layout problems
  • Buttons that don’t respond
  • Failed network requests
  • Incorrect app behavior
  • Slow loading screens
  • Laggy interactions
  • Memory warnings

Test Logs

View detailed logs of AI decision-making:

VCS Integration

Automatically post test results to your pull requests:

Setup

1

Connect VCS

Link your GitHub/GitLab repository in Tuist Cloud
2

Enable PR comments

Configure QA to post comments automatically:
3

Run QA in CI

Add QA tests to your CI pipeline:

PR Comments

QA results appear as comments on your pull request:

CI/CD Integration

Use --wait flag to make CI wait for QA completion. Without it, the command returns immediately.

Best Practices

Test Critical Flows

Focus on user journeys that must work:
  • Authentication
  • Core features
  • Checkout/payments
  • Navigation

Keep Tests Focused

Each test should validate one flow:
  • Don’t combine unrelated features
  • Break complex flows into steps
  • Test edge cases separately

Use Test Data

Configure your app for testing:
  • Mock network responses
  • Use test accounts
  • Enable debug features
  • Fast animations

Review Results

Act on QA findings:
  • Watch recordings for context
  • Fix high-priority issues
  • Update test prompts as needed
  • Track issue trends

Pricing & Limits

Automated QA uses AI tokens:
  • Token usage tracked per test run
  • Costs based on test complexity and duration
  • View usage in Tuist Cloud billing dashboard
Typical test run: 50,000-200,000 tokens depending on test complexity and duration.

Troubleshooting

Possible causes:
  • Test prompt too complex
  • App hangs or freezes
  • Network delays
Solutions:
  • Simplify test into smaller steps
  • Check app for blocking operations
  • Use mock data to avoid network calls
Possible causes:
  • Element not visible on screen
  • Accessibility labels missing
  • Dynamic content loaded slowly
Solutions:
  • Add accessibility labels to UI elements
  • Wait for content to load in test prompt
  • Use more specific descriptions
Possible causes:
  • Test prompt not specific enough
  • AI missed subtle issues
  • Visual-only problems
Solutions:
  • Add explicit verification steps
  • Review screenshots manually
  • Refine test prompt with more detail

Next Steps

App Previews

Learn more about creating shareable previews

Build Insights

Track QA test trends and success rates