# pdfvision

> Give AI agents human-like PDF vision

Turn silent failures in PDF extraction into errors agents can recover from. On its own, extraction reports empty scans, glyph garbage, and scrambled columns as success — pdfvision flags them page by page and names the next step: render the page, run OCR, or crop to the region that matters.

## Table of Contents

- [Agent Skills](https://pdfvision.dev/guide/agent-skill.md): Install and use the bundled pdfvision Agent Skills for Claude Code, Codex, Cursor, and other skill-aware agents.
- [Command Line Options](https://pdfvision.dev/guide/command-line-options.md): Complete pdfvision CLI option reference for PDF input, output formats, rendering, OCR, search, layout, metadata, and cache behavior.
- [FAQ](https://pdfvision.dev/guide/faq.md): Common pdfvision questions about empty text, scans, layout, OCR, rendering, cache behavior, and passwords.
- [Getting Started](https://pdfvision.dev/guide.md): Start using pdfvision to extract PDF text, layout, rendered pages, and OCR evidence for AI agents.
- [Installation](https://pdfvision.dev/guide/installation.md): Install or run pdfvision with npx, npm, or as a library dependency.
- [Layout and Warnings](https://pdfvision.dev/guide/layout-and-warnings.md): Understand pdfvision layout reconstruction, visual regions, geometry, and page warnings.
- [Library API](https://pdfvision.dev/guide/library-api.md): Use pdfvision from TypeScript or JavaScript with processDocument and processFile.
- [MCP Server](https://pdfvision.dev/guide/mcp-server.md): Serve pdfvision over the Model Context Protocol for hosts without a shell — Claude Desktop, Cursor, Cline, Zed, and workflow tools like n8n.
- [Output Formats](https://pdfvision.dev/guide/output.md): Choose between pdfvision Markdown, JSON, XML, and TOON output.
- [Prompt Examples](https://pdfvision.dev/guide/prompt-examples.md): Prompt templates for using pdfvision output with AI agents to inspect PDFs, verify layout, extract tables, read scans, and analyze forms.
- [Rendering and OCR](https://pdfvision.dev/guide/rendering-and-ocr.md): Render full pages, render visual regions, and OCR scanned PDF pages with pdfvision.
- [Search and Region Zoom](https://pdfvision.dev/guide/search-and-region-zoom.md): Use pdfvision to search PDF text, form fields, annotations, and OCR output, then render exact matching regions as PNG crops for AI vision models.
- [Security and Privacy](https://pdfvision.dev/guide/security-and-privacy.md): Understand how pdfvision handles local files, remote PDFs, passwords, cache directories, OCR traineddata, attachments, JavaScript actions, and sensitive output review.
- [Structured Output](https://pdfvision.dev/guide/structured-output.md): Understand pdfvision DocumentResult, PageResult, overview, quality, layout, OCR, warnings, coordinates, and optional PDF feature fields.
- [Usage](https://pdfvision.dev/guide/usage.md): Common pdfvision workflows for local PDFs, remote PDFs, page ranges, rendering, layout, OCR, and encrypted documents.
- [Use Cases](https://pdfvision.dev/guide/use-cases.md): Practical pdfvision workflows for AI agents reading research papers, slide decks, government forms, scanned PDFs, reports, tables, charts, and multilingual documents.