Skip to main contentGetting Started
Start your integration with simple test cases: straightforward flows and basic actions. As you become more familiar with the platform, you can gradually introduce more complex scenarios.
Writing Test Instructions
Break Down Steps
Write instruction step by step rather than combining multiple actions into one instruction. This improves reliability and makes debugging easier.
Keep Tests Focused
Avoid writing overly long and complex tests. Shorter, focused tests are more reliable and easier to maintain. Complex tests increase the probability of errors.
Be Specific When Needed
- Simple screens (e.g., onboarding with a couple of buttons): You can use general instructions like “go through onboarding”
- Complex screens with multiple flows: Describe the specific flow you need. The more context you provide, the better the agent understands your intent.
Application Context
Add application context in the app settings to help the agent understand your app better:
- Description of your app
- Overview of main screens
- Core functionality
- Any domain-specific terminology
This context significantly improves the agent’s decision-making.
Reusable Flows
If you have repeating flows across multiple tests, use reusable flows to avoid duplication and save time.
Deep Links for Efficiency
To avoid repeating onboarding or authentication steps in every test, implement deep links that set up the required state:
- Skip onboarding
- Auto-login users
- Navigate directly to specific screens
This saves time and makes tests more reliable.
Action Reference
The agent supports the following actions:
Fully Supported
- Taps: Always reference a specific UI element. Do NOT write “tap in the center of screen” – always bind to a concrete element
- Deep links: Open specific app states
- Scroll: Vertical and horizontal scrolling
- Form filling: Text input in fields
- App restart: Relaunch the application
In Development
- Swipes: Gesture-based navigation
- Drag and drop: Moving elements
- Long tap: Press and hold actions
Tips for Success
- Start simple: Begin with basic happy-path scenarios
- Add context: Provide app descriptions to improve agent understanding
- Use deep links: Skip repetitive setup steps
- Keep tests short: Focus on one feature or flow per test
- Be explicit: When in doubt, provide more detail rather than less
- Review failures: Check video recordings to understand what went wrong and refine instructions
Following these guidelines will help you create reliable, maintainable test suites that provide consistent results.