Question: pdf skill different from the one used in ChatGPT

[mistyreed63849]

Hi OpenAI team,

While using the ChatGPT web app, I noticed that the displayed reasoning trace for GPT-5.4-thinking appears to reference a pdfs skill whose SKILL.md starts with:

# PDF Skill (Read • Inspect • Extract • Edit • Render • Forms • OCR • Redact • Convert • Diff)
This skill is designed for reliable, workflow-driven PDF work: render -> verify -> operate -> re-render verify.

This version seems substantially more detailed and complex than the PDF skill currently published in this repo:

https://github.com/openai/skills/tree/main/skills/.curated/pdf

I'm wondering if this is a pending update or an intentional difference between the skill used in ChatGPT and Codex?

Thanks for the clarification!