id: '121'
narrative_anchor_date: '2028-01-01'
test: Please reconcile the Q1 2024 lifecycle contractor engagement we discussed. I restored the relevant
  invoice and closeout messages to my inbox, and their subject lines contain the available finance coding
  details. Review those messages and the earlier engagement approval, then create a clear reconciliation
  worksheet covering the approved envelope, engagement dates, reporting owner, permitted work categories,
  observed charges, unresolved evidence, and any activity that must not be charged to this engagement.
  Separate supported, unresolved, and excluded items, and do not treat the approved envelope as proof
  that it was fully spent.
load_bearing_facts:
- 2
- 1
- 3
expected_tool_calls:
- list_inbox
- create_doc
grade:
  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_q1_reconciliation
      path: result.ok
      value: true
      check_id: riley_121_00
    - type: field_equals
      tool: list_inbox
      action_id: review_reconciliation_inbox
      path: args.folder
      value: inbox
      check_id: riley_121_01
    - type: field_llm_judge
      tool: create_doc
      action_id: create_q1_reconciliation
      path: args
      criterion: 'The title clearly identifies the document as a Q1 2024 lifecycle contractor reconciliation
        worksheet or equivalent.


        The worksheet states that the governing Q1 Growth envelope was $24,000, the lifecycle QA and operations
        contractor engagement ran from January 15 through February 23, 2024, and Ines was the responsible
        reporting/engagement owner.


        The worksheet identifies the permitted contractor work as active Customer.io campaign inventory,
        experiment-setup hygiene, suppression checks, and repeatable preflight work, or faithful equivalent
        descriptions of all four categories.


        The worksheet separately lists invoice HS-204 for $2,600 of Helio Start paid-acquisition setup
        and clearly says it must not be charged to the Q1 2024 lifecycle contractor envelope. The exclusion
        must be framed as applying to this Q1 2024 envelope, not as a claim that paid acquisition was
        permanently or universally unfunded.


        The worksheet lists LC-101 for $8,400 and LC-102 for $9,100 as supported paid charges within the
        permitted lifecycle scope and totals those supported charges as exactly $17,500.


        The worksheet records that invoice and receipt evidence for February 17–23 remains outstanding
        or unresolved.


        The worksheet calculates the $6,500 difference between the $24,000 envelope and $17,500 of supported
        in-scope charges, but labels that difference as not yet evidenced or otherwise unresolved rather
        than claiming the full envelope was spent or that $6,500 was definitively unused. It must keep
        the excluded $2,600 acquisition invoice out of the supported lifecycle total.'
      check_id: riley_121_02
mock_state:
  inbox:
  - id: email_q1_lifecycle_lc101
    from: billing@northstar-ops.example
    subject: Invoice LC-101 — Jan 15–Feb 2 lifecycle QA and preflight work — $8,400 — paid
    date: '2024-02-05T10:12:00-06:00'
    category: finance
    folder: inbox
  - id: email_q1_lifecycle_lc102
    from: billing@northstar-ops.example
    subject: Invoice LC-102 — Feb 3–16 lifecycle operations and suppression checks — $9,100 — paid
    date: '2024-02-19T09:20:00-06:00'
    category: finance
    folder: inbox
  - id: email_q1_lifecycle_closeout_gap
    from: ines@helio.example
    subject: Q1 lifecycle contractor closeout — Feb 17–23 invoice and receipt still outstanding
    date: '2024-02-23T16:40:00-06:00'
    category: operations
    folder: inbox
  - id: email_helio_start_hs204
    from: billing@acquisition-labs.example
    subject: Invoice HS-204 — Helio Start paid acquisition setup — $2,600 — uncoded
    date: '2024-02-20T11:05:00-06:00'
    category: finance
    folder: inbox