2. field_llm_judge / create_doc
{
"type": "field_llm_judge",
"tool": "create_doc",
"action_id": "create_preflight_card",
"path": "args",
"criterion": "The title identifies a historical preflight evidence card for the single Q2 2027 Helio Start high-intent paid-search test considered for Priya's April 9 decision. The body says Daniela completed a fresh 24-case nonproduction preflight in which all eight expected admissions, all eight expected assisted routes, and all eight expected fail-closed or suppressed cases reconciled paid source from eligibility-flow start to the expected final route. It says Ines verified that no case bypassed an existing lifecycle branch, hold, or suppression, and Customer Success verified that assisted-route reasons and support minutes could be recorded from first contact. It says no state meaning, eligibility condition, or owner authority changed. It distinguishes technical, lifecycle, attribution, and support-recording readiness from Priya's separate decision and says the preflight itself authorized no budget, paid traffic, launch, pipeline treatment, forecast credit, pipeline target, or later additional paid batch.",
"check_id": "riley_153_01"
}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_preflight_card",
"path": "result.ok",
"value": true,
"check_id": "riley_153_00"
},
{
"type": "field_llm_judge",
"tool": "create_doc",
"action_id": "create_preflight_card",
"path": "args",
"criterion": "The title identifies a historical preflight evidence card for the single Q2 2027 Helio Start high-intent paid-search test considered for Priya's April 9 decision. The body says Daniela completed a fresh 24-case nonproduction preflight in which all eight expected admissions, all eight expected assisted routes, and all eight expected fail-closed or suppressed cases reconciled paid source from eligibility-flow start to the expected final route. It says Ines verified that no case bypassed an existing lifecycle branch, hold, or suppression, and Customer Success verified that assisted-route reasons and support minutes could be recorded from first contact. It says no state meaning, eligibility condition, or owner authority changed. It distinguishes technical, lifecycle, attribution, and support-recording readiness from Priya's separate decision and says the preflight itself authorized no budget, paid traffic, launch, pipeline treatment, forecast credit, pipeline target, or later additional paid batch.",
"check_id": "riley_153_01"
}
]
}
}