Web testing

Test web applications using Test Companion, from generating test cases to writing automation scripts and fixing failures, all without leaving your IDE.

Test Companion is an IDE extension that brings AI-powered testing into your development workflow. Describe what you want to test in plain language. Whether you provide a requirements document, a URL, a failing build, or a code diff, Test Companion figures out the best approach and handles the work.

Test Companion can:

• Generate test cases from a PRD, spec, user story, or live website.
• Validate your local code changes before you commit.
• Write automation scripts in any testing framework your project uses.
• Generate API tests from an endpoint or specification.
• Analyze and fix failing or flaky tests from your CI/CD builds.
• Scan your application for accessibility violations and generate code fixes.

How Test Companion works

When you type a prompt in the chat, Test Companion reads your message, evaluates the context you provide, and determines how to handle your request. You do not need to select a mode or choose an approach. Describe what you need, and Test Companion gets to work.

For independent subtasks you run repeatedly, you can create agents that operate as reusable playbooks with your project-specific context, credentials, and file paths. For more information, refer Agents documentation for details.

Chat interface

The Test Companion panel is in your IDE’s sidebar. It has two main areas: the Chat panel and the Test management panel.

The chat panel is where you interact with Test Companion. Type your prompt, attach context, and send the message. Test Companion streams its response, shows its action plan, and delivers results directly in the panel.

Quick actions

The home screen displays five quick action buttons for the most common tasks: Generate Test Cases, Write Automation Code, Fixing flaky or failing tests, Test the code changes (Dev Testing), and Fix accessibility issues. Clicking a quick action pre-fills the chat with a starter prompt for that task. You can edit the pre-filled message before sending, or type a free-form prompt of your own at any time.

Quick actions are suggestions, not selectors. Auto mode still routes the task based on the prompt and context, regardless of which quick action you started from.

Web and App toggle

A Web and App toggle sits below the chat input. Web is the default for browser-based testing on desktop sites. Switch to App when working on mobile applications. The capabilities available, the context options, and the connected devices change based on the selected mode.

Add context for better results

Test Companion produces better output when you give it relevant context. You can combine multiple context types in a single prompt.

To add context, click the + icon in the chat toolbar or type @ in the chat input. The available context options are:

The more specific your context, the more accurate the results. Attaching the exact file or referencing the exact Git commit gives Test Companion a focused scope to work with.

Capabilities

Each capability addresses a specific testing task. Describe what you need. Auto mode selects and runs the right one. The table below summarizes the five built-in capabilities.

Configure capability output

To set the format for generated test cases, click the gear icon (⚙) in the chat toolbar before sending your prompt. The Configure Test Cases panel opens with two format options:

• Default: Generates test cases in a standard tabular format with discrete fields for each step and expected result. This is the most common format for teams using BrowserStack Test Management.
• BDD: Generates test cases in Gherkin-style Given/When/Then syntax. Use this if your team follows behavior-driven development practices.

The configuration options shown depend on the capability auto mode selects for your prompt. See each capability’s child page for the full list of configuration options.

Customize Test Companion

Beyond the built-in capabilities, you can customize how Test Companion responds across every conversation, and you can define your own playbooks for repeated tasks.

• Use rules to enforce coding standards or other persistent guidance.
• Use agents to define reusable playbooks for independent subtasks and invoke them with a slash command.
• Use configuration and preferences to adjust interaction modes, approval controls, and other extension settings.

Next steps

• Generate test cases: Create test cases from requirements or by exploring a live site.
• Dev testing: Validate local code changes before you commit.
• Write automation code: Convert manual tests into automation scripts.
• Fix flaky or failing tests: Analyzes CI/CD failures and suggests code fixes for failing or intermittent tests. You can also start this workflow from the Failure Analysis panel in Test Reporting and Analytics.
• Fix accessibility violations: Scan for WCAG violations and apply code fixes.