QA Team
Your AI-powered quality assurance team.
QA Team is a team of AI specialists that tests your application through intelligent exploration. Instead of writing brittle test code, it understands your app semantically and tests it like a human QA specialist would.
Price: $99 (one-time)
Quick Start
npx @aiorg/cli@latest init qa-team ~/my-qa
cd ~/my-qa
claude
> /setup
Requirements
- Claude Code — Get it here
- Node.js 18+ (for CLI)
Kits are powered by Claude Code. Install Claude Code first, then use our CLI to download kits.
Overview
You don't need to know testing syntax. QA Team explores and verifies — no brittle selectors, no flaky tests.
| Traditional E2E Tests | QA Team |
|---|---|
| Brittle selectors break when UI changes | AI understands UI semantically |
| Tests verify happy paths | Destructive testing finds real bugs |
| You maintain test code | Zero maintenance |
| Tests run in background | Visible browser — watch AI explore |
Elite Tester Mindset
QA Team doesn't test to prove things work. It tests to find ALL the ways to break things.
- How can I BREAK this?
- What happens if I do things in the WRONG ORDER?
- What if I click 10 times before the page loads?
- What's the weirdest thing I can input?
- What would an ANGRY user do?
Destructive Tours
| Tour | What It Does |
|---|---|
| Antisocial | SQL injection, XSS, IDOR, access /admin |
| Interruption | Refresh during submit, back button, double-click |
| Chaos | Slow network, race conditions, concurrent tabs |
| Speed Demon | Everything very fast, don't wait for loading |
| Edge Bomber | 10,000 chars, unicode, null bytes, negative IDs |
What's Included
UI Testing (Browser)
| Command | Description |
|---|---|
/setup | Connect to target project, configure Playwright |
/explore | Explore app structure and discover flows |
/pair-test | Interactive pair testing with AI as expert driver |
/run | Execute tests in target project |
/flows | View and manage discovered flows |
/evidence | View screenshots, logs, bug reports |
API Testing (Backend)
| Command | Description |
|---|---|
/api-discover | Scan codebase, find all API endpoints |
/api-explore | Test all discovered endpoints |
/api-test [endpoint] | Deep-test specific endpoint (30+ scenarios) |
/api-generate | Generate stable test files |
Test Generation (Paid)
| Command | Description |
|---|---|
/plan | Generate test plan from app structure |
/generate | Generate Playwright tests and create PR |
/heal | Analyze and fix broken tests |
/coverage | Analyze test coverage by flow |
The Team
QA Team has two specialists:
UI Tester
Discovers your app by navigating pages, clicking buttons, filling forms. Uses Playwright MCP for browser control. Visible by default — you watch AI explore in a real browser.
API Tester
Tests your backend directly without a browser. Auto-discovers endpoints from code, tests auth/validation/security. Generates stable tests that don't depend on specific data.
Workflow
UI Testing
/setup → /explore → /run (repeat) → /generate (optional)
- Setup — Connect to target project
- Explore — AI discovers app structure (visible browser!)
- Run — AI re-executes flows, catches regressions
- Generate — (optional) Create Playwright tests for stable flows
API Testing
/setup → /api-discover → /api-explore → /api-generate (optional)
- Discover — Find all API endpoints in codebase
- Explore — Test each endpoint (auth, validation, security)
- Deep-test — Focus on specific endpoints
- Generate — Create stable test files
Pair Testing
For apps with external systems AI can't access alone — payments, mobile apps, email, databases — use /pair-test for interactive sessions where AI and human work as partners.
How It Works
| AI Does | You Do |
|---|---|
| Scans codebase before asking anything | Verify database state |
| Interviews you about what to test | Check Stripe/payment dashboards |
| Drives the browser via Playwright | Test the mobile app |
| Suggests edge cases to try | Check email inbox |
| Builds and adapts the test plan | Run backend commands |
| Records all bugs and results | Provide visual confirmation |
Workflow
/pair-test
- Recon — AI scans your codebase, git history, and previous sessions
- Interview — Smart questions about scope, concerns, and access
- Plan — AI builds a custom test plan with edge cases
- Execute — AI drives browser, pauses when it needs you to verify something
- Report — Summary with bugs, coverage map, and what to test next
When to Use Which Mode
/explore | /pair-test | /coop | |
|---|---|---|---|
| AI access | Full (browser) | Browser + human help | None |
| Human role | None | Partner (verifies backend) | Executor |
| Best for | UI-only apps | Apps with payments, email, mobile | Mobile-only apps |
Multiple Sessions
Each pair-test session is saved separately. Future sessions read previous results to recommend untested areas, track regression after fixes, and build cumulative coverage over time.
Evidence Collection
QA Team collects evidence like a professional tester:
| Evidence | When Captured |
|---|---|
| Screenshots | On errors, key pages, form submissions |
| Console logs | Continuously during exploration |
| Server logs | When dev server runs |
| Network HAR | All HTTP requests/responses |
| Bug reports | When issues are found |
Bug Reports Include
- Screenshot
- Console errors with stack trace
- Server logs
- Steps to reproduce
- Environment info
Security Testing
API Tester automatically checks for:
- Auth bypass (accessing protected endpoints without token)
- SQL injection
- Missing validation (empty, null, too long inputs)
- Information disclosure in errors
- Missing rate limiting
Avoiding Flaky Tests
API Tester generates tests that don't depend on specific data:
// FLAKY (traditional)
expect(response.users.length).toBe(5);
// STABLE (QA Team)
expect(response.users).toBeInstanceOf(Array);
Related Kits
Works with any project:
- SaaS Dev Team — Test your SaaS app
- Any web application
FAQ
Q: Do I need to write tests?
No. QA Team explores your app and finds bugs without test code. Generated tests are optional for stable flows.
Q: Can I watch the AI test?
Yes! By default, QA Team opens a visible browser window. Use --headless for background execution.
Q: How is this different from Playwright tests?
Traditional tests verify happy paths with brittle selectors. QA Team thinks like an elite destructive tester — actively trying to break your app.
Q: Does this work with any project?
Yes. QA Team is a companion kit — works with any web application. Point it at localhost or staging URL.