Turn screen time into learn time.
Screen time conflict. Outsourced.

AI tutors that quiz your kids via WhatsApp, Telegram, Signal, Discord, or any channel OpenClaw supports — and reward correct answers with screen time credits.

🌐 clawtutor.ai · 📦 GitHub · 🏠 Real setup · 📝 Blog · 📧 Contact · 📸 Instagram · ▶️ YouTube

Built by a father who was tired of the daily screen time negotiations with his two kids.

The Problem: Like many parents, I was caught in an endless negotiation loop. "Can I watch TV?" "Did you do your homework?" "But I already did math yesterday!" The struggle was real — and exhausting.

The Insight: Kids are motivated by screen time. What if we could channel that motivation into learning? Instead of limiting screen time as punishment, what if they could earn it?

The Solution: ClawTutor — AI tutors running on a Raspberry Pi that:

• Quiz the kids via WhatsApp, Telegram, Signal, or Discord — wherever they already are
• Reward correct answers with screen time credits
• Actually make them want to study (yes, really)
• Track progress automatically per topic

The Result: My kids now text their tutors voluntarily. They compete for streaks. They ask for extra quizzes. The screen time war is over — and learning won.

Device Control supports any device whose DNS points at Pi-hole:

• 📺 Streaming boxes (Apple TV, Fire TV, Chromecast, Android TV, Roku)
• 🎮 Gaming consoles (PlayStation, Xbox, Nintendo Switch) — WiFi or wired
• 📱 Phones & tablets — blocked on home WiFi
• 💻 Computers (laptops, PCs)
• 🔌 Smart plugs & home automation (Home Assistant, Shelly, TP-Link Kasa, IKEA Trådfri) — cuts power for offline/local games that bypass DNS

ClawTutor isn't just for parents. Anyone who wants to gamify learning and control their own screen time:

Don't have kids? Use it on yourself:

• Set up a tutor for your learning goals (language, coding, certifications)
• Quiz yourself or let the AI quiz you
• Earn credits toward your own devices
• Block your own PlayStation until you've done your Duolingo 😅

Compare to human tutoring at €30-50/hour 😉

┌─────────────────────────────────────────────────────────────┐
│                     Raspberry Pi 5                          │
│                                                             │
│  ┌────────────────────────────────────────────────────────┐ │
│  │                  OpenClaw Gateway                      │ │
│  │                                                        │ │
│  │   ┌─────────┐    ┌─────────┐    ┌─────────┐            │ │
│  │   │  Main   │    │  Tutor  │    │  Tutor  │            │ │
│  │   │ (Parent)│    │ (Kid 1) │    │ (Kid 2) │            │ │
│  │   │Opus 4.5 │    │Sonnet 4 │    │Sonnet 4 │            │ │
│  │   └────┬────┘    └────┬────┘    └────┬────┘            │ │
│  │        │              │              │                 │ │
│  │   WhatsApp 1     WhatsApp 2     WhatsApp 2             │ │
│  │   (personal)     (tutor)        (tutor)                │ │
│  └────────────────────────────────────────────────────────┘ │
│                              │                              │
│                    ┌─────────┴─────────┐                    │
│                    │  SQLite Database  │                    │
│                    │  (screen time     │                    │
│                    │   credits)        │                    │
│                    └───────────────────┘                    │
└─────────────────────────────────────────────────────────────┘

How it works:

1. Kid texts tutor on WhatsApp
2. Tutor quizzes them on current school topics
3. Correct answers → credits added to balance
4. Kid checks balance anytime ("How many credits do I have?")
5. Parent approves screen time based on balance

The secret sauce that makes kids want to study:

Planned — Chore Credits: Kids will be able to earn screen time by completing household tasks (empty the dishwasher, brush teeth, clean their room, etc.). A parent confirms the chore via WhatsApp chat, and the credits are added instantly. Each chore maps to a configurable credit value.

Kids earn credits. How you "spend" them is up to you:

• Manual: Kid asks, you check balance, you decide
• Honor system: Kid manages their own screen time
• Automated: Use the Device Control Add-on to enforce limits

When a test is coming up, ClawTutor shifts into high gear:

1. Register the exam in CALENDAR.md:

   ## Upcoming Exams
   - 2024-03-15: Math Test - Chapter 7 (Equations)
   - 2024-03-22: English Vocabulary Test

2. Tutor activates countdown mode (configurable: 14/7/3/1 days):

   • "📢 Maths test in 14 days — time for a first review!"
   • Increasing urgency as exam approaches
   • Focus shifts to exam-relevant topics

3. Intensified practice:

   • More questions on weak spots (tracked in PROGRESS.md)
   • Review of previously failed topics
   • Bonus credits for exam prep sessions

4. Progress visibility:

   • Topic mastery: 🔴 → 🟡 → 🟢 → ⭐
   • "You've got 3 out of 5 topics to green — 2 more to practise!"

Tutor: "⚠️ Maths test in 3 days!
        Topic: Linear equations
        
        Your progress:
        ✅ Setting up equations — ⭐
        ✅ Solving equations — 🟢
        ⚠️ Word problems — 🟡
        ❌ Inequalities — 🔴
        
        Shall we practise word problems?"

Kid: "ok"

Tutor: "A train leaves at 8:00 am travelling at 120 km/h..."

Works the same for university exams, certifications, or any test:

• Add your exam dates
• Let the AI quiz you on weak areas
• Track your progress toward exam readiness
• Reward yourself with guilt-free breaks when you're prepared

TODO: Add screenshots of WhatsApp chat and iOS app

One command installs ClawTutor and walks you through setup interactively:

# 1. Install OpenClaw (if not already)
npm install -g openclaw
openclaw onboard

# 2. Install ClawTutor and run the setup wizard
npm install -g clawtutor
clawtutor setup

The wizard asks for your children's names, grades, and messaging channel — then creates all workspaces, config, and the screen time database automatically.

Already running OpenClaw? Install and run directly from chat with your main agent:

/skill install clawtutor

The tutor agent will walk you through setup conversationally — no terminal needed.

npm install -g openclaw && openclaw onboard
git clone https://github.com/groonvandorp/ClawTutor.git
cat ClawTutor/SETUP.md

Requirements:

• Raspberry Pi 5 with 4 GB+ RAM (or any Linux machine with Node.js 22+)
• OpenClaw installed
• Anthropic API key
• A messaging channel for the tutor (WhatsApp, Telegram, Signal, or Discord)

ClawTutor/
├── README.md              # You are here
├── SETUP.md               # Step-by-step installation guide
├── LICENSE                # License (free non-commercial, commercial requires permission)
│
├── cli/                   # npm package — `npm install -g clawtutor`
│   ├── package.json
│   ├── bin/clawtutor.js   # Interactive setup wizard (no external deps)
│   └── scripts/           # Scripts bundled with the npm package
│
├── skill/                 # OpenClaw skill — `/skill install clawtutor`
│   └── SKILL.md           # Chat-based setup for existing OpenClaw users
│
├── addons/                # Optional add-ons
│   ├── DEVICE-CONTROL.md  # Pi-hole device control (TVs, gaming, mobile, etc.)
│   └── DASHBOARD.md       # SPI status display
│
├── scripts/               # Shell scripts (source — also bundled in cli/)
│   ├── screentime-db.py   # Credit balance management
│   ├── appletv-access.sh  # (Add-on) Pi-hole device control
│   ├── tv-session.sh      # (Add-on) Session management
│   └── tv-auto-off.sh     # (Add-on) Cron auto-off
│
├── artwork/               # Logo and icon source files
│   └── *.png              # Originals from Gemini image generation
│
├── docs/                  # GitHub Pages website (clawtutor.ai)
│   ├── index.html         # Main landing page
│   ├── olivers-setup.html # Real-world setup walkthrough
│   ├── courses.html       # Available courses (Math, English, Music)
│   ├── contact.html       # Contact form (via Formsubmit)
│   ├── ecosystem.html     # Projects ClawTutor is built on
│   ├── blog.html          # Dev log & updates index
│   ├── blog/              # Individual blog posts
│   ├── icon.png           # App icon (transparent background)
│   └── logo.png           # Horizontal wordmark (transparent background)
│
├── examples/              # Example workspace files
│   ├── openclaw-example.jsonc   # Channel config (language-neutral)
│   ├── CALENDAR.md              # Exam tracking (language-neutral)
│   ├── de/                      # 🇩🇪 German examples
│   │   ├── SOUL.md
│   │   ├── CURRICULUM.md
│   │   ├── PROGRESS.md
│   │   └── SCREENTIME.md
│   └── en/                      # 🇬🇧 English examples
│       ├── SOUL.md
│       ├── CURRICULUM.md
│       ├── CURRICULUM-ENGLISH.md
│       ├── PROGRESS.md
│       └── SCREENTIME.md
│
└── dashboard/             # (Add-on) Mission Control display
    └── mission_control.py

• Isolated agents — Each kid's tutor has no access to other workspaces
• Tool restrictions — Tutors can only read/write their curriculum files
• No shell access — Kids can't execute commands
• Prompt injection defense — SOUL.md contains jailbreak countermeasures
• Allowlist — Only registered phone numbers can chat with tutors

ClawTutor is a self-hosted system. Running it safely requires solid IT and IT-security knowledge.

This is not a plug-and-play consumer product. Operating ClawTutor responsibly requires:

• Linux / Raspberry Pi OS — comfortable with the terminal, SSH, systemd services, and file permissions
• Networking basics — understanding of DNS, firewall rules, local network configuration, and VPN (if you expose the gateway remotely)
• IT security awareness — you are connecting a messaging channel (WhatsApp, Telegram, etc.) to an AI with access to your local system. You need to understand what that means and configure it accordingly
• Dependency awareness — the stack includes OpenClaw, Node.js, Python, Pi-hole (optional), and Anthropic's API. You are responsible for understanding what each component does and does not do

ClawTutor is provided as-is, without any warranty. The authors accept no liability for:

• Data loss, system compromise, or unexpected AI behaviour
• Costs incurred through Anthropic API usage
• Any consequences arising from misconfiguration or security gaps in your setup

A self-hosted system is only as secure as its last update. You are responsible for:

Why Raspberry Pi? Running ClawTutor on a dedicated Raspberry Pi 5 (4 GB+) keeps it isolated from your main computers. If something goes wrong — a dependency vulnerability, a misconfiguration, an unexpected AI action — the blast radius is contained to a €80 device that can be wiped and re-flashed in minutes. It is also always on, low power, and easy to physically secure at home.

If you are not comfortable managing a Linux server and its security, consider waiting for a managed hosted version, or ask someone with the relevant skills to help with the initial setup and ongoing maintenance.

→ Full details on the Built on page.

Free for personal, family, and non-commercial use — attribution required. Commercial use requires written permission. See LICENSE for details.

• My kids, for being the ultimate QA testers (and for trying very hard to jailbreak Matteo)
• The OpenClaw team for making AI agents accessible
• Every parent who understands the screen time struggle

Made with 🦞 by a parent who turned "Can I watch TV?" into "Can I do another quiz?"