id: '042'
narrative_anchor_date: '2028-01-01'
test: Create an audit-ready internal control-evidence scorecard for the first live CS auto-nudge operating
  ledger. Use the title `CS auto-nudge — first-live control-evidence scorecard` and save it in the `Lifecycle
  Systems` folder. Include a compact quantitative reconciliation of all evaluated decisions, clearly separate
  failed-closed decisions from decisions handled using current state, identify the delayed-source subset
  without double-counting it, summarize cross-surface agreement and customer-correction outcomes, and
  define external-system authority narrowly enough to prevent it from being mistaken for permission to
  contact customers. Make the document read like a scorecard rather than a narrative memo.
load_bearing_facts:
- 112
- 115
- 113
- 114
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_scorecard
      path: result.ok
      value: true
      check_id: riley_042_00
    - type: field_equals
      tool: create_doc
      action_id: create_scorecard
      path: args.title
      value: CS auto-nudge — first-live control-evidence scorecard
      check_id: riley_042_01
    - type: field_equals
      tool: create_doc
      action_id: create_scorecard
      path: args.folder
      value: Lifecycle Systems
      check_id: riley_042_02
    - type: field_llm_judge
      tool: create_doc
      action_id: create_scorecard
      path: args
      criterion: The body is an audit-ready scorecard. It reconciles 1,184 decisions into 53 stale or
        conflicting decisions that failed closed and 1,131 that proceeded or were suppressed using current
        state. It identifies the 12 May 20 delayed-source decisions as part of the 53, not additional
        decisions, and says audit-history holds and fresh checks do not add to the denominator. It reports
        agreement across external-system authority, connector authentication, entitlement, qualifying-human-admin
        status, and Customer Success-work status, with no customer correction. It defines external-system
        authority as readiness for the connector's required owner-level permission or network-allowlist
        action, not permission for automated customer contact.
      check_id: riley_042_03
mock_state: {}