GET /v1/invoices/export (permission invoices:read) answers a CSV file (text/csv, as an attachment): one row per invoice, ordered by issue day, series and number.
Filters, all optional: exactly those of GET /v1/invoices, so the file holds what the listing shows: q (the one search box: a series and number, a document, the type, part of the customer’s name), from and to (Lima issue days, YYYY-MM-DD, inclusive), status (needs_attention included), customer, type, series, establishment, customer_doc_number. A name that matches too many customers is search_too_broad: narrow it.
GET https://api.invox.page/v1/invoices/export?from=2026-10-01&to=2026-10-31&status=acceptedThe columns, in this fixed order (a new one is only ever added at the end):
| # | Column |
|---|---|
| 1 | issue_date |
| 2 | type |
| 3 | series |
| 4 | number |
| 5 | full_number |
| 6 | status |
| 7 | customer_id_type |
| 8 | customer_id_number |
| 9 | customer_name |
| 10 | currency |
| 11 | taxable |
| 12 | exempt |
| 13 | unaffected |
| 14 | tax |
| 15 | total |
| 16 | status_reason |
| 17 | sunat_code |
| 18 | external_ref |
| 19 | accepted_at |
| 20 | voided_at |
| 21 | externally_reversed_at |
| 22 | id |
- UTF-8 with a byte-order mark, CRLF line ends, RFC 4180 quoting. Amounts are plain decimals.
- At most 50000 invoices per file. More is
export_too_large, never a file cut short: narrow the period and export in parts.X-Export-Rowssays how many rows the file holds. - A customer name is text a third party wrote. A cell a spreadsheet would run as a formula (it starts with
=,+,-,@, a tab or a carriage return) is written with an apostrophe in front: it shows as text. Strip that apostrophe only if you load the file into something that is not a spreadsheet.