Installation

Prerequisites

• Docker (with Docker Compose v2)
• curl and git
• An LLM API key (Anthropic or OpenAI)

Step 1: Clone

git clone https://github.com/nunocoracao/mobyclaw.git
cd mobyclaw

Step 2: Interactive Setup

./mobyclaw init

The init wizard walks you through:

1. LLM Provider — Choose Anthropic or OpenAI, enter your API key
2. Messaging — Optionally connect Telegram (more platforms coming)
3. Service Credentials — Optionally add GitHub token, AWS keys, etc.
4. Workspaces — Optionally mount project folders
5. Agent Settings — Heartbeat interval, data directory

Minimal setup

You only need an API key. Everything else is optional. Press Enter through the defaults and you'll have a working agent.

Step 3: Start Moby

./mobyclaw up

This builds the Docker images (first time only), starts the containers, and Moby is live.

Verify It Works

# One-shot prompt
./mobyclaw run "Hello! Who are you?"

# Interactive chat
./mobyclaw chat

# Check service status
./mobyclaw status

Connect Telegram

1. Message @BotFather on Telegram
2. Create a new bot with /newbot
3. Copy the bot token
4. Add it to your .env:

   TELEGRAM_BOT_TOKEN=123456:ABC-DEF...
   

5. Restart: ./mobyclaw down && ./mobyclaw up
6. Message your bot on Telegram — Moby responds!

Updating

cd mobyclaw
git pull
./mobyclaw down
./mobyclaw up

The containers rebuild automatically when the source changes.

Uninstall

./mobyclaw down
# Optionally remove your data:
# rm -rf ~/.mobyclaw