LLM School Notes Template — AI skill for Claude Code

LLM school notes - wiki template

Photograph your notebook, and an AI agent turns it into clear, curated study pages.

This is a template for a knowledge base (an "LLM wiki") built from one student's school notes. You drop in photos of exercise-book pages, downloaded notes, or voice notes; an LLM agent - [Claude Code](https://code.claude.com/), [Codex](https://developers.openai.com/codex/), or any agent that can read `AGENTS.md` (e.g. Hermes Agent) - files them, reads them, and maintains a wiki: one directory per subject, one page per topic, plus a page per set of lesson notes. The storage format is the [Open Knowledge Format 0.2](SPEC.md): plain markdown with YAML frontmatter, readable on GitHub, in Obsidian, or with `cat`.

The template contains no names, school, subjects, textbooks, or notes - everything specific is set up by the LLM on the first run.

What the pages look like

• Faithful and curated at the same time. A lesson-notes page records exactly what the notebook says. The topic pages teach: they explain terse notes, redo worked problems step by step, fill gaps, and fix errors - every addition visibly marked as not from the notebook (colored callouts: 💡 explanation, ➕ addition, ⚠️ correction).
• Two layers. Every topic page opens with a ⚡ In short box for the evening before a test; chapters get a summary page; the full detail stays on the topic pages.
• 🧠 Test yourself questions with hidden answers at the end of each topic page.
• Plain language at the student's level (ISO 24495-1 principles), phone-friendly tables, LaTeX math, Mermaid diagrams, and colorful header banners (tools/banner.py).
• Catch-up support: notes copied from a classmate's notebook are marked 🤒 until the student has caught up; homework from the e-diary goes into a 📌 table.
• Textbook references: textbooks (converted to text) in references/ are used to check readings and cite printed page numbers - never copied.

Getting started

1. Click **Us