BrowserQA: AI-assisted QA, Playwright runs, and a Chrome side panel that talks to your app

#ai #testing #playwright #test

I work on BrowserQA: a QA workspace where you can generate and run automated checks, iterate with AI, and keep execution close to how real users use the browser.

This post is a short overview of what it is, how the BrowserQA Copilot Bridge Chrome extension fits in, and where to grab installers if you want to try local Playwright runs.

What BrowserQA tries to solve

Traditional QA workflows often bounce between spreadsheets, ticketing tools, and "run this script locally" instructions. BrowserQA centralizes projects, runs, and results around Playwright-style automation, with room for hosted execution and local execution when you need your own machine (auth, VPN, headed debugging).

At a high level (from how the platform is built):

AI-assisted test generation using Browser-Use style agents and tooling.
Playwright execution as a first-class path for reliability.
Optional cloud runs where the stack is configured for hosted browsers.
Local desktop runner so you can execute on your hardware and stream logs and outcomes back into the product.

If you want the engineering-oriented feature list, the internal platform README also calls out things like TestRail integration, manual recording flows, and selector heuristics that favor stable attributes such as data-test-id.

The Chrome extension: Copilot Bridge

The extension is published as BrowserQA Copilot Bridge (Manifest V3). Current build in the repo is v0.1.23.

What it is for

Pick UI elements on real pages and send structured context into BrowserQA.
Work from the side panel: project, picks, optional notes, and actions to automate (when a manual case is attached) or generate manual test cases from the page.
In-page Copilot: open the assistant on the active tab so you can chat while staying on the site you are testing.

How it connects

Sign-in uses Chrome identity.launchWebAuthFlow, or you can pair via the /extension/connect flow on the site (see the extension README in the project).
Copilot requests go to your Backend /api/copilot/chat (on production that is https://browserqa.io). For debugging, remember MV3 fetch happens in the service worker, so you will not see those calls in the website tab Network panel unless you inspect the worker.

Important constraint (worth saying up front)

The injected page script is configured for https://* pages (not plain http). Stick to HTTPS sites when using the picker and in-page Copilot. Chrome also blocks scripting on chrome://, extension store pages, and similar; the extension surfaces a hint when that happens.

Privacy

Public policy: https://browserqa.io/legal/privacy (includes extension-related disclosure). If you ever ship on the Chrome Web Store, keep your single-purpose description and permission justifications aligned with what you actually do (host permissions vs identity vs scripting).

Download

ZIP and local runner installers: https://browserqa.io/downloads Example extension artifact name pattern: browserqa-extension-0.1.23.zip (version bumps with releases).

Development tip: Load unpacked on the built dist/ folder after npm run build in webapp/extension.

Local runner (Playwright on your machine)

The Browser QA Local Runner (desktop app) is aimed at teams that want:

Local execution while keeping orchestration and results in BrowserQA.
Faster iteration on flows that depend on machine-specific login or environment.
Auto-connect behavior after sign-in so the worker stays available.
Headed vs headless switching from the desktop app.

On /downloads the installers are currently versioned around Local Agent v0.2.0 in the codebase: Apple Silicon DMG and Windows installer, plus FAQs that note Playwright/Chromium bootstrap on first run for recent versions.

Who this is for

If you are a QA engineer, SDET, or a developer who owns releases, this stack is meant to shorten the path from "I see a bug on this page" to "here is a reproducible automated check and a run history", with optional AI help for drafting cases and suggesting steps.

Try it

Open https://browserqa.io and create or open a workspace.
Grab builds from https://browserqa.io/downloads.
Install BrowserQA Copilot Bridge, sign in, open the side panel, and try Pick + Generate or Automate on an HTTPS page you control or trust.

Disclosure

I am affiliated with BrowserQA / this project. If something here drifts from the live product (pricing, roadmap, Chrome Web Store listing), trust the browserqa.io site and release notes first.

Feedback welcome: what would make in-browser Copilot + Playwright runs irresistible for your team?