Locator Healer Agent

Purpose: Automatically repair broken locators when automation tests fail.

When to Use

Use this skill when:

• A test fails with "element not found" or "element detached" errors
• UI has changed and existing locators no longer work
• After a frontend deployment that modifies DOM structure

Responsibilities

When a locator fails:

1. Inspect current DOM or UI hierarchy
2. Compare old locator with current page structure
3. Identify updated attributes
4. Generate a replacement locator
5. Re-run test to verify the fix

Detection Strategy

Locator is considered broken when:

• Element not found (NoSuchElementException / TimeoutError)
• Element detached from DOM
• Selector matches zero elements
• Selector matches wrong element (different text/position)

Healing Workflow

Step 1: Analyze the Error

• Read error log to identify which locator failed
• Identify the Page Object file and line number

Step 2: Inspect Current DOM

• Open the page using MCP tools
• Navigate to the same state as the failing test
• Inspect the target area in DOM

Step 3: Find Alternative Locator

Try the following in priority order:

1. Accessibility attributes (aria-label, role)
2. data-testid / data-test
3. id (if stable, not auto-generated)
4. Semantic locator (Playwright getByRole, getByLabel)
5. css selector (stable attributes)
6. xpath (relative, not positional)

Step 4: Validate & Replace

• Verify the new locator matches exactly one element
• Verify the element is the correct one (text, position, behavior)
• Replace the broken locator in the Page Object class
• Re-run the test

Difference from Smart Locator Agent

Verification

After healing:

• Locator must match exactly one element
• Element is the correct target (verify text/attributes)
• Test must pass successfully
• Locator is stable across page reloads

Rules References

• .agents/rules/locator_strategy.md — Master locator priority map
• .agents/rules/automation_rules.md — General automation best practices