OpenClaw Docker Image

Docker image for OpenClaw - your personal AI assistant.

Built for use with Umbrel and other self-hosted platforms.

View available tags - follows OpenClaw's release cycle using calendar versioning (e.g., v2026.1.30).

Quick Start

Using Docker Compose (Recommended)

Clone this repository
Copy .env.example to .env and add your API keys
Run:

docker compose up -d

Open http://localhost:18789/chat

Using Docker Run

docker run -d \
-p 18789:18789 \
-e ANTHROPIC_API_KEY=your-key \
-e TELEGRAM_BOT_TOKEN=your-bot-token \
-v openclaw-data:/home/openclaw/.openclaw \
-v openclaw-workspace:/home/openclaw/clawd \
ghcr.io/zot24/openclaw-docker:latest

Supported Channels

Channel	Environment Variables	Notes
Telegram	`TELEGRAM_BOT_TOKEN`	Get token from @BotFather
WhatsApp	`WHATSAPP_ENABLED=true`	Requires QR code linking
Discord	`DISCORD_BOT_TOKEN`	From Discord Developer Portal
Slack	`SLACK_APP_TOKEN`, `SLACK_BOT_TOKEN`	Socket mode required
MS Teams	`MSTEAMS_APP_ID`, `MSTEAMS_APP_PASSWORD`	Azure AD app registration
Signal	`SIGNAL_NUMBER`	Requires signal-cli (not included)

Supported LLM Providers

Provider	Environment Variables	Model Example
Anthropic	`ANTHROPIC_API_KEY`	`anthropic/claude-sonnet-4`
MiniMax	`MINIMAX_API_KEY`	`minimax/MiniMax-M2.1`
OpenAI	`OPENAI_API_KEY`	`openai/gpt-4o`
OpenRouter	`OPENROUTER_API_KEY`	`openrouter/anthropic/claude-sonnet-4`
Moonshot	`MOONSHOT_API_KEY`	`moonshot/moonshot-v1-128k`
GLM	`GLM_API_KEY`	`glm/glm-4-plus`
Local (Ollama, etc.)	`OPENCODE_BASE_URL`	`opencode/llama3.1`

The image auto-selects a model based on available API keys. Override with OPENCLAW_MODEL.

Environment Variables

LLM Providers

Variable	Required	Description
`ANTHROPIC_API_KEY`	One provider required	Anthropic API key
`MINIMAX_API_KEY`		MiniMax API key
`OPENAI_API_KEY`		OpenAI API key
`OPENAI_ORG_ID`		OpenAI organization ID
`OPENAI_BASE_URL`		OpenAI-compatible endpoint
`OPENROUTER_API_KEY`		OpenRouter API key
`MOONSHOT_API_KEY`		Moonshot API key
`GLM_API_KEY`		GLM/ChatGLM API key
`OPENCODE_BASE_URL`		Local model endpoint (e.g., Ollama)
`OPENCODE_API_KEY`		API key for local endpoint
`OPENCODE_MODEL`		Model name for local provider
`OPENCLAW_MODEL`		Override auto-selected model

Channels

Variable	Default	Description
`TELEGRAM_BOT_TOKEN`		Telegram bot token
`TELEGRAM_ALLOWED_USERS`		Comma-separated user IDs
`OPENCLAW_DM_POLICY`	`pairing`	DM policy: pairing, allowlist, open
`WHATSAPP_ENABLED`	`false`	Enable WhatsApp channel
`WHATSAPP_DM_POLICY`	`pairing`	DM policy
`WHATSAPP_GROUP_POLICY`	`disabled`	Group policy: disabled, allowlist
`WHATSAPP_ALLOWED_NUMBERS`		Comma-separated phone numbers
`WHATSAPP_GROUPS`		Group IDs or `*` for all
`DISCORD_BOT_TOKEN`		Discord bot token
`DISCORD_ALLOWED_USERS`		Comma-separated user IDs
`DISCORD_DM_POLICY`	`pairing`	DM policy
`SLACK_APP_TOKEN`		Slack app token (xapp-...)
`SLACK_BOT_TOKEN`		Slack bot token (xoxb-...)
`SLACK_ALLOWED_USERS`		Comma-separated user IDs
`MSTEAMS_APP_ID`		Azure AD app ID
`MSTEAMS_APP_PASSWORD`		Azure AD app password
`SIGNAL_NUMBER`		Signal phone number

Gateway

Variable	Default	Description
`OPENCLAW_GATEWAY_TOKEN`	Auto-generated	Authentication token
`OPENCLAW_GATEWAY_PORT`	`18789`	Gateway port
`GATEWAY_MODE`	`local`	Gateway mode: local, cloud
`GATEWAY_BIND`	`lan`	Bind mode: lan, loopback, tailnet

Agent Runtime

Variable	Default	Description
`AGENT_TIMEOUT`	`600`	Agent timeout in seconds
`AGENT_TIMEZONE`	`UTC`	Timezone for agent
`ENABLE_MEMORY_SEARCH`	`false`	Enable memory search
`ENABLE_BROWSER`	`true`	Enable browser automation
`ENABLE_EXEC`	`true`	Enable shell execution
`EXEC_TIMEOUT`	`30000`	Exec timeout in milliseconds

Text-to-Speech (sag)

Variable	Default	Description
`ELEVENLABS_API_KEY`		ElevenLabs API key for TTS
`SAG_VOICE_ID`		Default voice ID for sag

Data Directories

Variable	Default	Description
`OPENCLAW_DATA_DIR`	`/home/openclaw/.openclaw`	Config and credentials
`OPENCLAW_WORKSPACE`	`/home/openclaw/clawd`	Workspace, memory, skills

Ports

18789: Gateway (HTTP + WebSocket + WebChat at /chat)
18790: Bridge (TCP for mobile nodes)

Volumes

/home/openclaw/.openclaw: Configuration and credentials
/home/openclaw/clawd: Workspace, memory, and skills

Channel Setup

Telegram

Message @BotFather on Telegram
Create a new bot with /newbot
Copy the bot token
Set TELEGRAM_BOT_TOKEN=your-token

WhatsApp

Set WHATSAPP_ENABLED=true
Start the container
Run docker exec -it <container> node dist/index.js channels login
Scan the QR code with WhatsApp

Discord

Go to Discord Developer Portal
Create a new application
Go to Bot section, create a bot
Copy the token and set DISCORD_BOT_TOKEN=your-token
Enable required intents (Message Content, etc.)
Invite bot to your server with appropriate permissions

Slack

Create a new Slack App at api.slack.com
Enable Socket Mode
Add required scopes: chat:write, app_mentions:read, im:history, etc.
Install to workspace
Copy App Token (xapp-...) and Bot Token (xoxb-...)
Set both SLACK_APP_TOKEN and SLACK_BOT_TOKEN

MS Teams

Register an app in Azure AD
Configure Bot Framework registration
Set MSTEAMS_APP_ID and MSTEAMS_APP_PASSWORD

Using Local Models (Ollama)

On Host Machine

To use Ollama running on your host machine:

docker run -d \
-p 18789:18789 \
-e OPENCODE_BASE_URL=http://host.docker.internal:11434/v1 \
-e OPENCODE_MODEL=llama3.1 \
-e TELEGRAM_BOT_TOKEN=your-token \
-v openclaw-data:/home/openclaw/.openclaw \
-v openclaw-workspace:/home/openclaw/clawd \
ghcr.io/zot24/openclaw-docker:latest

On Umbrel (Ollama in Another Container)

When running both OpenClaw and Ollama as Umbrel apps, they communicate via Docker's internal network.

1. Find the Ollama container name:

docker ps | grep ollama

Look for something like ollama_server_1 or ollama_web_1.

2. Find the Umbrel network:

docker network ls | grep umbrel

3. Get Ollama's network details:

docker inspect <ollama-container-name>| grep -A5 Networks

4. Configure OpenClaw:

Set these environment variables in your OpenClaw configuration:

OPENCODE_BASE_URL=http://<ollama-container-name>:11434/v1
OPENCODE_MODEL=llama3.1

Replace <ollama-container-name> with the actual container name from step 1 (e.g., ollama_server_1).

5. Ensure both containers are on the same network:

If OpenClaw can't reach Ollama, connect it to Umbrel's network:

docker network connect <umbrel-network><openclaw-container>

Tip: You can also use Ollama's internal IP address instead of the container name if DNS resolution doesn't work.

Security

This image runs as a dedicated non-root user (openclaw, UID 1000) for enhanced security:

Principle of Least Privilege: The container process has only the permissions it needs, not full root access
Container Escape Mitigation: If an attacker exploits a vulnerability in the application, they gain limited user privileges rather than root
Host System Protection: Volume mounts and any potential breakouts are constrained to non-root permissions
Compliance: Many security frameworks (CIS Docker Benchmark, PCI-DSS) require or recommend non-root containers
Defense in Depth: Adds another security layer on top of container isolation

The container uses UID/GID 1000, which matches the default user on most Linux systems, making volume permission management straightforward.

Features

Browser Automation: Playwright with Chromium for web scraping
Voice Messages: FFmpeg for audio processing
Image Processing: ImageMagick for image manipulation
Shell Execution: Run commands via the exec tool
Memory Search: Semantic search over conversation history
Text-to-Speech: ElevenLabs TTS via sag
Speech-to-Text: OpenAI Whisper for transcription
MCP Tools: mcporter for Model Context Protocol servers
Python Tools: uv/uvx for running Python MCP tools
GitHub CLI: gh for GitHub operations

Building Locally

# Build with docker-compose
docker compose build
# Or build directly
docker build -t openclaw:local .

The Dockerfile uses a multi-stage build for optimal caching:

deps: System packages, Go, Python, global tools (cached)
builder: Clone and build OpenClaw (rebuilds on version changes)
runtime: Final image with all tools

Included Tools

Tool	Purpose
Playwright + Chromium	Browser automation and web scraping
FFmpeg	Audio/video processing
ImageMagick	Image manipulation
sag	ElevenLabs TTS CLI
whisper	OpenAI speech-to-text
mcporter	MCP server management
uv/uvx	Python tool runner
gh	GitHub CLI
xvfb	Virtual framebuffer for headless browsers

Image Size

The full-featured image is approximately 1.5-2GB due to:

Node.js runtime
Go runtime and sag binary
Python and uv
Chromium browser
FFmpeg and ImageMagick
OpenClaw and dependencies

License

This Docker image builds OpenClaw from source. See the original repository for licensing information.

Name		Name	Last commit message	Last commit date
.github/workflows		.github/workflows
.env.example		.env.example
.gitignore		.gitignore
CLAUDE.md		CLAUDE.md
Dockerfile		Dockerfile
Dockerfile.base		Dockerfile.base
README.md		README.md
VERSION		VERSION
docker-compose.yml		docker-compose.yml
entrypoint.sh		entrypoint.sh
test.sh		test.sh

Navigation Menu

Uh oh!

Repository files navigation

OpenClaw Docker Image

Quick Start

Using Docker Compose (Recommended)

Using Docker Run

Supported Channels

Supported LLM Providers

Environment Variables

LLM Providers

Channels

Gateway

Agent Runtime

Text-to-Speech (sag)

Data Directories

Ports

Volumes

Channel Setup

Telegram

WhatsApp

Discord

Slack

MS Teams

Using Local Models (Ollama)

On Host Machine

On Umbrel (Ollama in Another Container)

Security

Features

Building Locally

Included Tools

Image Size

License

About

Releases

Packages

Contributors

Languages

Uh oh!

Latest commit

History

Folders and files

Repository files navigation

OpenClaw Docker Image

Quick Start

Using Docker Compose (Recommended)

Using Docker Run

Supported Channels

Supported LLM Providers

Environment Variables

LLM Providers

Channels

Gateway

Agent Runtime

Text-to-Speech (sag)

Data Directories

Ports

Volumes

Channel Setup

Telegram

WhatsApp

Discord

Slack

MS Teams

Using Local Models (Ollama)

On Host Machine

On Umbrel (Ollama in Another Container)

Security

Features

Building Locally

Included Tools

Image Size

License

About

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages