# OCR

> Skill `ocr` on Faro. 0 sub-skills.

Extracts text from scanned PDFs and images of documents using optical character recognition. Returns the recognized text as layout-aware markdown, preserving reading order and tables. Works on photographs of documents, scanned reports, and image-based PDFs.

**Category:** Documents  
**Tags:** ocr, text-extraction, scanned-pdf, markdown, tables  
**Use when:** You have a scanned PDF or a photo of a document with no selectable text layer and need its contents as text.  
**Not for:** PDFs that already have a selectable text layer, answering questions about a document, describing photos, or translating.  
**Returns:** information — Returns the recognized text as markdown per page, in reading order, with tables preserved. Character styling like bold or italic is not preserved.

## How to run
Skills run through one gateway with your Faro token. Hand it an `intent` in plain language; Faro routes to the right sub-skill, runs it, and bills per call. Raw tools are internal plumbing and are not directly callable.

```
POST https://skill.askfaro.com/skills/ocr/run
Authorization: Bearer faro_<your_key>
Content-Type: application/json

{"intent":{"prompt":"Extract the text from this scanned PDF"}}
```

Or from the CLI:

```bash
pip install askfaro-cli && askfaro auth login
askfaro run ocr "Extract the text from this scanned PDF"
```

Full run reference: https://askfaro.com/llms/run.md — Agent recipe: https://askfaro.com/llms/skill.md

## Example requests

- Extract the text from this scanned PDF
- Read the text in this photo of a receipt
- OCR this image of a contract and return the text

## What it can do

- Extract the text of a scanned PDF as layout-aware markdown, page by page.
- Extract the text of a standalone photo or image of a document.
- Reconstruct the tables in a scanned report as markdown tables.
- Read only specific pages of a long document to keep the cost down.

---
On the web: https://askfaro.com/search/ocr