Introduction

Docker Compose deployment for OpenClaw with Ollama integration. This project provides a containerized environment for running OpenClaw (autonomous AI assistant) alongside Ollama for local LLM inference.

Based on the structure of silvaengine_docker and incorporating patterns from mythrantic/ollama-docker.

Services

Service	Description	Port
openclaw	OpenClaw Gateway (Node.js + Python)	18789
ollama	Ollama LLM Server	11434
ollama-webui	Open WebUI Interface	8080

Installation

1. Clone this repository and navigate to the directory:

   cd docker-openclaw

2. Create the required directories:

   mkdir www/logs
   mkdir www/projects

3. (Optional) Place SSH keys in python/.ssh/ for GitHub access.

4. Create a .env file from the example:

   cp .env.example .env

   Configure your environment variables:

   PIP_INDEX_URL=https://pypi.org/simple/
   PYTHON=python3.12
   NODE_VERSION=22
   PROJECTS_FOLDER={path to the projects folder}
   ANTHROPIC_API_KEY=your_anthropic_api_key
   OPENAI_API_KEY=your_openai_api_key
   

   Example:

   • PIP_INDEX_URL: https://pypi.org/simple/
   • PROJECTS_FOLDER: "C:/Users/developer/GitHubRepos/openclaw_projects"
   • PYTHON: python3.12

5. Build the Docker images:

   docker compose build

6. Start the containers:

   docker compose up -d

Configure requirements.txt

Configure the python/requirements.txt file to manage the necessary packages. Ensure this file includes all dependencies required for your OpenClaw projects to run smoothly.

Usage

OpenClaw Gateway

The OpenClaw container includes:

• Node.js 22 with OpenClaw installed globally via pnpm
• Python 3.12 with virtual environment at /var/python3.12/openclaw/env
• Ollama, Anthropic, OpenAI Python SDKs
• Supervisor for process management

Ollama

Pull models via the API or Open WebUI:

# Pull a model
curl http://localhost:11434/api/pull -d '{"name": "llama3.2"}'

# List models
curl http://localhost:11434/api/tags

Open WebUI

Access the web interface at http://localhost:8080 to interact with Ollama models.

GPU Support

The ollama service is configured for NVIDIA GPU support. Ensure you have:

• NVIDIA drivers installed
• NVIDIA Container Toolkit installed

For CPU-only mode, remove the deploy section from the ollama service in docker-compose.yaml.

Directory Structure

docker-openclaw/
├── docker-compose.yaml
├── .env.example
├── .gitignore
├── README.md
├── python/
│   ├── Dockerfile
│   ├── openclaw.ini
│   ├── requirements.txt
│   └── .ssh/
└── www/
    ├── logs/
    └── projects/

License

MIT