Overview & Concepts

This guide explains the core concepts behind QAID and how it works.

How QAID Works

QAID follows a three-stage approach to test generation:

Stage 1: Discovery

QAID crawls your website to discover pages and understand the site structure. It handles:

Navigation menus (including dropdown menus)
Single Page Applications (SPAs)
Authenticated areas (with credentials you provide)
Dynamic content loaded via JavaScript

Stage 2: Analysis

For each discovered page, QAID:

Extracts interactive elements (buttons, links, forms, inputs)
Detects iframes and embedded content
Identifies bugs and quality issues (broken links, accessibility problems)
Analyzes page purpose and criticality using AI

Stage 3: Generation & Execution

Based on the analysis, QAID generates and executes:

Automated Tests (primary) - AI-powered user journey tests with self-healing selectors
Manual Tests - Structured manual test cases with steps and expected results
Visual Baselines - Screenshots for regression testing

Core Concepts

Projects

A project represents a single website or web application you want to test. Each project has:

A base URL (e.g., https://example.com)
Optional authentication configuration
Its own set of discovered pages, elements, and tests
Project-specific settings

Elements

Elements are interactive components on a page:

Types: link, button, input, select, checkbox, textarea, etc.
Importance: primary, secondary, or tertiary
Selectors: CSS/XPath/role-based selectors for targeting

Scenarios (Automated Tests)

Scenarios are the primary test type in QAID. They represent user journeys:

"User should be able to submit the contact form"
"User should navigate from homepage to pricing"
"User should complete checkout flow"

QAID's AI analyzes each page to understand its purpose and generates tests using selectors from elements actually discovered on your site. Tests self-heal automatically when selectors change.

Test Plans

Test plans organize tests into logical groups:

Folders for categorization
Priority-based organization
AI-powered auto-organization

Test Runs

Test runs execute a test plan and track results:

Parallel test execution
Real-time progress tracking
Pass/fail reporting with AI failure classification

Quality Score

A 0-100 composite score measuring your project's test health, combining:

Test Pass Rate — percentage of tests passing
Stability — consistency of results (detects flaky tests)
Coverage — element and flow coverage
Bug Freedom — open bug count and severity
Visual Consistency — visual regression pass rate

Test Pass Rate carries the most weight, with Stability and Coverage close behind.

The Workflow

QAID follows a workflow where steps unlock progressively:

Step	Requires	What It Does
Overview	Nothing	Dashboard with quality score and project health
Discovery	Nothing	Discover pages, extract elements, detect bugs
Diff Analysis	2+ element snapshots	Compare element changes between crawls
Automated Tests	Elements extracted	Create AI-powered scenario tests (primary)
Manual Tests	Pages discovered	Generate manual test documentation
Visual Regression	Pages discovered	Capture and compare screenshots
API Testing	Pages discovered	Test API endpoints with collections and assertions
Test Plan	Tests generated	Organize all tests into plans
Test Run	Tests generated	Execute test plans
Analysis	Tests generated	Review test health trends and patterns
Actions	Tests generated	Get AI recommendations for improvement

AI Integration

QAID uses Claude AI (by Anthropic) for intelligent test generation and analysis.

Zero-Hallucination Tests

All test selectors come from actual elements discovered on your site. QAID never invents or guesses selectors, making tests reliable and maintainable.

When an element can't be found with its original selector, QAID's self-healing engine automatically tries to re-locate it using live page analysis and content matching. Tests continue passing through routine UI updates — class name shifts, layout moves, A/B variants — without manual intervention.

AI Failure Classification

After test runs, AI can automatically classify failures as:

Real bugs in the application
Expected failures (site changes)
Flaky test issues

Page Status Lifecycle

Pages flow through these statuses:

New URL Found -> discovered
     |
     v
Successfully Crawled -> crawled
     |
     |-> Not in latest crawl -> removed (auto-restores if found again)
     |
     |-> User excludes -> blacklisted

Status changes cascade to related scenarios, tests, and manual test cases.

Production-Safe Mode

When enabled, QAID generates tests that only verify visibility and presence:

No form submissions
No button clicks that modify data
Safe to run against production environments