parse-schematic-pdf

Codex skill for turning schematic PDFs into Chinese, LLM-ready board briefs.

It accepts a local PDF path or a PDF URL, extracts the PDF text layer, optionally renders each page to PNG for visual review, and guides Codex to summarize board functions, power rails, peripherals, interfaces, firmware implications, risks, and evidence.

What It Produces

• pdf_text.md: extracted text grouped by PDF page.
• pdf_context.json: extraction quality, title candidates, parts, nets, pins, buses, warning lines, page summaries, and visual review targets.
• page_images/page_N.png: rendered pages when --render-pages is used and pypdfium2 is installed.
• visual_manifest.json: rendered image list and pages that should be visually reviewed.
• warnings.md: extraction or rendering warnings.

Install

Clone or copy this folder into Codex's skills directory:

Copy-Item -Recurse . "$env:USERPROFILE\.codex\skills\parse-schematic-pdf"

The skill folder must contain SKILL.md at its root.

Dependencies

Required:

python -m pip install pypdf

Recommended for visual page review:

python -m pip install pypdfium2

The Codex bundled Python runtime often already includes pypdf; install pypdfium2 if page rendering is needed.

Script Usage

python .\scripts\extract_schematic_pdf.py --input "schematic.pdf" --out ".\out" --render-pages

For URLs:

python .\scripts\extract_schematic_pdf.py --input "https://example.com/schematic.pdf" --out ".\out" --render-pages

Codex Usage

Ask Codex:

Use $parse-schematic-pdf to analyze this schematic PDF into a Chinese LLM-ready board brief.

When rendered page images are available, Codex should inspect the pages listed in visual_manifest.json before finalizing pin mappings, color channels, interface direction, and electrical warnings.

Scope

This skill is designed for functional peripheral-level understanding. It does not reconstruct a complete netlist from schematic geometry.