Know which overdue Dynamics GP balances to chase first

**Objective:** Produce a collections worklist of open receivables that are at least `{min-days-overdue}` days past due (default 30), for customers matching `{customer-name}` (default all customers), ranked by how much is overdue.

Pass `{company}` as `company` on every call (if it is empty, omit `company` and the default company is used). Pass a generous `max-records` (1000). If a result comes back with exactly that many rows, say the result may be truncated.

**Step 1: Pull candidate receivables documents**

Call `dynamics-gp-get-rm-trx`. If `{customer-name}` is given pass it as `customer-name.contains`. Pass `document-date.to` as today minus the overdue threshold: a document cannot fall due before its document date, so this narrows the read without losing anything that is overdue.

Keep only documents where:

* `current-transaction-amount` is greater than 0 (the unpaid remainder; `document-amount` is the original amount and must not be used as the balance)
* `void-status` does not mark the document as voided
* `due-date` is at least the threshold number of days before today

Treat credit-type documents (credit memos, returns, payments) with a remaining amount as unapplied credit. Do not add them to the overdue total; list them per customer as credit that could be applied.

**Step 2: Age and rank**

Per `customer-id`, bucket the overdue `current-transaction-amount` by days past `due-date`: 30 or less, 31-60, 61-90, over 90. Rank customers by total overdue, then by the over-90 amount.

**Step 3: Customer standing**

For the top 15 customers, call `dynamics-gp-get-rm-customer` filtered by `customer-id`. Take `contact-person`, `phone-1`, `main-email-address`, `salesperson-id`, `payment-terms-id`, `customer-balance`, `credit-limit-amount` and `hold`. Flag customers whose balance exceeds their credit limit (ignore a limit of 0, which usually means none is set) and customers already on hold.

**Step 4: Payment habits**

For the same customers, call `dynamics-gp-get-rm-trx-apply` filtered by `customer-id` with `document-date.from` set to 180 days ago. For each application, days to pay = `document-date` (the payment) minus `applied-to-document-date` (the invoice). Report the average days to pay and any `writeoff-amount`.

**Step 5: Report**

* A table: customer name, customer id, total overdue, the four aging buckets, unapplied credit, credit limit, over limit (yes/no), on hold (yes/no), average days to pay, contact, phone, email, salesperson.
* The five documents with the largest overdue remainders: document number, type, due date, days overdue, remaining amount, customer PO number.
* One line of recommendation per top-five customer: call, apply the credit, escalate to the salesperson, or consider a credit hold (the separate `place-customer-on-credit-hold` prompt performs that).

**Rules & Guidelines:**

* This prompt is read-only. Do not call `dynamics-gp-rm-customer-update`, `dynamics-gp-rm-customer-note`, `dynamics-gp-rm-trx-add`, `dynamics-gp-rm-trx-note` or any other tool whose name does not contain `get`.
* Amounts are in the document currency as returned. Do not convert them.
* Use `document-type` for the document type; do not guess it from the document number.

**Validation:**

* If nothing is overdue past the threshold, say so and give the total open (not overdue) balance instead.
* If `{customer-name}` matches several customers, include them all and say which names matched.