A Claude-driven template for turning scanned Catholic Latin (or Greek) texts into English — and publishing them as a clean website. You mostly just talk to Claude; it does the technical work and even helps you install what's missing.
- Claude Code installed. (This is the one thing Claude can't install for you — you need it to talk to Claude.)
That's it. Claude helps with everything else (PDF tools, image tools, GitHub).
- Make your own copy. On this repo's GitHub page, click the green Use this template button → it creates a fresh copy under your account. Download or clone it to your computer.
- Add your PDF. Put your scanned Latin/Greek PDF into the
source/folder. - Open the folder in Claude Code and say "let's get started."
From there Claude will:
- open a setup dashboard (a web page showing what's installed) and help install anything missing,
- help you pick how to read the scans,
- then walk you through: read the pages → translate to English → build the web pages → proofread side-by-side with the original scans → publish.
- Proofread. Claude opens a viewer where the generated text sits next to the original scan. Click a paragraph to see its page; click the flag (⚑) to mark anything wrong — Claude fixes the flagged spots for you.
- Publish (optional). Say "let's publish it" and Claude sets up a free GitHub Pages website with you.
- It's always safe to run
/compactwhen Claude suggests it — everything is saved to disk, so nothing is lost. - You can hold several books in one project; each lives under
work/<name>/.
Plain Python helper scripts under scripts/ do the mechanical steps (PDF → images, image cropping, HTML assembly). The judgment steps (reading, translating, formatting) are done by Claude. Two small local web apps (the setup dashboard and the proofreading viewer) are stdlib-only — no heavy setup. The guidance Claude follows lives in CLAUDE.md and .claude/skills/.