6. field_llm_judge / create_doc
{
"type": "field_llm_judge",
"tool": "create_doc",
"action_id": "create_development_memo",
"path": "args",
"criterion": "The memo must state that Ines’s H1 2027 routine-send audit development expectation was met and accurately cover all necessary evidence: 17 routine non-state-changing sends from April 1 through June 23; every audit remained visibly open until completion and closed within 24 hours; none became overdue; no later routine send was released against an overdue audit; and no review found a customer correction, state conflict, message defect, or incident issue requiring escalation. It must also explain that completion did not expand her authority beyond no more than three routine non-state-changing sends per week and did not transfer the customer-correction, incident-stop, disputed-semantics, engineering-release, or rollback lanes to her. Missing or contradictory claims fail.\n\nThe memo must state that Owen’s H2 2027 mature-outcome-packet development goal was met and accurately explain that his independently assembled packet used the correct denominator and dates, waited for every predeclared window, placed owner certifications and measurement limits before conclusions, and established that no semantic interpretation was disputed. It must state that any future semantic dispute requires Owen to leave the review incomplete and escalate it to Riley, and that completion transferred no board-readiness, causal, revenue, forecast, or executive-language decision authority to Owen. Missing or contradictory claims fail.",
"check_id": "riley_123_05"
}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_development_memo",
"path": "result.ok",
"value": true,
"check_id": "riley_123_00"
},
{
"type": "field_equals",
"tool": "send_slack_dm",
"action_id": "send_priya_summary",
"path": "result.ok",
"value": true,
"check_id": "riley_123_01"
},
{
"check_id": "riley_123_02",
"type": "field_equals",
"tool": "create_doc",
"action_id": "create_development_memo",
"path": "args.title",
"value": "2027 Lifecycle Systems development evidence — Ines and Owen"
},
{
"check_id": "riley_123_03",
"type": "field_equals",
"tool": "create_doc",
"action_id": "create_development_memo",
"path": "args.folder",
"value": "Growth"
},
{
"check_id": "riley_123_04",
"type": "field_equals",
"tool": "send_slack_dm",
"action_id": "send_priya_summary",
"path": "args.user",
"value": "priya"
},
{
"type": "field_llm_judge",
"tool": "create_doc",
"action_id": "create_development_memo",
"path": "args",
"criterion": "The memo must state that Ines’s H1 2027 routine-send audit development expectation was met and accurately cover all necessary evidence: 17 routine non-state-changing sends from April 1 through June 23; every audit remained visibly open until completion and closed within 24 hours; none became overdue; no later routine send was released against an overdue audit; and no review found a customer correction, state conflict, message defect, or incident issue requiring escalation. It must also explain that completion did not expand her authority beyond no more than three routine non-state-changing sends per week and did not transfer the customer-correction, incident-stop, disputed-semantics, engineering-release, or rollback lanes to her. Missing or contradictory claims fail.\n\nThe memo must state that Owen’s H2 2027 mature-outcome-packet development goal was met and accurately explain that his independently assembled packet used the correct denominator and dates, waited for every predeclared window, placed owner certifications and measurement limits before conclusions, and established that no semantic interpretation was disputed. It must state that any future semantic dispute requires Owen to leave the review incomplete and escalate it to Riley, and that completion transferred no board-readiness, causal, revenue, forecast, or executive-language decision authority to Owen. Missing or contradictory claims fail.",
"check_id": "riley_123_05"
},
{
"type": "field_llm_judge",
"tool": "send_slack_dm",
"action_id": "send_priya_summary",
"path": "args",
"criterion": "The message must concisely summarize that both Ines’s routine-send audit expectation and Owen’s mature-outcome-packet goal were met. For Ines, it must convey the material completion evidence (17 sends, timely audits with no overdue-audit release or escalation-triggering issue) and that her routine-send limit and exception/control boundaries remain unchanged. For Owen, it must convey the material packet-quality evidence, preserve escalation to Riley whenever semantics are disputed, and make clear that no board-readiness, causal, revenue, forecast, or executive-language authority transferred. Missing either person’s result or contradicting a continuing boundary fails.",
"check_id": "riley_123_06"
}
]
}
}