Invoice Parsing

Invoice Parsing

This skill extracts structured financial data from invoice documents. It identifies the vendor, enumerates line items, and reconciles the declared total against the computed sum so downstream automation can trust the numbers.

Overview

• Identify the vendor name, address, and invoice number from the header region.
• Extract every line item — description, quantity, unit price, and amount — into a table.
• Compute the subtotal, apply tax, and verify the printed grand total matches.

Steps

1. Read the uploaded document and detect whether it is a scanned image or native text.
2. Classify the invoice layout against the known vendor templates.
3. Extract each field into a normalized JSON structure with explicit types.
4. Validate arithmetic: the line-item sum plus tax must equal the stated total.

If a required field is missing, fall back to a clearly marked placeholder and flag the record for human review. When the layout is ambiguous or two templates match equally well, ask the user to confirm the vendor before continuing.

Output

Return a JSON object containing the vendor block, the ordered list of line items, the computed totals, and a confidence score for each extracted field.

from scripts.parse import parse_invoice

result = parse_invoice(document_text)

Run scripts/parse.py for the reference implementation. See [reference](reference.md) for the full field dictionary and [examples](examples.md) for worked input/output pairs.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: larsroettig
• Source: larsroettig/skillspector-quality
• License: MIT

Install and usage instructions live in the source repository linked above.