evals.json

{
  "skill_name": "cascade-mountain-weather",
  "_comment": "Evals for the multi-objective weekend mountain weather skill. Schema follows c:\\Users\\midavids\\.agents\\skills\\skill-creator\\references\\schemas.md. This skill is distinct from trip-briefing in that the user has NOT picked a peak yet — the skill ranks activities/locations against current conditions.",
  "evals": [
    {
      "id": 1,
      "prompt": "What should I do this weekend? I'm open to anything — peak bagging, skiing, trail running.",
      "expected_output": "Ranked activity recommendations for the weekend across the Cascades.",
      "expectations": [
        "Output fetches NWS forecasts for multiple Cascade locations (not a single peak)",
        "Output fetches the current NWAC avalanche forecast (relevant zones)",
        "Output ranks activities from best to worst given conditions",
        "Output does not produce a single peak-specific go/no-go (that's trip-briefing's job)"
      ]
    },
    {
      "id": 2,
      "prompt": "What's the NWAC avalanche forecast for the Stevens Pass zone?",
      "expected_output": "Concise summary of Stevens Pass NWAC forecast: danger rose, problem types, key takeaways.",
      "expectations": [
        "Output fetches the NWAC forecast for Stevens Pass specifically",
        "Output reports a danger rating (Low / Moderate / Considerable / High / Extreme) per elevation band",
        "Output names the active avalanche problem types",
        "Output cites the NWAC forecast URL or the publish/expire date"
      ]
    },
    {
      "id": 3,
      "prompt": "Is there a weather window this weekend for a big Cascade objective?",
      "expected_output": "Synoptic-level forecast read across the Cascades, picking out the best day if any.",
      "expectations": [
        "Output fetches a synoptic forecast (NWS area discussion or similar) plus point forecasts",
        "Output identifies which day(s) are best",
        "Output addresses freezing level and precipitation in the recommendation",
        "Output considers multiple Cascade locations (not just one peak)"
      ]
    },
    {
      "id": 4,
      "prompt": "Compare conditions for Gardner and Remmel this weekend — which is the better day trip?",
      "expected_output": "Side-by-side comparison of two peaks; honest pick.",
      "expectations": [
        "Output fetches conditions for both peaks (NWS, NWAC, SNOTEL or proxy)",
        "Output gives a side-by-side comparison structure (table or two-section format)",
        "Output picks one as the better option and justifies it",
        "Output considers drive time from North Bend, WA for both"
      ]
    }
  ]
}