clawbot — OpenClaw Docker Template

A ready-to-use Docker template for running an OpenClaw agent gateway with a sandboxed browser, local or cloud AI models, and custom skills.

---

Prerequisites

• Docker with Compose v2
• An OpenAI API key or a local model runner (e.g. Docker Model Runner / Ollama)

---

Quick Start

# 1. Copy the default env file and fill in your values
cp .env.default .env
# Edit .env — at minimum set OPENAI_API_KEY and OPENCLAW_GATEWAY_TOKEN

# 2. Start all services
docker compose up -d

# 3. Get the dashboard URL (wait ~30s for startup + npm update)
docker compose logs openclaw | grep "http://"

Open the printed URL in your browser to connect to the gateway.

---

Environment Variables

Always create .env from the template: cp .env.default .env, then fill in real values.

All variables are defined in .env.default. Copy it to .env and fill in real values.

Variable	Required	Default	Description
OPENAI_API_KEY	Yes*	—	OpenAI API key
OPENCLAW_GATEWAY_TOKEN	Yes	—	Secret token for gateway auth
OPENCLAW_PORT	No	18789	Host port for the gateway
TARGET_ENV	No	dev	Build stage (dev or production)

* Not required if using a local model runner (configure DMR_BASE_URL instead)

---

Dashboard

The gateway dashboard is available at:

http://localhost:<OPENCLAW_PORT>/

To get the tokenized URL (auto-opens connection):

docker compose exec -T openclaw openclaw dashboard --no-open 2>&1 | grep "http://"

Or check container logs — the URL is printed automatically on startup:

docker compose logs openclaw | grep "http://"

---

Customizing Your Agent

Identity

Edit openclaw/workspace/IDENTITY.md to define your agent's name, personality, and avatar.

Skills

Skills are Markdown playbooks that tell the agent how to handle specific tasks. They live in openclaw/workspace/skills/<skill-name>/SKILL.md.

A minimal skill:

---
name: my-skill
description: >-
  Describe when the agent should use this skill (used for trigger matching).
---

# My Skill

## Steps
1. Do this first
2. Then do this

## Triggers
Phrases that activate this skill: "run my skill", "do the thing".

See openclaw/workspace/skills/example-skill/SKILL.md for a working example.

Models

Edit openclaw/openclaw.json to configure model providers, add Ollama models, or change the default model.

---

Useful Commands

# Start
docker compose up -d

# Stop
docker compose down

# Restart openclaw only
docker compose restart openclaw

# View logs
docker compose logs -f openclaw

# Get dashboard URL
docker compose logs openclaw | grep "http://"

---

GPU / Local Model Runner

See docker-compose.gpu.yml and AGENTS.md for instructions on enabling GPU-accelerated local models.