Workflows

Create custom workflow templates to provide reusable context for common testing tasks and avoid repeating basic inputs.

Workflows in Test Companion allow you to define a series of steps that guide the AI through a repeatable process. Instead of repeatedly typing the same multi-step instructions into the chat for familiar tasks, you can write those instructions once in a workflow file and invoke it with a single command.

Test Companion features two types of workflows:

• Built-in workflows: Which handle essential testing tasks right out of the box.
• Custom workflow templates: Which you can create for your own repetitive processes.

Built-in workflows

Test Companion comes with pre-built workflows designed for the most common testing tasks. These workflows are fully integrated into the system and optimized for quality output. You do not need to create or configure them—they are available as soon as you install the extension.

The following built-in workflows are available:

You can activate the built-in workflows by selecting them from the quick-action buttons on the Test Companion home screen or by describing the task in chat, allowing the AI to route it to the appropriate workflow.

Custom workflow templates

Custom workflows are Markdown files (with a .md, .txt extension, or no extension) that contain step-by-step instructions for processes specific to your team or project. They are ideal for tasks that the built-in workflows do not cover or for wrapping a built-in workflow with project-specific context, ensuring you do not have to retype the information every session.

When to create a custom workflow

Consider creating a custom workflow when you find yourself doing any of the following repeatedly:

• Providing the same URLs, credentials, or environment details before every task
• Following a multi-step process that requires specific commands in a specific order
• Onboarding new team members who need a guided, standardized process
• Combining multiple actions into a single trigger (for example: pulling the latest code → running tests → summarizing results → creating a Jira ticket)

How to invoke a workflow

There are two ways to invoke a custom workflow:

• From the chat: Type / followed by the workflow’s file name in the Test Companion chat (for example, /regression-run.md).
• From the UI: Open the Manage Test Companion Rules & Workflows panel, switch to the Workflows tab, and click the toggle of the workflow you want to run.

Test Companion reads the file and follows the instructions step by step. Depending on your interaction mode, the AI either works through the steps autonomously (Default mode) or pauses at each step for your confirmation (Guided mode).

Global workflows vs. Workspace workflows

Workflows can be stored at two different scopes:

• Global workflows
• Workspace workflows

Global workflows

Global workflows are available in every project you open. Use these for processes that are not tied to a specific codebase, such as general code review checklists, PR preparation steps, or personal productivity routines.

Workspace workflows

Workspace workflows are stored within a specific project directory and are available only when that project is open. Use these for project-specific processes, such as deploying that particular service, running its unique test suite, or setting up its development environment.

Create a custom workflow

From the extension panel, follow these steps to create a new custom workflow:

1. Click the Manage Test Companion Rules & Workflows button (see annotation 1).

   

2. Switch to the Workflows tab (see annotation 2).
3. Under either Global workflows or Workspace workflows, type a descriptive file name in the input field.

4. Click the + button.

   

5. A new file will open in your editor. Write your workflow instructions and save the file.

Writing a workflow file

Workflow files are written in plain language with numbered or sequential steps. The AI reads them from top to bottom and follows each instruction. You can include:

• Descriptive instructions: Clearly communicate to the AI what to do using natural language.
• Specific commands: Include terminal commands that you want the AI to execute.
• Decision points: Prompt the AI to present options and wait for your input.
• References to files: Direct the AI to specific files that it should read from or modify.
• Placeholders: Use placeholder text (such as <URL> or <PR_NUMBER>) that you fill in when invoking the workflow.

Example workflows

Example 1: Supplying context for automation code generation

This is the simplest and most common type of custom workflow. It wraps a built-in workflow (Write Automation Code) with your project-specific context so you do not have to retype it each time.

File Name: write-tests.md

Write automation tests

Context

• Application URL: https://staging.myapp.com
• Login Credentials: testuser@example.com / examplePassword!
• Framework: Selenium WebDriver with Python + Pytest
• Browser: Chrome (headless)

Instructions

Write automation test scripts for the test cases I provide. Use the credentials above for any tests that require authentication. Place new test files in the tests/ directory following our existing naming convention (test_.py).

Run the tests after writing them to verify they pass. If any tests fail, fix the issues and re-run until all tests pass.

How to use it: Type /write-tests.md in the chat, then paste or describe the test cases you want to be automated. The AI already has your URL, credentials, and framework details, allowing you to skip the setup and go straight to the work.

Example 2: Regression test suite execution and reporting

File Name: regression-run.md

Regression test suite run

Step 1: Pull latest code

Run these commands to ensure we are testing against the latest version:

• git checkout main
• git pull origin main

Step 2: Install dependencies

Run: pip install -r requirements.txt

Step 3: Execute the regression suite

Run: pytest tests/regression/ –html=reports/regression-report.html –self-contained-html -v

Step 4: Analyze results

Read the test output and summarize:

• Total tests run
• Tests passed
• Tests failed (list each failure with a one-line description of the cause)
• Tests skipped

Step 5: Suggest fixes

For each failed test, analyze the failure and suggest a fix. Ask me whether I want to apply the fixes before making any changes.

Example 3: New feature test scaffolding

When your team starts work on a new feature, this workflow creates the entire test scaffolding in one step.

File Name: new-feature.md

New feature test scaffolding

I will describe a new feature. For that feature, please do the following:

Step 1: Create the Page Object

Create a new Page Object class in tests/pages/ following our existing patterns. Include locators for all interactive elements I describe and action methods for common user interactions.

Step 2: Generate test cases

Write at least 10 test cases covering:

• Happy path (primary user flow)
• Negative cases (invalid input, missing required fields)
• Boundary cases (min/max values, empty strings, special characters)
• Accessibility (keyboard navigation, screen reader compatibility)

Step 3: Create the test file

Create a test file in tests/ that automates all the test cases from Step 2 using our Selenium + Pytest setup.

Step 4: Verify

Run the new test file and report which tests pass and which need adjustments (some may fail if the feature is not yet fully implemented, which is expected).

How to Use It: Type /new-feature.md in the chat, then describe the feature. The AI will create the page object, write the test cases, generate the automation code, and run it—all in one flow.

Example 4: Cross-browser test run on BrowserStack

File Name: cross-browser.md

Cross-browser test run on BrowserStack

Configuration

• BrowserStack username: (use the value from Settings)
• BrowserStack access key: (use the value from Settings)

Target Environments

Run the specified tests on each of these configurations:

1. Chrome (latest) — Windows 11
2. Firefox (latest) — Windows 11
3. Safari (latest) — macOS Sonoma
4. Edge (latest) — Windows 11
5. Chrome (latest) — Samsung Galaxy S23 (Android 14)
6. Safari — iPhone 15 (iOS 17)

Instructions

1. Take the test file or test cases I provide.
2. Generate BrowserStack-compatible automation code using W3C capabilities.
3. Run the tests against each target environment.
4. Collect and summarize the results in a table showing pass/fail status per browser and any browser-specific failures.
5. For any browser-specific failures, suggest a fix.

Example 5: Daily test health check

File Name: health-check.md

Daily test health check

Step 1: Run the Smoke Test Suite

Run: pytest tests/smoke/ -v –tb=short

Step 2: Check for New Flaky Tests

Run: pytest tests/ –count=3 -v (This runs each test 3 times to detect intermittent failures.)

Step 3: Report

Create a summary with:

• Smoke suite status (all pass / X failures)
• Newly detected flaky tests (any test that passed some runs but failed others)
• Recommended priority fixes

Step 4: Ask what to do next

Based on the results, ask me whether I want to:

• Fix the flaky tests now
• Create tickets for the failures (list what each ticket should contain)
• Skip and continue with other work

Best Practices for writing workflows

• Start simple: Your first workflow should be brief—aim for 5 to 10 lines of context for a built-in workflow (like Example 1). Once that functions well, you can create more complex multi-step workflows.

• Use descriptive file names: The file name is what you type after the / to invoke the workflow. Choose names like regression-run.md or pr-review.md that are easy to remember. Avoid generic names like workflow1.md.

• Be explicit about order: Number your steps. The AI follows them sequentially, and numbered steps clarify which actions depend on previous results.

• Include decision points: For workflows where the outcome of one step determines the next action, instruct the AI to present options and wait for your input. This keeps you in control at critical junctions.

• Test your workflows: After creating a new workflow, run it once and observe the results. Adjust the instructions if the AI misinterprets a step or misses context.

• Version control your workspace workflows: Commit Workspace Workflows to your repository so the entire team can benefit from them. Changes to workflows should go through code review, just like modifications to test code.

Manage workflows

View and edit workflows

Click the Manage Test Companion Rules & Workflows button and switch to the Workflows tab to see all your active workflows. Click any workflow file name to open it in the editor. Changes are applied the next time you invoke the workflow.

Disable or delete a workflow

To disable a workflow, click the toggle switch (see annotation 1) next to the workflow name in the Workflows panel. A disabled workflow remains in your list but cannot be invoked. You can re-enable it at any time by clicking the toggle again.

To permanently delete a workflow, click the delete icon (see annotation 2) next to the workflow name. You will be prompted to confirm the deletion. Once deleted, the workflow file is removed and cannot be recovered.

Rules vs. Workflows

If you are unsure whether your instructions belong in a rule or a workflow, refer to this guide:

In many cases, you will use both together. Your rules ensure that any code the AI writes meets your standards while your workflows automate the process of producing that code for specific scenarios.