Guide 01

Create the Main Agent

Go from installed OpenClaw to a working main agent with visuals that show what the setup is supposed to look like at each stage.

Main Agent Setup

Turn OpenClaw into your command center

By this point the machine and OpenClaw are already set up. The win in this guide is turning that clean base into a main agent with a clear role, clear rules, and a first successful task.

Objective

Create the main agent on top of OpenClaw, use Opus as the lead orchestration brain, define its identity and rules, and establish the file structure that every future agent and workflow depends on.

Step-by-Step Build Path

Confirm the machine and OpenClaw are already set up and working.
Create the main agent identity files: purpose, user context, memory, rules, and tool notes.
Set Opus as the primary orchestration brain for the top-level agent.
Run one real task end-to-end before adding specialist agents.

Visual Checkpoints

OpenClaw launches cleanly into the correct workspace
Main agent has purpose, rules, and memory files
Opus is positioned as the lead orchestration brain
First successful task run is captured as proof

What This Guide Includes

Main-agent framing for normal users
OpenClaw-to-agent transition checkpoints
Starter mental model for identity + memory files
Windows adaptation notes where local setup differs

Terminal Walkthrough Moments

OpenClaw install terminal support visual — **Install:** Use this as the visual anchor for the package install and OpenClaw setup phase. It is best treated as a support visual, not exact command proof.

Main agent file creation terminal support visual — **Main agent setup:** Shows the moment where identity files, rules, memory, and first-task proof come together in one screen.