Your First Agent in 15 Minutes

This guide walks you through deploying your first OpenClaw agent and getting it running on your server.

Prerequisites

• A Linux server (Ubuntu 22.04+ recommended)
• Terminal/SSH access
• An Anthropic API key

Step 1: Install OpenClaw

Install the OpenClaw CLI as a systemd daemon:

openclaw onboard --install-daemon

This sets up auto-restart on failure, system logging via journald, and automatic startup on boot.

Step 2: Create Your First Agent

openclaw agents add coordinator

The wizard prompts you through configuration — model selection, auth setup, and workspace creation.

Step 3: Set Agent Identity

openclaw agents set-identity --agent coordinator --name "Coordinator Agent"

Step 4: Configure the Workspace

Navigate to your agent's workspace and create the core files:

cd ~/.openclaw/workspace-coordinator

Create these files:

• SOUL.md — Who the agent is (persona, tone, domain expertise)
• AGENTS.md — How the agent operates (protocols, rules, workflows)
• USER.md — Who the users are and their context
• TOOLS.md — Available tools and API references

Step 5: Connect a Channel

Bind your agent to a messaging channel:

openclaw agents bind --agent coordinator --bind slack:main

Step 6: Start the Gateway

openclaw gateway --port 18789

Step 7: Verify

openclaw doctor
openclaw channels status --probe

Rollout Strategy

1. Phase 1 (Weeks 1-2) — Single agent, one channel, validate memory and sessions
2. Phase 2 (Weeks 3-4) — Add a second specialist agent (e.g., engineering)
3. Phase 3 (Weeks 5-6) — Add remaining agents, build custom skills
4. Phase 4 (Weeks 7-8) — Enable sub-agent orchestration and automated workflows

Next Steps

• Architecture & Multi-Agent Gateway — Understand the gateway pattern
• Agent Roles — Design agents for different functions
• Create a Slack Agent — Full walkthrough with multi-bot config