Test Case Maintenance Agent

Keep your test repository aligned with changing requirements using AI that updates existing test cases and fills coverage gaps.

The Test Case Maintenance Agent updates the test cases you already have. You share your latest requirements. It reads them, scans every test case in your project, and finds the ones that no longer match your new spec. Then it drafts the updates and fills the coverage gaps. You review every edit before it lands.

Agentic capabilities

• Requirement-aware updates: The agent compares your new specs against the repository and revises the tests that drifted.
• Coverage-gap detection: The agent spots requirements that have no test and drafts new cases to cover them.
• Reasoned suggestions: Each change arrives with a plain-language reason and a side-by-side comparison.
• Iterative refinement: You steer the output with follow-up prompts, without starting over.
• Background runs: Generation runs in the background and stays tied to your session, so you can keep working in parallel.

The problem it solves

Test cases drift the moment a requirement changes. The repository keeps describing the old behavior while the product moves on. Someone has to find every affected test, rewrite it, and add coverage for the parts of the flow that are now missing. That work is slow. Teams under deadline pressure often skip it.

Stale test cases cause real damage:

• Tests fail for the wrong reason, because they assert behavior the product no longer has.
• Real regressions hide inside the noise of those false failures.
• New requirements ship with no test coverage at all.
• Testers waste review cycles reconciling old tests against the current spec.

The Test Case Maintenance Agent closes this gap. When your requirements change, it compares them against the test cases already sitting in your project. It then updates whatever no longer matches the new spec. New coverage gaps get filled too. But you review every suggestion before a single test changes.

The maintenance workflow

You run the agent on demand, whenever your requirements change. You share the latest spec, scope the folders it can touch, and start the run. The agent then follows four steps.

1. Share your latest requirements: Upload a requirement file, such as a PDF or DOCX, or link a Jira issue, Figma frame, or Confluence page. You can also paste requirements or issue links directly.
2. The agent scans your repository: It compares your new requirements against your current test cases to find which test cases no longer match.
3. The agent suggests updates and fills gaps: It flags the existing tests to update. It also drafts the new tests you are missing.
4. You accept the changes: Accepted updates apply to your repository automatically. Nothing changes until you accept it.

You stay in control of scope. The agent updates only the test cases inside the folders you select. It changes nothing until you accept it. Your requirements are never used to train BrowserStack AI models.

Prerequisites

Make sure you have:

• A project on the Team Ultimate plan.
• A set of existing test cases for the agent to update.
• Your latest requirements, ready as a file or as a link from any supported source.

Requirements you can share

The agent works from the requirements you give it. Share one source or several. Combine files with links when a change spans more than one place.

When you link a tool, the agent opens a short dialog so you can pick the exact item. A tool you have not connected yet shows a connect screen first. Connect it, then pick the item you need. Expand a source for the steps.

Link a Jira issue

• Click Add Requirements and select Jira.
• Select your Host, then the Project that holds the issue.
• Select one or more Issues.
• Click Link to pull the context into the run.

Link an Azure work item

• Click Add Requirements and select Azure.
• Select your Organization, then the Project.
• Select one or more Work Items.
• Click Link to pull the context into the run.

Link a Confluence page

• Click Add Requirements and select Confluence.
• Select your Host, then the Space.
• Select the Page that documents the change.
• Click Link to pull the content into the run.

Link a Figma design

• Click Add Requirements and select Figma.
• Enter the Figma link, then click Import.
• Select the frames you want, then click Save Selection.

File and attachment limits

The maintenance agent reads the same inputs as the rest of BrowserStack’s AI test case generation, so the same limits apply.

Each file you upload must stay within these limits:

Each generation also caps how many items you can attach from a linked tool:

If a file is larger than its limit, the upload fails. Split or compress it, then try again.

Run the maintenance agent

The whole run happens inside the Generate test cases with AI panel: share the changed spec, turn on maintenance mode, generate, then review and accept. You can refine the suggestions as many times as you need before you accept them.

Test Case Maintenance Agent vs. Test Case Generator Agent

Both agents turn requirements into test cases. But they solve different problems.

Use the Test Case Maintenance Agent when you already have a repository and a requirement changed. Use the Test Case Generator Agent when you are building coverage from scratch.

Best practices

• Scope each run to the folders the change affects. A tighter scope means a faster, cleaner review.
• Share the most specific requirement you have. A precise spec produces precise updates.
• Read the AI Reasoning note before you accept a scenario. It tells you why the agent thinks each test changed.
• Treat semantic rewrites with extra care. Compare the original and updated steps so an edit does not quietly change what a test verifies.
• Re-run the agent whenever the same area changes again. The repository stays current with far less manual effort.

Known limitations

• The agent runs on demand. It does not watch your linked requirements or update tests on its own.
• The agent changes only the test cases inside the Target Folders you select. Test cases in other folders stay untouched.
• Every change waits for your review. The agent does not edit a test case until you accept its suggestion.
• AI suggestions can be wrong. Compare each one against your requirement before you accept it.

Next steps

• Generate new test cases with the Test Case Generator Agent: Build coverage from scratch using prompts, requirement files, or linked issues.
• Remove duplicates with the Test Deduplication Agent: Find and merge exact and semantic duplicates to keep your repository lean.
• Explore all BrowserStack AI Agents: Learn about every agent available in Test Management.