---
title: "pdf"
description: "Create, extract, combine, fill and visually review PDF documents in the Xpert sandbox."
canonical: https://agentpluginsdirectory.com/plugins/pdf
last-updated: 2026-10-10
---

# pdf
Create, extract, combine, fill and visually review PDF documents in the Xpert sandbox.
- Slug: pdf
- Publisher: Xpert AI
- Repository: https://github.com/xpert-ai/xpert-plugins
- Manifest: agent-plugins/pdf/plugin.json
- Version: 1.0.2
- License: AGPL-3.0-only
- Category (editorial): other
- Skills: 1 (pdf)
- MCP servers: 0
- Stars: 8
- Repository created: 2025-09-19
- Repository last pushed: 2026-10-10
- Publisher type: Organization
- Listing: https://agentpluginsdirectory.com/plugins/pdf
- Schema: https://agent-plugins.org/schemas/1.0.0/plugin.schema.json

## What pdf does, in the publisher's words

An independently authored Agent Plugins 1.0.0 package for PDF generation, text/table extraction, rendering, page merge/selection and ordinary AcroForms. It contains no Codex runtime code, telemetry helper or prompt text. The plugin's display icon is reused from Codex; see icon provenance and implementation provenance.

Requires an enabled interactive Assistant sandbox, the installed @xpert-ai/plugin-view-image provider and an image-capable model. Selecting PDF adds SandboxShell, SandboxFile and ViewImageMiddleware for that run. It does not enable a disabled sandbox or require an MCP/OAuth connection.

In the Xpert desktop source checkout run:

- ReportLab creation with embedded Chinese TrueType font; PDFium page rendering.
- pdfplumber text/table extraction with explicit truncation flags.
- pypdf merge and page selection/reordering for ordinary unsigned PDFs.
- Interactive ASCII text fields and checkboxes, including shared Parent/Kids widgets across pages; optional explicit flattening with structural verification.

From the project README, punctuation lightly normalized. Full text: https://raw.githubusercontent.com/xpert-ai/xpert-plugins/HEAD/agent-plugins/pdf/README.md