kumQAt

Requirements to browser runs to bug reports

Describe what to test in plain English. Kumqat generates AI-written test cases, runs them in a real browser with the Browser Use Cloud agent, and delivers structured bug reports automatically.

Start Testing NowChat Interface

AI Test Generation

Describe a requirement in plain English. Gemini writes structured test cases with steps, expected outcomes, and failure signals — no test framework expertise required.

Real Browser Execution

Browser Use Cloud opens a real Chromium browser, navigates your site, clicks buttons, fills forms, and follows multi-step flows exactly like a human QA engineer.

Structured Bug Reports

Every test produces a detailed card: status, severity, confidence, expected vs actual, suspected root cause, business impact, screenshots, and video recordings.

Smart Result Classification

Outcomes are classified as pass, fail, blocked, or flaky. Blocked means a CAPTCHA or login wall stopped the agent. Flaky flags intermittent timing issues automatically.

Why Choose Kumqat?

< 5 min

Time to first report

From URL to structured bug report

4

Result types

Pass, fail, blocked, flaky

AI

Validated results

Gemini reads every agent trace

Kumqat is an AI-powered browser QA platform that turns product requirements into executable browser test cases, runs them against your live site with a real browser agent, and delivers structured, actionable bug reports in minutes.

Get Started Now

Frequently asked questions

Understanding your QA results

What does "pass / fail / blocked / flaky" mean?
  • pass: The agent completed all steps and found no issues matching the failure signals.
  • fail: One or more expected outcomes were not met, or an error was detected.
  • blocked: The agent could not proceed. Usually a login wall, CAPTCHA, or bot-detection challenge prevented access.
  • flaky: The result was inconsistent, sometimes passing and sometimes failing. Usually a timing or race condition.
What does severity (low / medium / high) mean?
  • high: Likely blocks users completely from using a feature.
  • medium: Degrades the user experience but does not fully prevent use.
  • low: Minor or cosmetic issue with minimal user impact.
What does confidence % mean?

When Gemini validates results, confidence reflects how certain it is of the classification (e.g. 90% means very confident it's a real failure). When the AI key is unavailable and the heuristic fallback is used, confidence is lower (50-65%) because the classification is rule-based, not AI-validated.

What do P0 / P1 / P2 test priorities mean?
  • P0: Critical user flow; usually release-blocking if it fails.
  • P1: Important functionality; high impact but typically not full release blockers.
  • P2: Lower-priority coverage, edge cases, or polish checks.
  • In the current version, these priorities are shown as triage labels in the UI and do not automatically change execution order.
What does the agent actually do?

Each test case runs in two stages. First, Playwright takes a screenshot and checks the HTTP status as a quick baseline. Second, the Browser Use Cloud agent opens a real browser, follows the test steps, navigates subpages, clicks buttons, fills forms, and reports what it found. The agent trace in each result card shows every step the agent took.

Why is a test marked "blocked"?

Sites like LinkedIn, Google, or university portals often detect automated browsers and show a CAPTCHA or login wall. The agent stops and marks the case blocked rather than guessing. You can try re-running with credentials provided in the test requirement, or the Browser Use Cloud agent may bypass bot detection better than a plain headless browser.

What is "Requirement-focused check", "Smoke test", etc.?

When you click Generate test cases, Gemini reads your requirement and creates named test cases. "Reach application" checks the page loads. "Primary navigation sanity" clicks your main nav. "Requirement-focused check" directly exercises the specific feature you described. "Form or input resilience" tests form validation. "Mobile viewport sanity" checks the layout on a narrow screen.