Moltbot Wizard Setup Guide: Build Your Own Private AI Assistant

Imagine mentioning your AI in a group chat and having it book flights, check the weather, write code, and manage your schedule — all while remembering every conversation from last week. That is exactly what Moltbot delivers: a fully self-hosted, private AI assistant you actually own.

This guide walks you through the Moltbot Wizard — the interactive CLI that turns a complex multi-step setup into a few simple prompts.

Why Moltbot?

The Problem with Existing AI Assistants

We all use AI assistants daily — ChatGPT, Claude, Gemini. But they share a fundamental limitation: you are a tenant, not the owner.

• No data ownership: Your conversations and preferences live on someone else’s servers
• Limited integrations: Want AI in your WeChat group? Not happening
• Platform fragmentation: Slack for work, Telegram for personal — each needs separate setup
• No customization: The AI’s personality and capabilities are entirely platform-controlled

How Moltbot Solves This

Moltbot (formerly Clawdbot, renamed due to trademark issues) is an open-source project created by PSPDFKit founder Peter Steinberger, with over 68,000 stars on GitHub.

Its core philosophy: give you full control over your AI assistant.

┌─────────────────────────────────────────────────────────────┐
│                       Your Device                            │
│  ┌─────────────────────────────────────────────────────┐   │
│  │                   Moltbot Gateway                    │   │
│  │  ┌─────────┐  ┌─────────┐  ┌─────────┐            │   │
│  │  │ Claude  │  │ GPT-5   │  │ Gemini  │  ...more    │   │
│  │  └────┬────┘  └────┬────┘  └────┬────┘            │   │
│  │       └────────────┼────────────┘                  │   │
│  │                    ▼                                │   │
│  │            ┌──────────────┐                        │   │
│  │            │   AI Agent   │ ← Your custom persona   │   │
│  │            └──────┬───────┘                        │   │
│  └───────────────────┼─────────────────────────────────┘   │
│                      ▼                                      │
│  ┌─────────┐  ┌─────────┐  ┌─────────┐  ┌─────────┐      │
│  │ WhatsApp│  │Telegram │  │ Discord │  │ iMessage│ ...   │
│  └─────────┘  └─────────┘  └─────────┘  └─────────┘      │
└─────────────────────────────────────────────────────────────┘

In short, Moltbot acts as an AI relay hub:

• Upstream: Connects to AI models (Claude, GPT, Gemini, etc.)
• Downstream: Connects to your messaging apps (WhatsApp, Telegram, Discord, etc.)
• Core: Your custom AI Agent with its own memory, skills, and personality

What Is the Wizard?

One-Command Configuration

Setting up Moltbot manually requires writing JSON configs, configuring API tokens for each platform, setting up daemons, debugging network connections, and installing dependencies. For non-technical users, that is a nightmare.

The Wizard is an interactive CLI tool that asks you questions step by step and handles all configuration automatically. Think of it as the “Next, Next, Finish” installer experience.

moltbot onboard

One command to start the magic.

What Does the Wizard Configure?

Prerequisites

System Requirements

• OS: macOS, Linux, or Windows (WSL2 required)
• Node.js: Version 22 or higher
• Network: Access to your chosen AI provider’s API

Getting an API Key

You need at least one AI model API key. Anthropic (Claude) is recommended:

1. Visit Anthropic Console
2. Create an account and generate an API key
3. Save the key — you will need it shortly

Tip: If you already use the Claude Code CLI, the Wizard can automatically reuse its authentication credentials.

Installing Moltbot

Linux/macOS:

curl -fsSL https://molt.bot/install.sh | bash

Windows (PowerShell):

iwr -useb https://molt.bot/install.ps1 | iex

Verify the installation:

moltbot --version

Step-by-Step Wizard Walkthrough

Launch the Wizard

moltbot onboard

You will see a mode selection screen:

? How would you like to proceed?
> QuickStart (recommended defaults)
  Advanced (full control)

QuickStart vs Advanced

For first-time users, choose QuickStart. The Wizard will automatically configure:

• Local Gateway on port 18789
• Token authentication (auto-generated)
• Allowlist mode for Telegram/WhatsApp

Step 1: Choose Your AI Model

? Which model provider would you like to use?
> Anthropic API key (recommended)
  Anthropic OAuth (Claude Code CLI)
  OpenAI API key
  OpenAI Code (Codex) subscription
  Vercel AI Gateway (multi-model proxy)
  MiniMax M2.1
  Moonshot (Kimi K2)
  Skip

Provider comparison:

After selecting “Anthropic API key”, paste your key:

? Enter your Anthropic API key: sk-ant-api03-xxxxx

The Wizard automatically validates the key.

Step 2: Set Up the Workspace

? Where should the agent workspace be?
> ~/clawd (default)
  Custom location

The workspace stores Agent configurations, memory, and sessions. The default ~/clawd works well for most users.

Workspace structure:

~/clawd/
├── IDENTITY.md      # Agent persona definition
├── BOOT.md          # Startup instructions
├── TOOLS.md         # Available tool descriptions
├── SOUL.md          # Core behavior guidelines
└── sessions/        # Conversation history

Step 3: Configure the Gateway

The Gateway is Moltbot’s core — all messages pass through it.

? Gateway port: 18789
? Gateway bind address:
> loopback (127.0.0.1, recommended for single user)
  all interfaces (0.0.0.0, for remote access)
? Enable authentication?
> Yes (recommended)
  No

Key concepts:

The Gateway works like a home router — just as all internet traffic flows through the router, all messages flow through the Gateway.

Loopback vs All Interfaces:

• loopback: Only accessible from your machine — most secure
• all interfaces: Allows access from other devices — must be combined with authentication

Why enable authentication? Even on localhost, other programs running on your machine could potentially interact with the Gateway. Token authentication prevents unauthorized access.

Step 4: Connect Messaging Channels

This is where Moltbot comes alive — bringing AI into your chat apps.

? Which channels would you like to set up?
  ◉ WhatsApp
  ◉ Telegram
  ◯ Discord
  ◯ Google Chat
  ◯ Signal
  ◯ iMessage

Configuring Telegram

If you selected Telegram, the Wizard will prompt:

? Enter your Telegram bot token:

How to get a Bot Token:

1. Search for @BotFather on Telegram
2. Send /newbot
3. Follow the prompts to name your bot
4. BotFather returns a token like 123456789:ABCdefGHIjklMNOpqrsTUVwxyz
5. Paste the token into the Wizard

Configuring WhatsApp

WhatsApp requires QR code authentication:

? Set up WhatsApp now?
> Yes
  Skip for now

Select Yes and the terminal displays a QR code. Scan it with WhatsApp (Settings > Linked Devices).

Note: WhatsApp pairing data is stored in ~/.clawdbot/credentials/whatsapp/.

Step 5: Install the Daemon

? Install as background service?
> Yes (recommended)
  No
? Runtime:
> Node (recommended)
  Bun

The daemon keeps Moltbot running in the background and starts it automatically on boot.

• macOS: Creates a LaunchAgent
• Linux: Creates a systemd user service

Choose Node as the runtime (Bun support is still maturing).

Step 6: Install Skills

Skills are capability plugins that extend what your AI can do.

? Install recommended skills?
> Yes
  No
? Node package manager:
> npm
  pnpm

Recommended Skills include:

• web_search: Search the web
• web_fetch: Fetch and parse web pages
• exec: Execute system commands
• browser: Browser automation

Setup Complete

After configuration, the Wizard shows a summary:

Configuration complete!

Gateway:     ws://127.0.0.1:18789
Workspace:   ~/clawd
Channels:    Telegram, WhatsApp
Model:       claude-sonnet-4-20250514

Next steps:
1. Run 'moltbot dashboard' to open the web UI
2. Send a message to your bot on Telegram or WhatsApp
3. Run 'moltbot status' to check system health

Verifying Your Installation

Check Status

moltbot status

Example output:

Gateway:     Running (ws://127.0.0.1:18789)
Agent:       Ready
Channels:
  Telegram:  Connected (@your_bot)
  WhatsApp:  Connected (+1 555xxxx)
Model:       claude-sonnet-4-20250514

Open the Dashboard

moltbot dashboard

This opens http://127.0.0.1:18789/ in your browser — the Moltbot web console where you can:

• Monitor conversations in real time
• Manage sessions
• Adjust Agent settings

Send Your First Message

Open Telegram, find your bot, and send:

Hello! Tell me about yourself.

If everything is working, you will receive an AI response. Congratulations — your private AI assistant is live.

Advanced Configuration

Non-Interactive Installation

For automated server deployments, use --non-interactive mode:

moltbot onboard --non-interactive \
  --mode local \
  --auth-choice anthropic-api-key \
  --anthropic-api-key "$ANTHROPIC_API_KEY" \
  --gateway-port 18789 \
  --gateway-bind loopback \
  --install-daemon \
  --daemon-runtime node

Running Multiple Agents

A single Moltbot instance can run multiple Agents, each with its own workspace and sessions. For example, create separate work and personal Agents:

# Add a work Agent
moltbot agents add work --workspace ~/clawd-work

# Add a personal Agent
moltbot agents add life --workspace ~/clawd-life

Then use routing rules to direct messages from different channels to different Agents.

Remote Mode

To run the Gateway on a VPS and connect locally:

# Run on your local machine
moltbot onboard --mode remote

Then enter the remote Gateway’s WebSocket address:

? Remote Gateway URL: ws://your-vps-ip:18789
? Authentication token: your-token

Modifying Configuration

After initial setup, use the configure command to change settings:

# Reconfigure everything
moltbot configure

# Reconfigure specific sections
moltbot configure --section gateway
moltbot configure --section channels
moltbot configure --section skills

Troubleshooting

Gateway fails to start — port already in use

# Check what is using the port
lsof -i :18789

# Change the port
moltbot configure --section gateway

WhatsApp won’t connect after scanning the QR code

1. Ensure your phone and computer are on the same network
2. Check firewall settings
3. Try re-pairing:

   moltbot channels login whatsapp
   

Slow AI responses

1. Check network latency to your AI provider
2. Consider using a closer model provider
3. Configure a proxy if needed:

   export HTTPS_PROXY=http://your-proxy:port
   moltbot gateway
   

How to customize the AI persona

Edit IDENTITY.md in your workspace:

cd ~/clawd
nano IDENTITY.md

Example persona:

You are Alex's personal assistant, named "Jarvis".

Personality:
- Concise and well-organized responses
- Professional but friendly tone
- Remembers user preferences

Responsibilities:
- Schedule management
- Technical Q&A
- Writing and translation assistance

Where are the configuration files?

Security Best Practices

Moltbot runs on your device with significant permissions. Follow these guidelines:

1. Always enable Gateway authentication, even in loopback mode
2. Never expose the Gateway to the public internet unless you fully understand the risks
3. Keep Moltbot updated to the latest version
4. Be cautious with exec permissions — this allows the AI to run system commands
5. Audit Skill sources — only install plugins from trusted providers

For group chat scenarios, enable sandbox mode:

{
  "agents": {
    "defaults": {
      "sandbox": {
        "mode": "non-main"
      }
    }
  }
}

This isolates non-primary Agent sessions inside Docker containers.

Summary

The Moltbot Wizard transforms a complex multi-step configuration into answering a few simple prompts. After following this guide, you should have:

• A clear understanding of Moltbot’s architecture
• A fully configured instance via the Wizard
• Connected messaging channels (Telegram, WhatsApp, etc.)
• Knowledge of advanced configuration and troubleshooting

You now own a truly private AI assistant — one that knows you, remembers you, and works exclusively for you.

Further Reading

• Clawdbot (Moltbot): Build Your Own Private AI Assistant
• Claude Code Complete Guide

References:

• Moltbot Official Documentation
• Moltbot Wizard Configuration Page
• Moltbot GitHub Repository
• TechCrunch: Everything you need to know about Moltbot

Related Reading

• Moltbot Explained: 80K Stars, Renaming Drama & Security Guide — Deep dive into Moltbot’s architecture and security
• MoltBot Explained: What It Is, How It Works, and Why It Renamed to OpenClaw — The full story of the Moltbot-to-OpenClaw transition
• OpenClaw Setup Guide: Install and Configure Your AI Agent — Getting started with OpenClaw (the successor to Moltbot)
• OpenClaw vs CrewAI vs AutoGPT 2026: 6 AI Agent Frameworks Compared — How OpenClaw compares to other agent frameworks