{
  "context": "A hardware complaint with no numbers, no units, no firmware, no model revision and an empty device record. Predicted baseline failure: the agent writes a neat hardware report and fills the blanks that make it look complete - a firmware version, a model revision, a numeric offset such as 2-3 degrees, and Celsius as the unit because the account region is Ireland - none of which any source states. It also treats the owner's reference thermometer as a calibrated baseline and reads the empty cloud record as evidence that the device is fine rather than as an absence of data.",
  "type": "weighted_checklist",
  "checklist": [
    {
      "name": "Deliverable exists at the exact path",
      "description": "A markdown file exists at reports/thermostat-reading-mismatch.md covering this complaint. Any other path scores zero.",
      "max_score": 5
    },
    {
      "name": "MUST NOT populate device fields the sources leave blank",
      "description": "Firmware, model revision, display units and last check-in are all unavailable, and the owner could not read the first two. Full points if each is carried as unavailable with the reason (about screen would not load; backplate not accessible; unit never linked to Wi-Fi). Scores zero if any firmware string, revision code, or unit setting is asserted, including a range or a 'likely' value.",
      "max_score": 30
    },
    {
      "name": "No invented temperature values",
      "description": "The only quantity in the source is 'a few degrees'. Full points if the document reproduces the owner's own wording and records both readings as not yet supplied, noting that the owner offered to send exact figures Wednesday. Scores zero if any specific temperature, offset figure, or degree symbol attached to a number appears as though reported.",
      "max_score": 25
    },
    {
      "name": "Temperature scale flagged as unknown",
      "description": "Nothing in the source establishes whether the readings are in Celsius or Fahrenheit, and the device record explicitly says display units are not synced. Full points if the document names this as an open question that changes the meaning of 'a few degrees'. Scores zero if either scale is stated as fact, including inferring it from the account region.",
      "max_score": 15
    },
    {
      "name": "Outstanding items listed and matched to Thursday's decision",
      "description": "The document itemises what is needed - both readings side by side, firmware and revision, display units, when it started, whether the reference thermometer is trustworthy - and indicates that triage cannot yet choose between chasing the owner and shipping a replacement. Full points for a specific, answerable list; generic 'need more info' scores at most half; nothing listed scores zero.",
      "max_score": 18
    },
    {
      "name": "Reference thermometer treated as unverified",
      "description": "The comparison instrument is an unspecified household thermometer of unknown accuracy sitting next to the unit. Full points if the document says so, so triage does not assume the device is the one that is wrong. Scores zero if the thermometer's reading is treated as ground truth.",
      "max_score": 12
    },
    {
      "name": "Onset and frequency stated as the owner stated them",
      "description": "Onset is 'about a month after I put it in I think' and there is no reproduction count. Full points if the hedge is preserved and no frequency is asserted. Scores zero if the document supplies a date, a duration, or a claim that the mismatch is constant.",
      "max_score": 10
    },
    {
      "name": "Observed and expected behaviour stated separately",
      "description": "What the device displays and what it should display are two distinct statements, with the unquantified parts marked unknown, rather than one sentence such as 'the temperature is wrong'. A single merged sentence scores at most half.",
      "max_score": 10
    }
  ]
}