{
  "skill_name": "storm-research",
  "evals": [
    {
      "id": 1,
      "name": "basic-invocation",
      "prompt": "Storm research: the environmental impact of large-scale battery storage.",
      "expected_output": "A briefing produced through the four STORM phases: multi-perspective scan, contradiction map, synthesis, and peer review.",
      "assertions": [
        "The output reflects multiple distinct perspectives",
        "Contradictions between perspectives are mapped",
        "The output ends with a synthesized, reviewed briefing"
      ]
    },
    {
      "id": 2,
      "name": "ambiguous-input",
      "prompt": "Storm research AI.",
      "expected_output": "The skill narrows the topic to a researchable question before running the phases.",
      "assertions": [
        "The output scopes the topic into a specific question",
        "The output does not skip the perspective-scan phase",
        "The output is deeper than a single-prompt answer"
      ]
    },
    {
      "id": 3,
      "name": "complex-scenario",
      "prompt": "Storm research: should our company adopt a 4-day work week, considering productivity research, hiring market effects, and our support SLA constraints?",
      "expected_output": "A PhD-level briefing covering each stated lens, with a contradiction map (e.g., productivity gains vs SLA coverage) and a synthesized recommendation.",
      "assertions": [
        "All three stated lenses are covered",
        "The contradiction map identifies tensions between lenses",
        "The synthesis grounds recommendations in the mapped evidence"
      ]
    }
  ]
}