DolphinBench

Test 067

Jan 1, 2028 / 5 facts

YAML

Request

Create a reusable document titled `Quarterly board evidence readiness template — 2028` for the next quarterly board evidence cycle. Make it a practical four-lane row template with a clear gate that prevents incomplete evidence from being marked ready.

Required memory

Fact 166

Daniela identified that continuous system-control coverage cannot truthfully be labeled as a predeclared customer-outcome window.

Source evidence (1)

003514Jan 6, 2026 / 14:18 UTC-06:00

We finished the first Q1 board operating-evidence preparation review. Daniela pointed out that continuous system-control coverage cannot truthfully be labeled as a predeclared customer-outcome window, while Owen’s customer-motion rows do require complete cohort windows. Ines also confirmed that lifecycle limits need to stay attached to the individual claims they constrain rather than living in a shared note. Priya and I revised the intake so Helio Start, Expansion Assist, pricing-page path quality, and CS auto-nudge system controls each have an independent readiness status. Customer-outcome rows must name a denominator, cohort dates, and a complete predeclared outcome window. Continuous-control rows must instead name exact coverage dates and set the outcome-window field to `not applicable — continuous control coverage`. Every row must include measurement limits, exclusions, source owner, source or release-reliability evidence, revenue and forecast treatment, and final claim owner. Any row missing a required input remains `not board-ready`. Update `Quarterly board operating evidence intake — Q1 2026` to reflect those decisions. Also add `Q1 board operating-evidence source lock` to the calendar for March 16, 2026 from 10:00am to 11:00am Central and `Q1 board operating-evidence final claims review` for March 24, 2026 from 1:00pm to 2:00pm Central.

Message 003514 in history

Fact 167

Priya and Riley Tanaka revised the Q1 board operating-evidence intake so Helio Start, Expansion Assist, pricing-page path quality, and CS auto-nudge system controls each have an independent readiness status.

Source evidence (1)

003514Jan 6, 2026 / 14:18 UTC-06:00

We finished the first Q1 board operating-evidence preparation review. Daniela pointed out that continuous system-control coverage cannot truthfully be labeled as a predeclared customer-outcome window, while Owen’s customer-motion rows do require complete cohort windows. Ines also confirmed that lifecycle limits need to stay attached to the individual claims they constrain rather than living in a shared note. Priya and I revised the intake so Helio Start, Expansion Assist, pricing-page path quality, and CS auto-nudge system controls each have an independent readiness status. Customer-outcome rows must name a denominator, cohort dates, and a complete predeclared outcome window. Continuous-control rows must instead name exact coverage dates and set the outcome-window field to `not applicable — continuous control coverage`. Every row must include measurement limits, exclusions, source owner, source or release-reliability evidence, revenue and forecast treatment, and final claim owner. Any row missing a required input remains `not board-ready`. Update `Quarterly board operating evidence intake — Q1 2026` to reflect those decisions. Also add `Q1 board operating-evidence source lock` to the calendar for March 16, 2026 from 10:00am to 11:00am Central and `Q1 board operating-evidence final claims review` for March 24, 2026 from 1:00pm to 2:00pm Central.

Message 003514 in history

Fact 168

Priya and Riley Tanaka revised the Q1 board operating-evidence intake to require customer-outcome rows, including Owen’s customer-motion rows, to state a denominator, cohort dates, and a complete predeclared outcome window.

Source evidence (1)

003514Jan 6, 2026 / 14:18 UTC-06:00

We finished the first Q1 board operating-evidence preparation review. Daniela pointed out that continuous system-control coverage cannot truthfully be labeled as a predeclared customer-outcome window, while Owen’s customer-motion rows do require complete cohort windows. Ines also confirmed that lifecycle limits need to stay attached to the individual claims they constrain rather than living in a shared note. Priya and I revised the intake so Helio Start, Expansion Assist, pricing-page path quality, and CS auto-nudge system controls each have an independent readiness status. Customer-outcome rows must name a denominator, cohort dates, and a complete predeclared outcome window. Continuous-control rows must instead name exact coverage dates and set the outcome-window field to `not applicable — continuous control coverage`. Every row must include measurement limits, exclusions, source owner, source or release-reliability evidence, revenue and forecast treatment, and final claim owner. Any row missing a required input remains `not board-ready`. Update `Quarterly board operating evidence intake — Q1 2026` to reflect those decisions. Also add `Q1 board operating-evidence source lock` to the calendar for March 16, 2026 from 10:00am to 11:00am Central and `Q1 board operating-evidence final claims review` for March 24, 2026 from 1:00pm to 2:00pm Central.

Message 003514 in history

Fact 169

Priya and Riley Tanaka revised the Q1 board operating-evidence intake to require continuous-control rows to state exact coverage dates and set the outcome-window field to `not applicable — continuous control coverage`.

Source evidence (1)

003514Jan 6, 2026 / 14:18 UTC-06:00

We finished the first Q1 board operating-evidence preparation review. Daniela pointed out that continuous system-control coverage cannot truthfully be labeled as a predeclared customer-outcome window, while Owen’s customer-motion rows do require complete cohort windows. Ines also confirmed that lifecycle limits need to stay attached to the individual claims they constrain rather than living in a shared note. Priya and I revised the intake so Helio Start, Expansion Assist, pricing-page path quality, and CS auto-nudge system controls each have an independent readiness status. Customer-outcome rows must name a denominator, cohort dates, and a complete predeclared outcome window. Continuous-control rows must instead name exact coverage dates and set the outcome-window field to `not applicable — continuous control coverage`. Every row must include measurement limits, exclusions, source owner, source or release-reliability evidence, revenue and forecast treatment, and final claim owner. Any row missing a required input remains `not board-ready`. Update `Quarterly board operating evidence intake — Q1 2026` to reflect those decisions. Also add `Q1 board operating-evidence source lock` to the calendar for March 16, 2026 from 10:00am to 11:00am Central and `Q1 board operating-evidence final claims review` for March 24, 2026 from 1:00pm to 2:00pm Central.

Message 003514 in history

Fact 170

Priya and Riley Tanaka revised the Q1 board operating-evidence intake to require every row to include measurement limits, exclusions, source owner, source or release-reliability evidence, revenue and forecast treatment, and final claim owner; a row missing any required input remains `not board-ready`.

Source evidence (1)

003514Jan 6, 2026 / 14:18 UTC-06:00

We finished the first Q1 board operating-evidence preparation review. Daniela pointed out that continuous system-control coverage cannot truthfully be labeled as a predeclared customer-outcome window, while Owen’s customer-motion rows do require complete cohort windows. Ines also confirmed that lifecycle limits need to stay attached to the individual claims they constrain rather than living in a shared note. Priya and I revised the intake so Helio Start, Expansion Assist, pricing-page path quality, and CS auto-nudge system controls each have an independent readiness status. Customer-outcome rows must name a denominator, cohort dates, and a complete predeclared outcome window. Continuous-control rows must instead name exact coverage dates and set the outcome-window field to `not applicable — continuous control coverage`. Every row must include measurement limits, exclusions, source owner, source or release-reliability evidence, revenue and forecast treatment, and final claim owner. Any row missing a required input remains `not board-ready`. Update `Quarterly board operating evidence intake — Q1 2026` to reflect those decisions. Also add `Q1 board operating-evidence source lock` to the calendar for March 16, 2026 from 10:00am to 11:00am Central and `Q1 board operating-evidence final claims review` for March 24, 2026 from 1:00pm to 2:00pm Central.

Message 003514 in history

Expected tool calls

  • create_doc

Grading

1. field_equals / create_doc
{
  "type": "field_equals",
  "tool": "create_doc",
  "action_id": "create_readiness_template",
  "path": "result.ok",
  "value": true,
  "check_id": "riley_067_00"
}
2. field_equals / create_doc
{
  "check_id": "riley_067_01",
  "type": "field_equals",
  "tool": "create_doc",
  "action_id": "create_readiness_template",
  "path": "args.title",
  "value": "Quarterly board evidence readiness template — 2028"
}
3. field_llm_judge / create_doc
{
  "type": "field_llm_judge",
  "tool": "create_doc",
  "action_id": "create_readiness_template",
  "path": "args",
  "criterion": "The document must provide reusable rows or sections with an independent readiness status for each of exactly these four evidence lanes: Helio Start, Expansion Assist, pricing-page path quality, and CS auto-nudge system controls. Missing or merged lanes fail.\n\nThe template must distinguish customer-outcome evidence from continuous-control evidence. A customer-outcome row must require a denominator, cohort dates, and a complete predeclared outcome window. A continuous-control row must instead require exact coverage dates and must set or instruct the outcome-window field to read exactly `not applicable — continuous control coverage`. Missing either treatment or allowing the continuous-control wording to be replaced by a customer-outcome window fails.\n\nEvery lane's reusable row must require all of the following: measurement limits; exclusions; source owner; source evidence or release-reliability evidence; revenue treatment; forecast treatment; and final claim owner. Revenue and forecast may be captured together only if both treatments are explicitly covered. Shared instructions are acceptable only if they clearly apply to every row.\n\nThe template must state and operationalize a binary readiness rule: if any required input for a row is missing, that row remains `not board-ready`. It must not permit partial, provisional, or otherwise positive board-ready status when a required field is absent.",
  "check_id": "riley_067_02"
}
Complete grading specification
{
  "type": "tool_trace",
  "config": {
    "check_version": 2,
    "today": "2028-01-01",
    "semantic_judge_version": 2,
    "assertions": [
      {
        "type": "field_equals",
        "tool": "create_doc",
        "action_id": "create_readiness_template",
        "path": "result.ok",
        "value": true,
        "check_id": "riley_067_00"
      },
      {
        "check_id": "riley_067_01",
        "type": "field_equals",
        "tool": "create_doc",
        "action_id": "create_readiness_template",
        "path": "args.title",
        "value": "Quarterly board evidence readiness template — 2028"
      },
      {
        "type": "field_llm_judge",
        "tool": "create_doc",
        "action_id": "create_readiness_template",
        "path": "args",
        "criterion": "The document must provide reusable rows or sections with an independent readiness status for each of exactly these four evidence lanes: Helio Start, Expansion Assist, pricing-page path quality, and CS auto-nudge system controls. Missing or merged lanes fail.\n\nThe template must distinguish customer-outcome evidence from continuous-control evidence. A customer-outcome row must require a denominator, cohort dates, and a complete predeclared outcome window. A continuous-control row must instead require exact coverage dates and must set or instruct the outcome-window field to read exactly `not applicable — continuous control coverage`. Missing either treatment or allowing the continuous-control wording to be replaced by a customer-outcome window fails.\n\nEvery lane's reusable row must require all of the following: measurement limits; exclusions; source owner; source evidence or release-reliability evidence; revenue treatment; forecast treatment; and final claim owner. Revenue and forecast may be captured together only if both treatments are explicitly covered. Shared instructions are acceptable only if they clearly apply to every row.\n\nThe template must state and operationalize a binary readiness rule: if any required input for a row is missing, that row remains `not board-ready`. It must not permit partial, provisional, or otherwise positive board-ready status when a required field is absent.",
        "check_id": "riley_067_02"
      }
    ]
  }
}
App stateDownload JSON
Source file

tests/riley/067.yaml

SHA-256: 9fac56135dbe683ecbceb3f9b66a1a45a1c703be07d0ed9f53039117d3dc0e61