financeiro-ops

Use a minimal schema of client, document, due date, amount, status, payment method, and notes, with statuses like open, overdue, paid, partial, or cancelled. Normalize dates to YYYY-MM-DD and values to decimal BRL before grouping totals by status or client.

Load the file with pandas, normalize dates and amounts, then group by status or category to validate row counts and sums. Deduplicate entries using a composite key of client or supplier, document, due date, and value.

No. It only drafts collection messages for human review and confirmation before any external send, and it never executes payments. Actual payment always requires human action in the bank or ERP system.

Identifiers such as CPF, CNPJ, bank details, PIX keys, and barcodes are masked to partial forms like ***1234 before appearing in any output. Full documents, statements, and credentials are never published to group chats or reports.

Totals fail when values like R$ 1.234,56 are treated as text instead of numbers. Normalize Brazilian currency strings to decimal format before summing, and confirm whether the user wants accrual (competência) or due-date (vencimento) views.