Claude PDF skill

When you hand Claude a PDF and ask for the numbers in its tables, or ask for a finished report as a PDF, the work goes through Anthropic's PDF skill. Its ID is pdf, and it is one of four pre-built document skills, along with Word, Excel and PowerPoint. This page covers what it does, how to use it on each surface and where it stops. For how skills load in general, see how Claude Skills work.

What the skill does

The skill's description covers almost any PDF job: reading text and tables, merging and splitting files, rotating pages, adding watermarks, creating new PDFs, filling forms, encrypting and decrypting, extracting images, and running OCR on scanned pages so they become searchable. The SKILL.md maps each job to a tool Claude runs in its sandbox:

• pdfplumber for text and tables, with layout kept.
• pypdf and qpdf for merging, splitting and rotating.
• reportlab for building new PDFs page by page.
• pytesseract for OCR, after converting pages to images.

Form filling has its own guide. For a fillable form, Claude reads the field list, writes the values and fills them with a script. For a flat form with no fields, it first extracts the labels and lines with their coordinates, and falls back to estimating positions from page images when the PDF is a scan.

When to use it

• Pulling tables out of a report or invoice into rows you can work with. Anthropic's help center gives the example of extracting all the data from a PDF into an Excel sheet with a summary chart, which brings in the Excel skill as well.
• Combining several PDFs into one, or splitting a long one by section.
• Filling a standard form from data you paste in.
• Producing a short formatted report as a PDF rather than a Word file.

If you need an editable document, ask for the Word skill instead, and for slides, the PowerPoint skill. Our productivity prompts help you summarize what you extract.

How to install it

Claude.ai: nothing to install. Turn on "Code execution and file creation" under Settings > Capabilities (on Team and Enterprise, under organization settings). The help center lists file creation on Free, Pro, Max, Team and Enterprise plans.

Claude API: add pdf as a skill in the container parameter together with the code execution tool.

Claude Code: the platform docs say the pre-built document skills are not available in Claude Code. Anthropic publishes source copies you can install as a plugin:

/plugin marketplace add anthropics/skills
/plugin install document-skills@anthropic-agent-skills

Example request

Here is [file name].pdf. Extract every table on pages [page range] into one spreadsheet, one sheet per table, and keep the column headers as they appear. Then list anything you could not read cleanly.

Limits

• Files can be up to 30MB for uploads and downloads. The help center says larger PDFs can still be processed in Claude's computing environment without loading them into the conversation.
• On the API, skills run with no network access and only the preinstalled packages.
• Coordinates on flat forms are estimated, so check a filled form before you send it.
• The repository copy is source-available, not open source, and Anthropic notes that the behavior you get in Claude may differ from it.
• File creation counts against your plan's usage limits, and the help center says it uses more than a normal chat.

FAQ

Can Claude read a scanned PDF?

Yes. The skill converts the pages to images and runs OCR on them.

Is the PDF skill free?

It comes with Claude at no extra charge, and the help center lists file creation on the Free plan. Usage counts against your plan's limits.