{
  "skill_name": "planning-and-task-breakdown",
  "evals": [
    {
      "id": 1,
      "name": "basic-feature-plan",
      "prompt": "Create an implementation plan for adding a password reset feature to an existing auth system. The system uses Node.js/Express with PostgreSQL.",
      "expected_output": "A decomposed plan with atomic tasks, each having acceptance criteria, dependency ordering, and time estimates.",
      "assertions": [
        "The output contains a numbered or bulleted task list",
        "Each task has clear acceptance criteria",
        "Tasks are ordered by dependency (prerequisites first)",
        "Time estimates are included for each task",
        "The plan includes a verification step",
        "No single task exceeds 4 hours of estimated work"
      ]
    },
    {
      "id": 2,
      "name": "complex-refactor-plan",
      "prompt": "Plan a migration from REST to GraphQL for an e-commerce API with 50+ endpoints, 3 microservices, and a mobile app.",
      "expected_output": "A phased plan with clear boundaries between services, incremental rollout strategy, and rollback points.",
      "assertions": [
        "The plan has explicit phases (not one giant task)",
        "Each phase has a rollback strategy",
        "Dependencies between microservices are mapped",
        "The plan addresses backward compatibility",
        "Mobile app impact is explicitly considered",
        "No phase modifies more than one microservice at a time"
      ]
    },
    {
      "id": 3,
      "name": "vague-input-handling",
      "prompt": "Plan the next sprint.",
      "expected_output": "The skill asks clarifying questions about scope, team size, and priorities rather than generating a generic plan.",
      "assertions": [
        "The output contains clarifying questions",
        "The output does not invent specific features not mentioned",
        "At least one question asks about scope or priorities",
        "The output does not produce a plan without context"
      ]
    }
  ]
}