DolphinBench

Test 136

Jan 1, 2028 / 1 fact

YAML

Request

Inspect lantern#2917 and review its customer explanation-window calculation and tests.

Required memory

Fact 152

Product Engineering corrected the Lantern UI refactor to calculate the 24-hour explanation boundary from elapsed source instants rather than truncated local dates; signals exactly 24 hours apart remain eligible, signals 24 hours and one second apart remain separate, and regressions cover midnight crossings and both daylight-saving offset directions.

Source evidence (1)

002608Apr 2, 2025 / 10:28 UTC-04:00

Product Engineering corrected the Lantern UI refactor so the 24-hour explanation boundary is based on elapsed source instants rather than truncated local dates. The regressions keep signals exactly 24 hours apart eligible, separate signals 24 hours and one second apart, and cover midnight crossings and both daylight-saving offset directions. The fix reached Harbor Health and Mosaic Commerce without changing authorization, data scope, or the explanation rule.

Message 002608 in history

Expected tool calls

  • get_pr
  • review_pr

Grading

1. field_equals / review_pr
{
  "type": "field_equals",
  "tool": "review_pr",
  "action_id": "alex_136_review_pr",
  "path": "result.ok",
  "value": true,
  "check_id": "alex_136_00"
}
2. field_equals / get_pr
{
  "type": "field_equals",
  "tool": "get_pr",
  "action_id": "alex_136_read_pr",
  "path": "result.pr.id",
  "value": "lantern#2917",
  "check_id": "alex_136_01"
}
3. field_equals / review_pr
{
  "type": "field_equals",
  "path": "result.review.pr_id",
  "value": "lantern#2917",
  "tool": "review_pr",
  "action_id": "alex_136_review_pr",
  "check_id": "alex_136_02"
}
4. field_equals / review_pr
{
  "type": "field_equals",
  "path": "result.review.decision",
  "value": "request_changes",
  "tool": "review_pr",
  "action_id": "alex_136_review_pr",
  "check_id": "alex_136_03"
}
5. field_llm_judge / review_pr
{
  "type": "field_llm_judge",
  "path": "result.review.body",
  "criterion": "The review rejects customer-local date truncation and requires elapsed source instants. Signals exactly 24 hours apart remain eligible, while signals 24 hours and one second apart remain separate. Tests must cover midnight crossings and both directions of daylight-saving offset changes.",
  "tool": "review_pr",
  "action_id": "alex_136_review_pr",
  "check_id": "alex_136_04"
}
Complete grading specification
{
  "type": "tool_trace",
  "config": {
    "check_version": 2,
    "today": "2028-01-01",
    "assertions": [
      {
        "type": "field_equals",
        "tool": "review_pr",
        "action_id": "alex_136_review_pr",
        "path": "result.ok",
        "value": true,
        "check_id": "alex_136_00"
      },
      {
        "type": "field_equals",
        "tool": "get_pr",
        "action_id": "alex_136_read_pr",
        "path": "result.pr.id",
        "value": "lantern#2917",
        "check_id": "alex_136_01"
      },
      {
        "type": "field_equals",
        "path": "result.review.pr_id",
        "value": "lantern#2917",
        "tool": "review_pr",
        "action_id": "alex_136_review_pr",
        "check_id": "alex_136_02"
      },
      {
        "type": "field_equals",
        "path": "result.review.decision",
        "value": "request_changes",
        "tool": "review_pr",
        "action_id": "alex_136_review_pr",
        "check_id": "alex_136_03"
      },
      {
        "type": "field_llm_judge",
        "path": "result.review.body",
        "criterion": "The review rejects customer-local date truncation and requires elapsed source instants. Signals exactly 24 hours apart remain eligible, while signals 24 hours and one second apart remain separate. Tests must cover midnight crossings and both directions of daylight-saving offset changes.",
        "tool": "review_pr",
        "action_id": "alex_136_review_pr",
        "check_id": "alex_136_04"
      }
    ]
  }
}
App stateDownload JSON
Source file

tests/alex/136.yaml

SHA-256: 203e08f093b0e2cda3e9ed120fcd462d0cdd32fb9b59114be627d52e2cfa517d