Quick Start

goClaw ships with a CLI (@clawrm/cli) that handles project scaffolding, channel configuration, and initial knowledge base setup. You can be running an agent in under 5 minutes.

Prerequisites

• Node.js 22 or later
• pnpm (recommended) or npm
• An Anthropic API key (Claude is the primary agent LLM)
• Optional: Twilio account (SMS/WhatsApp), Resend account (Email), Telegram bot token

Step 1 — Initialize a new project

npx @clawrm/cli init my-agent

The CLI will prompt you for:

1. Project name — becomes your agent's identity
2. Primary use case — GTM, Support, Sales Dev, or custom
3. Channels to enable — Email, SMS, Telegram, WhatsApp (can add more later)
4. LLM API key — Anthropic API key (required)

The wizard scaffolds a complete project directory:

my-agent/
├── .env                    # API keys and channel credentials
├── config/
│   ├── agent.yaml          # Agent persona, goals, behaviors
│   ├── channels.yaml       # Channel configuration
│   └── permissions.yaml    # Group and sandbox settings
├── knowledge/              # Initial knowledge base (empty)
├── skills/                 # Agent skills (empty)
└── package.json

Step 2 — Install dependencies

cd my-agent
pnpm install

Step 3 — Configure environment variables

The CLI creates a .env file with placeholders. Fill in your credentials:

# Required
ANTHROPIC_API_KEY=sk-ant-...

# Email (optional)
RESEND_API_KEY=re_...
IMAP_HOST=imap.gmail.com
IMAP_USER=your@email.com
IMAP_PASSWORD=...

# SMS / WhatsApp (optional)
TWILIO_ACCOUNT_SID=AC...
TWILIO_AUTH_TOKEN=...
TWILIO_PHONE_NUMBER=+1...

# Telegram (optional)
TELEGRAM_BOT_TOKEN=...

Step 4 — Start the agent

pnpm dev

This starts:

• The agent runtime (NanoClaw Core)
• The admin dashboard on http://localhost:3000
• Channel listeners for any configured channels

Step 5 — Open the admin dashboard

Navigate to http://localhost:3000. You'll see:

• Inbox — incoming messages across all channels
• Contacts — your CRM
• Knowledge — current knowledge base files
• Research Queue — pending curiosity items
• Settings — agent configuration

Running in production

For production deployments, goClaw supports two modes:

Docker (recommended)

# Build the image
pnpm build:docker

# Run with Docker Compose
docker-compose up -d

The default docker-compose.yml includes the agent runtime, admin dashboard, and optional nginx reverse proxy.

Bare Node.js on EC2

# Build
pnpm build

# Start with PM2
pm2 start dist/agent.js --name my-agent
pm2 start dist/admin.js --name my-agent-admin

A t3.small EC2 instance handles most production workloads. Add a t3.medium for heavy multi-agent deployments.

Managed hosting (coming soon)

Indigo will offer managed goClaw hosting — you provide the credentials and config, we handle the infrastructure. See Pricing for upcoming tiers.

Next steps

• Configure your agent — fine-tune persona, goals, and behaviors
• Architecture overview — understand what's running under the hood
• Pick a use case — see how others are deploying goClaw