FAQ
Getting Started
What is DebuggAI?
AI-powered browser testing platform. Write tests in plain English, run them against your localhost or staging environments. No test scripts, no brittle selectors, no complex setup.
How does AI test generation work?
- You describe what to test in natural language ("Test user can log in and see dashboard")
- AI analyzes your page, finds elements by understanding context rather than selectors
- Remote browser executes the test against your app via secure tunnel
- Results with screenshots appear in your IDE or dashboard
What frameworks are supported?
All web frameworks work: Next.js, React, Vue, Angular, Django, Rails, Flask, Node.js, static sites. If it runs on localhost, DebuggAI can test it.
Testing
How do I write good test descriptions?
Be specific about user, action, and expected outcome:
Good: "Test logged-in user can add item to cart and see cart count increase"
Bad: "Test the cart works"
Include user context, clear action, and success criteria. See Test Descriptions for patterns.
Can I test authenticated pages?
Yes. Configure login credentials in project settings:
- Set login URL and form selectors
- Store credentials as environment variables
- Tests authenticate automatically before running
Add /logout to exclude patterns to prevent session termination during tests.
How long do tests take to run?
- Tunnel setup: 5-15 seconds
- Test execution: 30-120 seconds depending on flow complexity
- Results delivery: Near-instant
What happens if a test fails?
You receive:
- Screenshot at the failure point
- Step-by-step execution log showing where it broke
- AI analysis of what went wrong
- Suggestions for test description improvements
Browser Agents
What are browser agents?
Autonomous AI systems that navigate and interact with your app without predefined scripts. They observe page state, reason about actions, and adapt dynamically to UI changes.
How are they different from E2E tests?
| E2E Tests | Browser Agents |
|---|---|
| Follow predefined descriptions | Make decisions in real-time |
| Best for critical, stable paths | Best for exploratory testing |
| Faster execution | More thorough exploration |
| Consistent, repeatable | Adaptive to changes |
Use E2E tests for CI/CD pipelines and critical paths. Use agents for discovering edge cases and testing after UI redesigns.
Do I need to train them?
Optional but recommended. Untrained agents use general knowledge. Training with example interactions improves accuracy for your app's specific patterns.
Provide examples of successful workflows, especially for:
- Complex forms and multi-step processes
- Custom UI components
- Edge cases you've encountered
Knowledge Graph
What is the knowledge graph?
A visual map of your application's pages, states, and navigation paths. The crawler discovers all reachable pages and builds an interactive graph showing:
- Every route and view
- Navigation transitions between pages
- Test coverage per page (green = tested, yellow = untested)
How does crawling work?
- Start from your entry URL
- Crawler clicks links and buttons to discover new pages
- Each unique state becomes a node in the graph
- Navigation paths become edges
- Graph updates continuously as your app evolves
Configure depth limits, exclude destructive actions (/logout, /delete/*), and set authentication for protected sections.
Is it safe to crawl my production site?
Yes, with precautions:
- Set crawl to read-only mode
- Exclude all destructive endpoints (
/delete,/admin/destroy) - Exclude logout URLs
- Use staging when possible
The crawler never submits forms unless explicitly configured.
Billing & Plans
What counts as a test run?
One test execution = one test run. Running the same test multiple times counts as multiple runs. Failed tests also count.
Can I upgrade/downgrade anytime?
Yes. Changes take effect immediately. Upgrading prorates billing; downgrading applies at next billing cycle.
Current pricing
| Plan | Test Runs | Price |
|---|---|---|
| Free | 100/month | $0 |
| Pro | 1,000/month | $20/month |
| Grow | 5,000/month | $40/month |
All plans include full feature access. Enterprise plans with custom limits available.
Security
How are my credentials stored?
- Login credentials are stored encrypted and never logged
- Use environment variables for sensitive data
- Credentials are only used during test execution
- No plaintext storage anywhere in the system
Is my code sent to external servers?
No. Only:
- DOM structure (what the browser sees)
- Screenshots of rendered pages
- Your test descriptions
Source code files never leave your machine. The secure tunnel only transmits what's needed to render and interact with your running app.
Troubleshooting
Tests failing unexpectedly?
- Ensure your app is fully loaded before testing
- Make test descriptions more specific
- Review error screenshots in results
- Break complex tests into smaller ones
Can't connect to localhost?
- Verify app is running:
curl localhost:3000 - Check port matches your settings
- Disable VPN if active
- Run "DebuggAI: Test Connection" from VS Code command palette
Need help?
- In-app chat in VS Code extension (
Cmd+L) - Web dashboard help center
- status.debugg.ai for system status