Skip to main content
Send every monetary amount as a decimal string ("100.00", "-19.00", "0.19"). There is no wire-level cap on fractional digits.

​
tax_amount consistency

For each line_items[].taxes[] on a sale / return / exchange, OpenFiskal recomputes the expected tax_amount from the line’s total_amount (gross) and the rates on the same line, and rejects with 422 tax_amount_precision_invalid if it disagrees. 
expected_tax_i = total_amount × rate_i / (1 + Σ rate_j)
The divisor sums every rate on the line. Single-tax lines collapse to gross × rate / (1 + rate); multi-tax stacking (e.g. US state + municipal) checks every row against the full rate set. Comparison is strict at 8 decimal places: expected is rounded half-up to 8 dp and compared by numeric equality. Submit tax_amount rounded to 8 dp.
total_amountrateSubmitWhy
"119.00""0.19""19.00"Math is exact (19.00000000); trailing zeros are equal.
"7.00""0.19""1.11764706"7 × 0.19 / 1.19 = 1.117647058… half-up at 8 dp. Truncating to "1.11764705" is rejected.
"107.50""0.05" + "0.025" stacked"5.00" and "2.50"107.50 / (1 + 0.05 + 0.025) = 100 net, then net × rate_i.
The operation-level identity pretax_amount + tax_amount + tip_amount = total_amount is enforced separately and exactly.