playbooks

home / skills / phodal / auto-dev / pdf

pdf skill

/examples/skills/pdf

This skill helps you extract tables, summarize content, and analyze PDF metadata to reveal structured insights.

npx playbooks add skill phodal/auto-dev --skill pdf

Review the files below or copy the command above to add this skill to your agents.

Files (1)
SKILL.md
541 B
---
name: pdf
description: Extract and analyze information from PDF documents
---

# PDF Document Analysis

You are an expert at extracting and analyzing information from PDF documents.

## Task

$ARGUMENTS

## Instructions

1. Understand the request carefully
2. Extract the requested information accurately
3. Present in a clear, structured format
4. Verify completeness

## Common Tasks

- Extract tables preserving structure
- Summarize document content
- Extract specific data points
- Analyze document metadata

Project: $PROJECT_NAME

Overview

This skill extracts and analyzes information from PDF documents to produce clean, actionable outputs. It supports table extraction with preserved structure, content summarization, specific data point retrieval, and metadata analysis. Designed for accuracy and clear, structured presentation of results.

How this skill works

The skill ingests a PDF and parses its layout, text blocks, tables, images, and metadata. It applies layout-aware extraction to preserve table rows/columns and uses NLP techniques to summarize content or locate requested data points. Outputs are validated for completeness and returned in structured formats (tables, JSON, or plain text) suitable for downstream processing.

When to use it

  • You need structured tables extracted from complex PDFs while preserving rows and columns.
  • You want concise summaries of long reports, contracts, or research papers.
  • You must extract specific data points (dates, amounts, IDs) from invoices, forms, or statements.
  • You need document metadata (author, creation date, software) for auditing or cataloging.
  • You require data-ready outputs for ingestion into databases or analytics pipelines.

Best practices

  • Provide clear extraction instructions and examples of the desired output format.
  • Supply sample pages or indicate which sections/pages to prioritize when PDFs are long.
  • Specify table header rows and expected columns when available to improve table fidelity.
  • Validate extracted numeric values and dates using a quick rule-set or regex patterns.
  • Request output format explicitly (CSV, JSON, markdown table) to avoid post-processing.

Example use cases

  • Extract invoice line items into CSV with columns: item, qty, unit price, total.
  • Summarize a 50-page technical report into a one-page executive summary with key findings and recommendations.
  • Pull contract metadata and key clauses (term, termination date, renewal conditions) into a checklist.
  • Convert scanned form data into structured JSON after OCR and field mapping.
  • Aggregate financial tables across quarterly PDFs into a single normalized dataset for analysis.

FAQ

Can the skill handle scanned PDFs or images?

Yes. Scanned pages require OCR. Provide high-quality scans for best accuracy; specify OCR language if not English.

How are extracted tables returned?

Tables can be returned as CSV, JSON arrays, or markdown-style tables. Indicate preferred format when requesting extraction.