For content creators

AI-guided curriculum authoring

Drop one of these packages into your curriculum repository and your AI coding assistant will know the exact JSON+Markdown shapes, locale-overlay rules, and personal-data boundaries apprenti.dev expects — before it writes a single file.

Claude Code

A real Claude Code Skill — loads automatically when you're working in a curriculum repository, with deeper reference files for JSON shapes and validation.

.claude/skills/apprenti-curriculum-author/ Download for Claude Code

Cursor

A project rule scoped to curriculum files (task.json, term.json, resources, competencies, locale overlays).

.cursor/rules/apprenti-curriculum-author.mdc Download for Cursor

OpenAI Codex

An AGENTS.md Codex reads automatically at the start of every session in this repository.

AGENTS.md Download for OpenAI Codex

OpenCode

The same AGENTS.md convention OpenCode reads, with setup notes for opencode.json and /init.

AGENTS.md Download for OpenCode

GitHub Copilot

A repo-wide copilot-instructions.md plus a path-scoped instructions file for curriculum content specifically.

.github/copilot-instructions.md Download for GitHub Copilot

What's actually in these

Every package teaches the same facts, adapted to how each tool reads project instructions: the single- and multi-curriculum repository layouts, the exact JSON shapes for tasks, terms, resources, and competencies, the locale-overlay pattern, the AI policy shape, a pre-flight validation checklist (duplicate ids, ghost/orphan tasks, dangling references, bad overlays), and the product's voice guidance — so generated task text never drifts into "LMS" or gamified language.

They also encode hard boundaries every package agrees on: never touch an apprentice's or mentor's personal-data folders, never invent an id from a file path, and never commit or publish anything without a human deciding to.

ChatGPT (Custom GPT)

Plain ChatGPT can't discover a file in your repository the way the five tools above do — there's no equivalent mechanism. This package is a template for your own personal Custom GPT instead: an Instructions file plus three Knowledge files you paste and upload once. It works the same way pedagogically, but you'll paste or upload the specific file you want help with into each chat yourself, since this Custom GPT has no direct access to your repository.

Paste into a Custom GPT's Instructions + Knowledge fields Download for ChatGPT

Want the full walkthrough?

Installation steps per tool, what to expect the first time you open a curriculum repo, and how to keep a package updated.