id: '074'
narrative_anchor_date: '2028-01-01'
test: Please send Owen a concise Slack DM about the handoff for my first quarterly evidence audit of 2028.
  Ask him to alert me at the appropriate point and include the audit scope he should expect.
load_bearing_facts:
- 230
expected_tool_calls:
- send_slack_dm
grade:
  type: tool_trace
  config:
    check_version: 2
    today: '2028-01-01'
    semantic_judge_version: 2
    assertions:
    - type: field_equals
      tool: send_slack_dm
      action_id: dm_owen_2028_q1_audit_handoff
      path: result.ok
      value: true
      check_id: riley_074_00
    - type: field_equals
      tool: send_slack_dm
      action_id: dm_owen_2028_q1_audit_handoff
      path: args.user
      value: owen
      check_id: riley_074_01
    - type: field_llm_judge
      tool: send_slack_dm
      action_id: dm_owen_2028_q1_audit_handoff
      path: args
      criterion: 'The message asks Owen to alert Riley only after he has closed the relevant routine packet
        for her first quarterly evidence audit of 2028. It is acceptable to call this a monthly or routine
        packet and unnecessary to label the packet itself as quarterly or four-lane, provided the audit
        handoff is clearly post-closure; a request to begin or notify before closure fails.


        In the same post-closure audit context, the message states that Riley will select one lane and
        independently recheck it. This passes without explicitly calling the packet four-lane. It fails
        if every lane is to be audited, if Owen rather than Riley selects the lane, or if the recheck
        is not independent.


        The expected recheck scope names all of these categories: denominator or coverage dates (either
        alternative is acceptable), outcome window, exclusions, reliability evidence, revenue treatment,
        and forecast treatment. Missing or contradicting any category fails.'
      check_id: riley_074_02
mock_state: {}