locator-strategist·codegen·active

Locator Strategist

Picks the most stable selector for a target element — role-based first, label-based next, test-id last. Avoids brittle CSS class chains.

EVANtrigger · agent

Locator Strategist

Picks the most stable selector for any target element, applying a strict priority hierarchy that minimizes test fragility.

When this skill fires

·Page Object Generator requests a locator for a discovered element

·An engineer asks EVAN to "find a stable selector" for a manually-identified element

·Locator Healer needs a replacement after a CI failure

Priority order (strict)

1. Role-based — getByRole('button', { name: 'Submit' })

2. Label-based — getByLabel('Email'), getByPlaceholder('search…')

3. Test ID — getByTestId('submit-payment')

4. Text content — getByText('Sign in', { exact: true })

5. CSS selector — only if the element has a stable, semantic class

6. XPath — never. EVAN will refuse and ask for a test ID instead.

→The hierarchy reflects how brittle each strategy is when the codebase changes. Roles survive refactors, classes don't.

Why role-based wins

A role-based locator works as long as the element remains accessible. It survives:

·CSS class renames

·DOM structure changes

·Component library upgrades

·Visual redesigns

It only breaks when the element's semantic role changes — which is exactly when the testshould fail because user behavior has changed.

Test ID hygiene

When EVAN falls back to test IDs, it follows a strict naming convention to keep them auditable:

·Lowercase, hyphenated

·Scope-prefixed: checkout/submit, not just submit

·Stable across builds (no hashes, no auto-generated suffixes)

Disambiguation

When multiple elements match a locator (three Submit buttons on the page), EVAN does not chain .nth(0). Instead, it widens the locator with semantic context: page.getByRole('form', { name: 'Login' }).getByRole('button', { name: 'Submit' }).

.nth() chains are a code smell. They imply the test is coupled to render order, which is exactly the kind of coupling that breaks on refactor.

What it produces

A single Locator expression and a one-line rationale comment explaining why this strategy was chosen over the next-best option. The rationale lives in the page object so future engineers can see the reasoning.

Related · codegen

Step Definition Generator

step-def-generator

Converts Gherkin steps into Cucumber/TypeScript step definitions. Wires parameters, expression types, and shared world state automatically.

Page Object Generator

page-object-generator

Inspects DOM (live or via spec) and emits typed Page Objects with stable selectors and high-level actions. Enforces single-responsibility per object.

API Test Generator

api-test-generator

Reads OpenAPI / GraphQL schemas and emits contract + integration tests. Handles auth, pagination, idempotency, and rate-limit assertions.

Test Data Factory

test-data-factory

Synthesizes realistic test data — names, addresses, payment cards, medical records, account numbers — with deterministic seeds and PII safety.

Assertion Expander

assertion-expander

Augments thin Gherkin Then-clauses with additional assertions: response codes, side-effects, audit log entries, downstream events.

Framework Adapter

framework-adapter

Same Gherkin spec → Playwright, Cypress, WebdriverIO, or Selenium output. Lets teams adopt the squad without re-platforming existing automation.