How does a schedule workflow appear as linked tables in a reseller's catalogue?

A schedule workflow usually lands as four linked tables, which suits catalogue listings because each can be described, sampled and priced separately by you.

- `schedules`: `schedule_id`, `account_id`, `product_type`, `start_date`, `frequency`, `status`, `policy_id` - `instalments`: `instalment_id`, `schedule_id`, `due_date`, `amount_due`, `principal`, `interest`, `status` - `amendments`: `amendment_id`, `schedule_id`, `request_type`, `requested_by_role`, `approved_by_role`, `decision`, `effective_date` - `policies`: `policy_id`, `rule_text`, `threshold_value`, `approval_role`, `valid_from`

Foreign keys run from instalments and amendments back to schedules, and from schedules to policies. Keep the identifiers stable across files so that your customers can join them without guessing.

Delivery formats are typically JSONL for event-style records and Parquet for larger tables. A manifest file can list each file, its schema version and row count, so you can validate a drop before it reaches your marketplace.

For evaluation use, an additional `expected_outcomes` file could hold the correct post-amendment schedule for each task, keyed by `amendment_id`.

Ask for a schema document in a sample so your catalogue entries and your customers' loaders match what is delivered. All of it is generated material; keep that statement in the metadata you pass onward.