openclaw-docker

This repository provides a Docker image build for OpenClaw. The image can be built locally or via GitHub Actions and pushed to GitHub Container Registry (GHCR).

Features

• Automated build via GitHub Actions (workflow_dispatch)
• Push to GHCR after successful build
• Build cache support to speed up subsequent builds

How to build (GitHub Actions)

1. Open the repository's Actions tab on GitHub.
2. Select the Build OpenClaw Docker Image workflow.
3. Click Run workflow, choose the branch and start the job.

The workflow is manually triggered (workflow_dispatch) and uses the docker/Dockerfile in this repository.

Image info

• Registry: ghcr.io
• Image name: vxwork/openclaw-docker/online-base
• Tag example: linux_latest

Pull example:

docker pull ghcr.io/vxwork/openclaw-docker/online-base:linux_latest

Run locally

Run with docker run (example):

docker run -d --name openclaw -p 18789:18789 \
  -v openclaw-config:/app/config -v openclaw-data:/app/data \
  ghcr.io/vxwork/openclaw-docker/online-base:linux_latest

Or use docker-compose (recommended when using this repository's compose file). This repo includes docker-compose.yml that defines a service named openclawbot. By default the compose file mounts ./config to /root/.openclaw inside the container so the entrypoint can persist configuration and pairing tokens there.

# start in detached mode
docker compose up -d

# stop and remove
docker compose down

If you prefer to build locally before running with docker run:

docker build -t openclaw:local -f docker/Dockerfile .
docker run -d --name openclaw -p 18789:18789 openclaw:local

First-run configuration

This image includes an entrypoint script that initializes configuration on first start.

• The compose file mounts ./config to /root/.openclaw inside the container. On first start, the entrypoint checks for /root/.openclaw/openclaw.json. If missing it:

  • creates /root/.openclaw/workspace and /root/.openclaw/pairing
  • generates a device token and writes /root/.openclaw/pairing/device.json
  • starts the gateway with --allow-unconfigured so you can complete initial setup

• To manually run the config helper inside a running container (if needed):

# for docker run example
docker exec -it openclaw /bin/sh -c "pnpm openclaw config"

# for docker compose example (service name: openclawbot)
docker compose exec openclawbot /bin/sh -c "pnpm openclaw config"

Make sure the host ./config directory is writable by the container, so configuration persists across restarts.

Notes on Dockerfile

• The docker/Dockerfile in this repository is based on almalinux:10.1-minimal and installs Node.js via NodeSource. It then installs pnpm, clones the openclaw repository, installs dependencies and builds the project.
• The image exposes port 18789 and defines VOLUME entries for persistent config and data.

Migration from qverisbot-docker

This repository has been updated to follow the layout and conventions from the qverisbot-docker main branch. The Dockerfile uses almalinux for packaging as requested; if you want an exact 1:1 copy of that repo's files, provide the repo or point to specific diffs and I will sync them precisely.

Recommendation: Qveris AI for OpenClaw

For an enhanced OpenClaw experience, consider using Qveris AI — a toolset that integrates with OpenClaw and provides automation and AI-driven helpers. Qveris AI is recommended as a companion for OpenClaw deployments.

Troubleshooting

• If the service does not start, inspect container logs:

# view logs for docker run example
docker logs -f openclaw

# view logs for docker compose example (service name: openclawbot)
docker compose logs -f openclawbot

• Interactive shell:

docker exec -it openclaw /bin/sh       # docker run example
docker compose exec openclawbot /bin/sh # compose example