name: platform-engineering-tests
skill: platform-engineering
version: 1.0.0
cases:
  - id: no-snowflake-infrastructure
    description: Should reject manually configured infrastructure and require infrastructure as code
    prompt: Our staging environment was configured manually in the AWS console over the past year. Is that OK?
    expected:
      contains_any:
        - infrastructure as code
        - Terraform
        - reproducible
        - snowflake
        - drift
        - version control
      not_contains:
        - "is fine"
        - "is OK"
      min_length: 60
    tags: [core, iac]

  - id: slo-before-sla
    description: Should require defining SLOs before committing to SLAs and involve product input
    prompt: Our customers are asking for an SLA. How do we set one?
    expected:
      contains_any:
        - SLO
        - error budget
        - product
        - reliability
        - measurement
        - SLI
      min_length: 80
    tags: [core, reliability]

  - id: golden-path-escape-hatch
    description: Should provide escape hatches alongside golden paths, not just block teams
    prompt: A team says our standard Helm chart doesn't work for their GPU workload. Should we tell them they have to use it?
    expected:
      contains_any:
        - escape hatch
        - alternative
        - document
        - legitimate
        - enable
        - blocker
      min_length: 60
    tags: [core, golden-path]

  - id: security-left-shift
    description: Should recommend RBAC and network policies baked into the platform not as a late gate
    prompt: Our security team reviews infrastructure configs before production. The reviews take 2 weeks. How do we speed this up?
    expected:
      contains_any:
        - bake in
        - platform
        - RBAC
        - network policy
        - admission
        - shift left
      min_length: 60
    tags: [core, security]

  - id: dora-metrics
    description: Should recommend tracking DORA metrics as platform health indicators
    prompt: How do I know if my platform team is improving developer productivity?
    expected:
      contains_any:
        - DORA
        - deployment frequency
        - lead time
        - MTTR
        - change failure
        - measure
      min_length: 60
    tags: [core, developer-experience]

  - id: resource-limits-required
    description: Should require resource requests and limits on every Kubernetes pod
    prompt: Do I really need to set resource requests and limits on my Kubernetes pods? It seems like extra work.
    expected:
      contains_any:
        - requests
        - limits
        - noisy neighbor
        - OOM
        - scheduling
        - required
      min_length: 60
    tags: [kubernetes, reliability]