A reusable regression library
Save tests, organize suites and run checks sequentially. Use CSV data and environment profiles, inspect each outcome and rerun selected failures.
Browser testing · Chrome extension
QAgentExt
Record a browser flow. Replay it as a test. Turn the result into your next automation script, all from Chrome's side panel.
From a manual check to a repeatable test
Keep the flow. Check the outcome.
Give your everyday browser checks a repeatable path, an expected result and evidence you can review when something changes.
- 01 / Record
Capture the interaction
Record a short flow, stop and review the steps. Add an assertion such as confirmation text appearing after a form is saved.
- 02 / Replay
Verify what happened
Run the steps on your target page, inspect the result and save useful checks to your library. A click alone does not prove success.
- 03 / Export
Move toward automation
Copy Playwright or Cypress starter code. Review the generated script, install its dependencies and adapt it to your project.
75-second walkthrough
Watch a real test take shape
Record a form, replay it, export the code, investigate a failed assertion and run two saved checks. The copied Playwright script was also executed independently.
2.1 release preview. The store listing shows the version currently available to install.
Read the demo transcript
- Record: Start with a real flow. In Q Agent Extension, press Record, update the release name, and save the draft. Stop recording, then add the expected result: Draft saved. Review the captured steps before replaying.
- Replay: Replay on a reset page. Q Agent returns to the starting address, enters the recorded value, and clicks Save draft. The assertion checks the actual outcome. Save the passing test to your local library.
- Export: Choose a framework and copy the script. This is the actual Playwright export, with navigation, recorded selectors, and the expected outcome. We also run this copied script independently. Review generated code before using it in your own automation project.
- Debug: A passing click is not enough. Here we deliberately expect an approval message that the sample application does not show. The test fails. Review the assertion and the evidence in the side panel before changing the test or the application.
- Run a suite: Run saved tests together. This local example replays the recorded draft flow and checks a sample API. Review the outcome of each case and the overall quality gate. These are real execution results, not a mock interface.
A testing workspace beside your application
Evidence for the failed step
Review the expected outcome, available screenshots, console messages and network diagnostics. Export reports and reproduction evidence for your team.
Live contrast inspection
Hover over an element for its contrast ratio and text thresholds. Feedback updates as styles change; click to keep the detailed report.
Screen-reader semantics
Inspect roles, names, descriptions and states using the pointer or keyboard. Inspection modes have active states and Escape cleanup.
Page and API checks
Combine supported plain-language commands and assertions with API requests, response checks, visual comparisons and accessibility diagnostics.
Continue in CI
The separately installed companion CLI runs supported regression plans in headless Chromium and exports JSON or JUnit results for your pipeline.
Sequential tests share browser state. Generated scripts need review, and export support varies by action and framework. Automated accessibility and security checks provide diagnostic findings; manual review is still needed.
Follow the result from draft to suite
Screenshots from the 2.1 preview. Select an image to inspect the full view.
Before your first test
Do I need an account or an AI service?
Core testing runs in Chrome without a QAgentExt account or an external AI service. Your application may still require its own sign-in and network connection.
Can I use the generated code in my automation project?
Yes, after review and setup. Export coverage varies by action and framework. The 2.1 preview adds an execution guard for unsupported Playwright and Cypress steps; those steps need manual implementation. The demo's copied Playwright flow was run independently.
Does QAgentExt send data over the internet?
Core processing and saved tests are local. Page actions and API tests contact the applications you choose. Optional monitor webhooks send run summaries, and the extension panel requests Google Fonts. Reports can contain application data. See the privacy policy for details.
Does inspection replace a real screen reader?
No. Semantic inspection helps you examine roles, accessible names and states. It does not reproduce a complete NVDA, JAWS or VoiceOver session or establish accessibility conformance.
Can I test an internal application?
Use Chrome on an application you can access and are authorized to test, including permitted internal environments. The application still needs its own connectivity and permissions. Chrome-protected pages and some embedded content cannot be inspected.
How do I start, or report a problem?
Open a safe test page and the QAgentExt side panel, record a short flow, add an expected result and replay it with dummy data. For help, email shubix.connect@gmail.com with the extension version and reproduction steps. Review screenshots and exported files for sensitive data before sharing.
Start with one useful check
Choose a flow you repeat often, define its expected outcome and keep it in your test library.