Engineering & AI

ShardBrowser: choosing how to isolate browser test sessions

ShardBrowser manages separate browser profiles so QA teams can save and reopen signed-in sessions for different roles when testing a website.

Summary

  • Isolate sessions when one test involves both an administrator and an ordinary user.
  • ShardBrowser manages profiles and uses a modified Chromium build. Use the browser your users actually run as the baseline for product testing.
  • On staging, verify whether signing out or changing roles in one session affects the other.
ShardBrowser: choosing how to isolate browser test sessions

You need to test both sides of a workflow, such as an administrator approving an order while a customer tracks its status, but two tabs can share one login. When evaluating ShardBrowser, ask whether the sessions are independent and whether important results reproduce in the browser your customers use.

What does ShardBrowser offer for session isolation?

ShardBrowser manages profiles in a modified build of Chromium. Its author describes it as an anti-detect browser. Each profile has its own data and can retain its session between launches. That provides a model for keeping two long-lived sessions separate. With a modified browser build, confirm important results and failures in the target browser. ShardBrowser repository

When should you use separate profiles or BrowserContext?

  • Two roles in the same test. Use two Playwright browser contexts. Each has its own cookies and browser storage. Run the test on your own site or an authorized staging environment. Playwright isolation guide
  • Reopen a session for manual checks. Two profiles in a standard browser are often enough. Record the role and sample data for each profile so you can repeat the check.
  • Confirm a customer-reported bug. Repeat the steps in the specific browser and version you support. If the result only appears in the modified Chromium build, record the environment and reproduce it in the target browser before drawing a conclusion. Playwright browser support

Try one test and check for session carry-over

  1. Prepare two staging accounts. Give one an administrator role and the other a standard user role. Open each in its own context and use one sample order.
  2. Change the order in one session. Update its status as administrator, then reload the customer’s page to confirm the new status. Check that the standard user only sees features available to their role.
  3. Sign out as administrator. Reload the standard user’s page. If the app is designed for independent sessions, the user should stay signed in. If the result differs from what you expect, record the steps and check them again in the standard browser before attributing the problem to the app.

Preparing the AI agent

Provide the staging URL, roles, sample data, target browser version, and expected results. Ask the agent to check the workflow in two independent contexts and record which session failed and which step exposed the problem. Use test accounts in place of customer session data.

HubCom reviewed the documentation at commit b07f818 on November 20, 2026. HubCom has yet to install or run ShardBrowser.

Terms in this article (3)
ShardBrowser
A profile manager using a modified Chromium build.
BrowserContext
A Playwright environment that isolates session data for one or more pages.
Staging
A product environment for testing changes before public release.

Frequently asked questions

Are two tabs the same as two sessions?

Tabs in one profile usually share cookies for the same site. Use separate profiles or browser contexts for two independent roles.

Do I need ShardBrowser to test multiple roles?

Playwright supports multiple browser contexts in one test. Standard browser profiles are suitable for manual checks across launches.

What should I record if a bug appears in only one environment?

Record the browser, version, role, and reproduction steps. Repeat the test in the environment customers use before marking the bug as confirmed there.

Sources

Talk with HubCom ↗