Extract tables
~1 credits / page (up to 50)Detects tables in a PDF using geometry analysis and returns each one as an array of rows, ready to use as structured data. Works best on clean, ruled tables with clear borders. Optionally target specific pages to reduce cost when you know where the table lives.
Use when
You want tabular data from a clean, ruled table as structured rows.
Not for
Scanned tables, merged-cell or borderless layouts (those need OCR).
Cost
~1 credits / page (up to 50)
Billed at 0.1 credits per page scanned for tables.
Estimated; the actual charge depends on your input and is shown in the response.
What it accepts
Set these inside the intent when you run it.
URL of the PDF containing the table(s).
Optional 1-indexed page selector to target the page(s) with tables, e.g. "3" or "2-4".
What you get back
A list of detected tables, each as an array of row arrays, inline as JSON.
Run it
Run this sub-skill directly: pin it with operation and pass its inputs in the intent. (Omit operation and the PDF skill will route from your intent instead.)
curl -X POST "https://skill.askfaro.com/skills/pdf/run" \
-H "Authorization: Bearer $FARO_TOKEN" \
-H "Content-Type: application/json" \
-d '{"intent":{"operation":"extract_tables","source_pdf":"https://example.com/report.pdf","pages":"3"}}'Example requests
- ›Pull the table from page 3 of this PDF
- ›Extract all tables from this financial report
- ›Get the data out of the pricing table in this document