Tov-learn

Interactive AI tutor, built as a Claude Code skill.

Built and maintained by TovTech

Getting Started

Step 1 — Prerequisites

• Claude Code installed (Pro plan or higher)
• Git

Step 2 — Clone and open

git clone https://github.com/TovTechOrg/Tov-learn.git

Open the cloned folder in Claude Code.

Step 3 — Run setup

/learn setup

This will:

• Ask for your preferred session language (Hebrew / English)
• Ask for your course content folder path
• Optionally configure a TTS voice
• Install /learn globally — from this point on, /learn works in any project on your machine

That's it. Type /learn 0.1 to start the first lesson.

Usage

/learn setup        — configure language, TTS, and course path (run once)
/learn              — smart resume: shows what to do next based on your progress
/learn 0.1          — start lesson 0.1 directly
/learn status       — generate a visual progress dashboard (HTML)

Choosing a lesson

Lessons are numbered by module and index — 0.1, 0.2, 1.1, etc. Browse available lessons in courses/ai-engineer/COURSE.md, which lists all modules and their topics.

To start a lesson:

/learn 0.1

The tutor loads the script, greets you, asks what you already know, and walks through each section interactively. At any point you can type quiz me to test yourself, or stop to end the session and get a next-step recommendation.

Analyzing your own project

If you want the tutor to help you understand how a concept applies to a project you're building, open Claude Code inside that project and run:

/learn

Choose project analysis from the menu. The tutor will:

1. Silently scan your codebase (package.json, folder structure, config files)
2. Ask you 5 short questions about the project's purpose, users, and pain points
3. Generate a visual architecture map saved as an HTML file you can open in a browser

This is useful before starting a lesson — the tutor uses your project as a concrete example throughout the session.

Smart Resume — /learn

Typing /learn with no arguments doesn't just ask "what do you want to do?" — it reads your actual progress and tells you:

• Which lessons are overdue for review (spaced repetition)
• Which lessons are in progress
• What the next new lesson is

Example output after a few sessions:

Welcome back. Here's where things stand:

🔁 Due for review: Lesson 0.2 — What is ML (3 days overdue)
📖 Continue: Lesson 1.1 — Prompt Engineering Basics (in progress)
📖 Next new: Lesson 1.2 — Few-shot and Chain of Thought
🔍 Project analysis

What would you like to do?

Progress Dashboard — /learn status

Generates an HTML file at ~/skill-tutor-tutorials/dashboard.html showing all lessons color-coded by status (mastered / in progress / not started), quiz scores, and what's due for review. Open it in a browser.

Commands during a session

What the Tutor Does

Every section is taught using the Journey Format:

1. The problem — why this matters
2. The insight — what experts understand that beginners don't
3. In your project — connects the concept to your actual work
4. Question — one thinking question before moving on

After quizzes, the tutor tracks your score and tells you when to review the lesson again (spaced repetition: 2 / 13 / 34 / 89 days based on score).

Course Content Structure

Lessons live in courses/[course-name]/lessons/:

courses/
  ai-engineer/
    COURSE.md
    lessons/
      00-ai-fundamentals/
        0.1-intro-to-ai/
          0.1_script.txt       ← lesson script (split by [מעבר שקף])
          0.1_exercises.md     ← exercises
      01-prompt-engineering/
        ...

The tutor auto-detects lesson files by number — /learn 0.1 finds 0.1_script.txt automatically.

Files Created at Runtime

All learner data is saved to ~/skill-tutor-tutorials/ (outside the repo):

~/skill-tutor-tutorials/
├── settings.json           — language, TTS config, course path
├── learner_profile.md      — background, current project, learning style
├── tutorials/              — per-lesson notes, key insights, Q&A
├── progress/               — quiz scores and next review dates
├── topics/knowledge_map.md — full map of mastered vs. in-progress topics
└── architectures/          — HTML architecture diagrams (from project analysis)

Project Structure

.claude/commands/
  learn.md                  ← entry point + routing
  learn/
    setup.md                ← first-run configuration
    teaching.md             ← lesson loop + Journey format
    quiz.md                 ← quiz format + spaced repetition
    progress.md             ← saving tutorials and knowledge map
    project-analysis.md     ← codebase scan + architecture map
    display.md              ← visual formatting conventions
courses/
  ai-engineer/              ← course content
CLAUDE.md                   ← architecture overview for contributors

Adding a New Module

1. Create .claude/commands/learn/[module-name].md
2. Add a row to the modules table in CLAUDE.md
3. Add a routing entry in learn.md (Step 2 — Route table)
4. Update the global install command in setup.md

Requirements

• Claude Code (Pro plan or higher)
• Windows (for TTS voice support) — TTS can be disabled on any OS

License

MIT