extract-numbers-from-pdf

---
name: extract-numbers-from-pdf
description: Download PDF reports (annual reports, filings) and put the figures into a spreadsheet with the page number for each figure. Use when numbers must be pulled from several PDFs and be checkable.
---

# Extract numbers from pdf

This job runs on the user's own MyClawn cloud computer: a real browser that stays signed in, its own email address and files. It can outlast the chat and comes back with the answer, a screenshot of the final screen and a link to a screen recording.

## Setup

The user must have the MyClawn Computer connector signed in. If it is not, follow the setup in the cloud-computer-task skill of this repository and let the user do the sign-in; do not create accounts for them.

## Task

Hand this task to the connector's task tool, in the user's own words:

> Download the latest annual reports of Andritz, Palfinger and Rosenbauer. Put revenue, operating result and headcount for the last two years into a spreadsheet with the page number you took each figure from.

Replace the companies and the figures with the user's own. Show the final wording to the user before sending anything that submits a form, books something or writes to a third party.

## Steps

1. Write the task with the goal, the sites, what the result must look like and what must not be done.
2. Send it with the connector's task tool and wait for the result.
3. Report the answer, the screenshot and the recording link.

## Limits

Tell the user to check the figures they will rely on; the page numbers make that quick. A captcha is the user's to solve on the live screen.

Full description of this job, with the steps and the limits: https://www.myclawn.com/extract-numbers-from-pdf-reports?utm_source=agent-skills&utm_medium=probe&utm_campaign=extract-numbers-from-pdf