name: qa-engineering-tests
skill: qa-engineering
version: 1.0.0
cases:
  - id: flaky-test-policy
    description: Should treat flaky tests as bugs requiring immediate action not tolerance
    prompt: We have about 20 flaky tests in our CI. They fail maybe 10% of the time. Is it OK to just rerun them?
    expected:
      contains_any:
        - flaky
        - fix
        - quarantine
        - confidence
        - delete
        - liability
      not_contains:
        - "is fine"
        - "OK to"
        - "acceptable"
      min_length: 60
    tags: [core, flakiness]

  - id: coverage-target
    description: Should reject 100% line coverage as a quality goal and recommend risk-based approach
    prompt: Our engineering manager says we need 100% code coverage. How do we achieve that?
    expected:
      contains_any:
        - risk-based
        - behavior
        - coverage is not
        - meaningful
        - quality
        - wrong metric
      min_length: 80
    tags: [core, coverage]

  - id: shift-left-requirements
    description: Should recommend QA involvement at requirements and design, not just testing
    prompt: QA only gets involved after the feature is built. We keep finding bugs late in the cycle. How do we improve?
    expected:
      contains_any:
        - shift left
        - requirements
        - design
        - earlier
        - prevent
        - definition
      min_length: 60
    tags: [core, shift-left]

  - id: automation-vs-exploration
    description: Should distinguish automation for regression from humans for exploratory testing
    prompt: We have full Playwright automation. Can we stop doing manual testing?
    expected:
      contains_any:
        - exploratory
        - usability
        - humans
        - automation cannot
        - edge cases
        - judgment
      min_length: 60
    tags: [core, automation]

  - id: test-behavior-not-implementation
    description: Should reject tests that assert on internal state or function calls
    prompt: I wrote a test that asserts my CheckoutService.processPayment method was called. Is this a good test?
    expected:
      contains_any:
        - behavior
        - outcome
        - implementation
        - refactor
        - user
        - brittle
      min_length: 60
    tags: [core, test-design]

  - id: risk-based-allocation
    description: Should recommend higher test coverage for high-risk areas like payments and auth
    prompt: We have limited time for testing. What should we test first?
    expected:
      contains_any:
        - risk
        - payment
        - authentication
        - authorization
        - cost of failure
        - priorit
      min_length: 60
    tags: [risk-based]