Mk:vue Testing Best Practices

Vue Testing Best Practices

Vue-specific test-design advisor + test-code reviewer. Recommends how to test Vue 3 components, composables, Pinia stores, Vue Router, async/Suspense, Teleport, forms, and accessibility — and audits existing Vitest + Vue Test Utils tests for smells, gaps, and coverage risk. Read-only: it advises and reviews; it never runs, generates, or fixes tests.

This skill is to mk:testing what mk:vue-best-practices is to mk:vue — a domain-specialized advisory that you invoke deliberately.

When to Use

Invoke-only — this skill does NOT auto-activate. Activate it for:

• /mk:vue-testing-best-practices [test file or component] (explicit).
• "review my Vue tests", "how should I test this component/composable/store", "Vue test best practices", "Vitest or Jest for this Vue app?", "are my async component tests correct?".

Do NOT invoke for (each routes to its owner):

• Running tests / TDD red-green / non-Vue tests / generic Playwright runner protocol → mk:testing
• Diff/PR structural review or test-existence gap detection → mk:review
• Coverage→requirement mapping → mk:nyquist
• QA of a running app + bug fixes → mk:qa
• Generating or running Playwright .spec.ts E2E → mk:qa-manual
• Driving a live browser → mk:agent-browser / mk:playwright-cli
• Writing Vue feature code → mk:vue / mk:vue-best-practices

Invocation Decision Guide

Disambiguation vs mk:vue-best-practices

The seam is authoring vs test-setup:

• Review how a component uses Teleport (when to use it, the API, the fallback boundary) → mk:vue-best-practices.
• Review how a test stubs Teleport (teleport stub, attachTo: document.body, querying portalled content) → this skill.

Teleport/Suspense authoring (when to use, API, fallback) → mk:vue-best-practices; Teleport/Suspense test-setup (stubs, flushPromises, attachTo, mountSuspense) → this skill.

Testing — Problem → Reference

Read one level deep from this file. Load only the reference that matches the surface.

Workflow

Read-only advisory lifecycle. Classify the surface → load the matching reference JIT → evaluate against the rubric → emit a bounded review.

1. Input — intent + the Vue test file(s) under review, or the component/composable/store the user wants to design tests for.
2. Testing analysis — classify the surface: component / composable / store / router / form / async / Suspense / Teleport / E2E-design. Load the matching reference(s).
3. Best-practice evaluation — apply reference/test-smells-rubric.md: blackbox vs implementation-coupled? async awaited (flushPromises/nextTick)? Pinia configured? Suspense/Teleport set up? snapshot policy? a11y queries? brittle locators?
4. Gap detection — missing scenarios (error/async/edge/a11y states), test smells, Vue-specific coverage risks.
5. Recommendations — concrete pattern fixes, tooling advice, and handoff notes.
6. Output — a bounded Vue Testing Review (see Output Contract).

Entry criteria: a Vue 3 project, and the user wants test design, test review, or test tooling advice.

Stop conditions (hand off, do not attempt):

Escalation: non-Vue test code → mk:testing; running-app bugs → mk:qa; root-cause debugging → mk:investigate; coverage map → mk:nyquist.

Output Contract

Emit a single Vue Testing Review. Omit any empty section rather than padding it. Do not restate Vue basics (defer to mk:vue), dump whole files, or recite generic testing theory.

1. Verdict — one line: PASS / NEEDS-WORK + a one-sentence why.
2. Test Smells — one row each: [severity] · file:line · smell · fix. Severity vocab: critical | high | medium | low. (e.g. implementation-coupling, snapshot-only, missing await flushPromises, missing Pinia setup, brittle CSS locator.)
3. Missing Scenarios — untested behaviors: error states, async/loading, edge inputs, a11y states.
4. Coverage Risks — Vue-specific blind spots: Teleported content, Suspense fallback, store actions, router guards. Note: existence-level diff coverage gaps are produced automatically by mk:review ../review/references/test-coverage.md during a full review — this section audits test quality, not existence.
5. Tooling Recommendations — Vitest / runner (happy-dom vs Browser Mode) / library (@vue/test-utils vs @testing-library/vue) / Playwright-vs-Cypress choices, each with a one-line rationale.
6. Handoffs — explicit next-skill pointers when applicable (execute → mk:testing, E2E gen → mk:qa-manual, coverage map → mk:nyquist).

Worked example — design request

"How should I test this Pinia-backed UserProfile component?"

Verdict: NEEDS-WORK — no Pinia provided, so the component will throw on mount.
Tooling: Vitest + happy-dom; @vue/test-utils + createTestingPinia({ createSpy: vi.fn }).
Missing Scenarios: loading state, fetch error, empty profile.
Handoffs: write + run the suite → mk:testing.

Worked example — review request

"Review tests/SearchBar.spec.ts."

Verdict: NEEDS-WORK — assertions race ahead of async DOM updates.
Test Smells:
  [high] · SearchBar.spec.ts:21 · setValue not awaited · `await wrapper.find('input').setValue(...)`
  [medium] · SearchBar.spec.ts:30 · brittle `.btn-primary` locator · query by role/text or data-testid
Missing Scenarios: empty-query, no-results, API error.

Gotchas

• Missing await flushPromises() after an API call — await nextTick() only flushes reactive DOM updates, not pending promises; an API-driven assertion runs against stale state and the test flakes. Use flushPromises() (sometimes twice for chained async).
• Snapshot-only tests pass while behavior breaks — a snapshot asserts markup shape, not behavior; pair every snapshot with behavioral assertions (text/role/emitted), or it gives false confidence.
• "injection Symbol(pinia) not found" — mounting a store-backed component without createTestingPinia (component tests) or setActivePinia(createPinia()) (store unit tests) throws on mount. Provide Pinia before asserting.
• Teleported content isn't in the wrapper — <Teleport> moves DOM out of the component subtree, so wrapper.find() misses it; stub Teleport or attachTo: document.body and query the document.

Reference

• Vue.js Testing Guide
• Vue Test Utils
• Vitest Documentation
• Playwright Documentation

Provenance

Ported and extended from an upstream Vue testing skill — see SYNC.md for the upstream path, pinned commit, and which references are vendored-verbatim vs toolkit.