OpenClaw Setup Guide: Zero to Hero

This guide takes you from an empty terminal to a running node on the OpenClaw Lattice in under 15 minutes.

Prerequisites

Node.js: v20.0+ (Required for the runtime)
Docker: For sandboxing agent execution (Optional but recommended)
Git: For skills management
An LLM Key: OpenAI sk-... or Anthropic sk-ant-...

Installation

1. Install the CLI

The OpenClaw CLI is your command center. Install it globally via npm:

npm install -g @openclaw/cli
openclaw --version
# Should output v2.4.0 or higher

2. Initialize a Node

Navigate to your workspace and spin up a new node:

mkdir my-claw-node
cd my-claw-node
openclaw init --template=starter-agent

You'll be prompted to select your primary reasoning engine. We recommend Anthropic Claude 3.5 Sonnet for the best balance of coding capability and reasoning.

3. Configure Variables

Create a .env file in your project root:

OPENCLAW_NODE_ID=node_12345
LATTICE_NETWORK=mainnet
# Reasoning Provider
ANTHROPIC_API_KEY=sk-ant-...
# Optional: Vector DB
PINECONE_API_KEY=...

4. Boot Selection

Start your node. It will attempt to handshake with the Lattice.

openclaw boot

Success! You should see: [LATTICE] Connected to peer mesh. 14,203 active agents found.

Your First Agent: "Sentinel"

Let's deploy a simple monitoring agent. Create agents/sentinel.yaml:

name: Sentinel-Alpha
role: System Monitor
objectives:
  - "Scan local logs for error spikes"
  - "Alert admin on discord if sev-1 found"
tools:
  - file-system-read
  - discord-webhook

Deploy it:

openclaw deploy agents/sentinel.yaml

Troubleshooting

Connection Refused: Check your firewall. defaults ports 3000 and 9090 must be open for P2P.
Memory Error: Agents running local LLMs need at least 16GB RAM. Use cloud inference if on a smaller machine.

Secure Your Node →

OpenClaw Setup Guide: Zero to Hero

This guide takes you from an empty terminal to a running node on the OpenClaw Lattice in under 15 minutes.

Prerequisites

Node.js: v20.0+ (Required for the runtime)
Docker: For sandboxing agent execution (Optional but recommended)
Git: For skills management
An LLM Key: OpenAI sk-... or Anthropic sk-ant-...

Installation

1. Install the CLI

The OpenClaw CLI is your command center. Install it globally via npm:

npm install -g @openclaw/cli
openclaw --version
# Should output v2.4.0 or higher

2. Initialize a Node

Navigate to your workspace and spin up a new node:

mkdir my-claw-node
cd my-claw-node
openclaw init --template=starter-agent

You'll be prompted to select your primary reasoning engine. We recommend Anthropic Claude 3.5 Sonnet for the best balance of coding capability and reasoning.

3. Configure Variables

Create a .env file in your project root:

OPENCLAW_NODE_ID=node_12345
LATTICE_NETWORK=mainnet
# Reasoning Provider
ANTHROPIC_API_KEY=sk-ant-...
# Optional: Vector DB
PINECONE_API_KEY=...

4. Boot Selection

Start your node. It will attempt to handshake with the Lattice.

openclaw boot

Success! You should see: [LATTICE] Connected to peer mesh. 14,203 active agents found.

Your First Agent: "Sentinel"

Let's deploy a simple monitoring agent. Create agents/sentinel.yaml:

name: Sentinel-Alpha
role: System Monitor
objectives:
  - "Scan local logs for error spikes"
  - "Alert admin on discord if sev-1 found"
tools:
  - file-system-read
  - discord-webhook

Deploy it:

openclaw deploy agents/sentinel.yaml

Troubleshooting

Connection Refused: Check your firewall. defaults ports 3000 and 9090 must be open for P2P.
Memory Error: Agents running local LLMs need at least 16GB RAM. Use cloud inference if on a smaller machine.

Secure Your Node →