PDF Toolkit

PDF Toolkit

✓ Verified

Read, create, merge, split, and manipulate PDF files. Extract text and tables, fill forms, add watermarks, and OCR scanned documents.

Claude / Claude Codetested
GitHub Copilot
Cursor
VS Code
OpenAI Codex
Document CreationDeveloperTeacher / EducatorBusiness OwnerLegal Professional

What it does

PDF is a binary format. Claude can write Python code that claims to generate a PDF, but without knowing the specific library APIs and their edge cases, the code frequently produces malformed output, fails on non-ASCII characters, or generates files that open incorrectly in certain readers. This skill loads the correct patterns for reading, creating, merging, splitting, and manipulating PDFs — including text extraction, form filling, OCR for scanned documents, and watermarking — using libraries Claude can actually call.

Use case

Workflows that produce or consume PDF documents: generating client reports, extracting data from invoices, merging documents, or automating document pipelines. Made by Anthropic.

The Prompt

Copy and use immediately
"Read this invoice PDF and extract the line items as a structured JSON object."
"Generate a PDF report from this data with a header, summary, and appendix."
"Merge these three PDFs in order and add sequential page numbers."
"Extract all text from this scanned document using OCR."
"Add a CONFIDENTIAL watermark to every page of this document."

How to use

  1. 1

    Provide a file path for existing PDFs, or describe the document you want to create.

  2. 2

    For extraction: Claude reads the file and returns structured data — text, tables, or specific fields.

  3. 3

    For creation: describe content and layout. Claude generates the file, not just the code to generate it.

Input / Output

Input

A PDF file path, a description of the document you want to create, or structured data to format as a PDF.

Output

Extracted text or structured data from existing PDFs, or newly generated PDF files with correct formatting. Scanned documents processed via OCR return searchable text with confidence scores for ambiguous regions.

Install via CLI

npx skillsadd anthropics/skills/pdf
Added 15 Mar 2026Submitted by anthropics👁 45📋 0

Details

Platforms
Claude / Claude CodeGitHub CopilotCursorVS CodeOpenAI Codex
Category
Document Creation
License
apache-2.0
Author
Anthropic

Stats

📋 Copies0
👁 Views45
👍 Upvotes0

Install with skills.sh

npx skillsadd anthropics/skills/pdf

Requires skills.sh CLI

Community Notes

Sign in with GitHub to leave a note.

No notes yet. Be the first to contribute.