id: 059
narrative_anchor_date: '2028-01-01'
test: Create a concise design-input card for a bounded Expansion Assist proposal before design work begins.
  Capture the key observed pilot pattern and the evidence limitations that reviewers must preserve. Title
  it `Expansion Assist — bounded proposal evidence card`.
load_bearing_facts:
- 50
- 51
expected_tool_calls:
- 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_evidence_card
      path: result.ok
      value: true
      check_id: riley_059_00
    - check_id: riley_059_01
      type: field_equals
      tool: create_doc
      action_id: create_evidence_card
      path: args.title
      value: Expansion Assist — bounded proposal evidence card
    - type: field_llm_judge
      tool: create_doc
      action_id: create_evidence_card
      path: args
      criterion: 'The document accurately states the observed concentration: all six expansions and seven
        of the eight help requests came from the 12 exposed accounts whose product-visible usage ceiling
        had been crossed within the preceding 14 days. It may also note that the other 18 exposed accounts
        produced one help request and no expansions. Missing or materially contradicting the concentration
        must fail.


        The document explicitly limits interpretation of the pilot: the result is descriptive or observational,
        does not establish causal lift, and does not justify a broad rollout. Missing either the non-causal
        limitation or the no-broad-rollout boundary must fail.'
      check_id: riley_059_02
mock_state: {}