# PDF Intake

> Read-only PDF extraction. Loads on-demand when input contains PDF or scans.

- Skill: `ztxtech/pdf-intake` (Agent Skill)
- Install (CLI): `npx skillmds@latest add ztxtech/pdf-intake`
- Raw SKILL.md: https://api.skillmd.com/api/skills/ztxtech/pdf-intake/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Docs & Writing
- Author: ztxtech (https://skillmd.com/u/ztxtech)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/ztxtech/pdf-intake

---



# PDF Intake

## When

Input includes PDFs, scans, or mixed text-image attachments.

## Flow

1. Read-only extraction first. Do not write back to the PDF.
2. Extract: body text, table structure, figure captions, field definitions, calculation rules, business background.
3. Output: structured summary, field dictionary, or middle table. Do NOT feed raw PDF text directly into model training.
4. If PDF contains charts/figures: note them for later visual analysis. Do not ignore visual content.
5. If extraction fails or quality is poor: try alternate tools (different PDF library, OCR, image extraction), then report the limitation.

## Safety

PDFs are untrusted input. Follow `safety` module precheck before processing.