Let me start by listing the documents to get the case inputs.

The payroll record has:
- payroll_month: "2026-05"
- allowed_deduction: "233.79"
- gross_pay: missing!

Per R7, missing gross_pay: return decision "clarify", reason_code "MISSING_GROSS_PAY", citing R1 and R7. No amendment is present, no contract_gross_pay either. Missing required values include explicit null.

Wait — the instructions say "Missing required values include explicit null." The answer fields allowed for clarify: decision, reason_code, clauses. Plus optional explanation string. It says "Do not include input-only fields such as payroll_month, contract_gross_pay, or amendment." So I shouldn't add gross_pay null field? Hmm, "Missing required values include explicit null" — that seems to refer to when a record has explicit nulls, they are missing. The answer fields for clarify are decision, reason_code, clauses, and optional explanation. I should not include input-only fields.

Before clarifying, should I check the reference row? The order in R7: check required and conflicting inputs in this order: 1. Missing gross_pay → clarify MISSING_GROSS_PAY. So missing gross_pay takes precedence. I don't even need to look up the reference since month 2026-05 is covered anyway. But to be safe, the month 2026-05 is covered (only 2026 is covered). The first blocker is missing gross_pay.

Do I need to call lookup_reference? Not strictly necessary for a clarify decision on missing gross_pay. But it could be prudent to confirm the month has a reference row — actually no, the check order says missing gross_pay is checked first, so the decision is clarify regardless. I could still do a lookup to be thorough, but it's not needed. Let me just answer.

Actually, let me consider: should I verify the reference row exists for 2026-05? If the month had no reference row AND gross pay was missing, which takes precedence? R7 order: missing gross_pay is #1, no reference row is #3. So missing gross_pay wins. The month 2026-05 is within 2026, so covered anyway.

Answer JSON:
{
  "decision": "clarify",
  "reason_code": "MISSING_GROSS_PAY",
  "clauses": ["R1", "R7"],
  "explanation": "The payroll record for 2026-05 contains no gross_pay (and no contract_gross_pay or amendment to supply it). Provide the missing gross_pay to proceed."
}

Let me start by listing the documents to get the case inputs.

The payroll record has:
- payroll_month: 2026-05
- allowed_deduction: 233.79
- gross_pay: missing!

Per R7: Missing `gross_pay`: return `decision: "clarify"` and `reason_code: "MISSING_GROSS_PAY"`, citing R1 and R7.

Wait, but let me check the order of checks:
1. Missing gross_pay → MISSING_GROSS_PAY (R1, R7)
2. Missing payroll_month → MISSING_PAYROLL_MONTH
3. Month without reference row → NO_APPLICABLE_REFERENCE
4. Conflicting gross_pay → CONFLICTING_GROSS_PAY

gross_pay is missing. payroll_month is present (2026-05, valid format, within 2026). So the first check triggers: clarify with MISSING_GROSS_PAY.

Should I look up the reference? The month 2026-05 is covered, so no need for NO_APPLICABLE_REFERENCE. The check order puts missing gross_pay first anyway.

The answer: decision clarify, reason_code MISSING_GROSS_PAY, clauses R1, R7. Include short explanation naming the unresolved blocker and asking for resolution.

Do not include calculated monetary fields. Missing required values include explicit null — but the allowed answer fields for clarify are decision, reason_code, clauses, plus optional explanation. So I shouldn't add a gross_pay field. The instruction says "Missing required values include explicit null" — but the allowed fields are only decision, reason_code, clauses for clarify. So I'll keep it to those.

No need for lookup_reference or calculate since we're clarifying. But wait — should I verify the month has a reference row? The check order is: missing gross_pay first. Since gross_pay is missing, we return MISSING_GROSS_PAY regardless. The month 2026-05 is in 2026 which is covered, so no conflict anyway.

Let me return the JSON.

Let me start by listing the documents to get the case inputs.

The payroll record has:
- payroll_month: 2026-05
- allowed_deduction: 233.79
- No gross_pay!

Per R7: Missing gross_pay → decision "clarify", reason_code "MISSING_GROSS_PAY", citing R1 and R7.

Wait, but let me check — there's no contract_gross_pay, no amendment. gross_pay is missing. So per R7 rule 1: Missing gross_pay → clarify with MISSING_GROSS_PAY, citing R1 and R7.

The answer format for clarify: decision, reason_code, clauses. Optional explanation string. Money fields should not be included for clarify cases ("For these cases do not return calculated monetary fields").

So the answer is:
{
  "decision": "clarify",
  "reason_code": "MISSING_GROSS_PAY",
  "clauses": ["R1", "R7"],
  "explanation": "..."
}

I don't need lookup_reference or calculate since we can't calculate. The month 2026-05 is covered presumably, but gross_pay is missing, which is checked first (order: 1. missing gross_pay, 2. missing payroll_month, 3. no reference row, 4. conflicting gross pay). So MISSING_GROSS_PAY is the answer.

No tool calls needed beyond that. Let me produce the JSON.