# Qwarm

> Qwarm is an AI-powered QA and UX testing platform. Write test flows in plain English, AI agents execute them in real browsers, and you get visual reports with actionable insights.

Qwarm helps teams ship bug-free software without writing test scripts. Describe what a user would do in natural language, and AI agents handle element resolution, browser automation, screenshots, and analysis.

## Core Concepts

- [Landing Page](https://qwarm.ai): Product overview, interactive demo, pricing, and FAQ.
- [Sign Up](https://qwarm.ai/signup): Create an account. No credit card required for the Free plan.
- [Log In](https://qwarm.ai/login): Access your dashboard.
- [Dashboard](https://qwarm.ai/dashboard): Manage all your projects from one place.

## Features

### Projects
A project represents one website or app you want to test (e.g. `https://myapp.com`). Each project has its own flows, variables, runs, and reports. On creation, Qwarm automatically health-checks the URL and fetches metadata (title, favicon) to populate the project card.

### Variable Vault
A secure, per-project key-value store for credentials and config. Sensitive values (passwords, API keys) are encrypted at rest with AES-256-GCM and masked in the UI. Reference variables in test flows using `{{VARIABLE_NAME}}` syntax — values are injected at runtime and never exposed in reports or logs.

### Flow Builder
Write test steps in plain English — no selectors, no scripts. Example:
1. Go to {{BASE_URL}}/login
2. Type {{TEST_USER_EMAIL}} into the email field
3. Type {{TEST_USER_PASS}} into the password field
4. Click "Sign In"
5. Verify the page says "Dashboard"

### Execution Engine
Each test run launches a real Chromium browser via Playwright. The AI reads the page's accessibility tree for semantic element resolution, with vision-based fallback using screenshots. Every step produces a screenshot and status event streamed in real time.

### Live Run View
Watch test runs execute in real time via server-sent events. See step-by-step progress, screenshots, and AI decisions as they happen.

### Visual Reports
After a run completes, get a visual report with comic-strip step replay, pass/fail status per step, AI-generated insights about what went wrong and why, and full technical logs.

### Swarm Mode
The killer feature. Run the same test flow through multiple AI personas simultaneously — each with different behaviors, expectations, and perspectives. Example personas:
- "Impatient Shopper" — races through steps, notices slow loading
- "First-time User" — gets confused by unclear navigation
- "Accessibility Auditor" — checks screen reader compatibility
- "Mobile-first User" — tests touch interactions and responsive layout
Swarm Mode surfaces bugs and UX issues a single test perspective would miss.

### Run History & Regression Detection
Browse all past runs with filtering by status, date, and flow. Compare any two runs side-by-side to spot regressions — see which steps changed status and how screenshots differ across time.

### Scheduled & Recurring Runs
Set up test flows to run on a schedule (daily, weekly, monthly) or at a specific time. Aligned with your deploy cycle for continuous quality assurance.

### Discovery Mode
Point Qwarm at your site's URL and AI crawls the pages to automatically suggest test flows based on the interactions and structure it finds. Accept suggestions to create flows instantly.

## How It Works
1. **Create a project** — paste your site URL, Qwarm auto-detects metadata.
2. **Write a flow** — describe test steps in plain English.
3. **Run it** — AI agents execute steps in a real browser, taking screenshots at every step.
4. **Review the report** — visual step replay, AI insights, pass/fail breakdown.
5. **Scale with Swarm** — launch multiple personas against the same flow to find edge-case bugs.

## Technology
- **Browser automation**: Playwright (real Chromium)
- **Element resolution**: Hybrid — accessibility tree (semantic) + vision model (screenshot fallback)
- **AI model**: Model-agnostic, currently using Anthropic Claude
- **Real-time**: Server-Sent Events for live run streaming
- **Security**: AES-256-GCM encryption for secrets, RLS-scoped database access

## Pricing

### Free — $0/month
- 1 project
- 3 runs per month
- QA mode only
- Community support
- 7-day report retention

### Pro — $29/month
- 5 projects
- 50 runs per month
- Swarm Mode (multi-persona testing)
- Scheduled & recurring runs
- Discovery Mode (5/month)
- 90-day report retention
- Email support
- 14-day free trial
- Run Packs available for extra runs

### Team — $79/month
- Unlimited projects
- 200 runs per month
- Everything in Pro
- Discovery Mode (15/month)
- API access (coming soon)
- Priority support
- Unlimited report retention
- 14-day free trial
- Run Packs available for extra runs

### Run Packs (one-time add-ons for Pro & Team)
- 25 runs — $9
- 75 runs — $19
- 200 runs — $39

## FAQ

- **Do I need to write code?** No. Qwarm is designed for teams that want to test without writing Selenium scripts or Cypress tests.
- **Does it test in real browsers?** Yes. Every run launches real Chromium via Playwright — real DOM, real pages, real screenshots.
- **What is Swarm Mode?** It runs the same test flow through multiple AI personas simultaneously, each with different behaviors and perspectives, surfacing bugs a single tester would miss.
- **Can I integrate with CI/CD?** Scheduled runs align with deploy cycles. API access for direct CI/CD integration will be available on the Team plan soon.
- **Is there a free plan?** Yes. 1 project, 3 runs/month, no credit card required.
- **What are Run Packs?** One-time purchases that add extra runs to your current billing period. Available for Pro and Team plans starting at $9 for 25 runs.

## Links
- Website: https://qwarm.ai
- Sign Up: https://qwarm.ai/signup
- Log In: https://qwarm.ai/login
- Dashboard: https://qwarm.ai/dashboard