Back to templates
OpenClaw icon

Deploy OpenClaw on Sealos

AI agent gateway for browser-based agent control and chat-channel automation

AI
Launch a private OpenClaw AI workspace with one-click deployment.Keep models, prompts, and workflow data in your own Sealos environment.Scale GPU or CPU resources as your AI workload grows.Use an open-source template without managing Kubernetes YAML.
Official WebsiteGitHubREADME

Why deploy on Sealos

Sealos makes deploying any app effortless, secure, and production-ready. From one-click launch to ongoing operations, we handle the heavy lifting so you can focus on what matters.

One-Click Deploy
Sealos logo
Compute
Networking
Storage
Security
Observability
Your Application is Live

One-Click Deployment

Deploy any app template in seconds. No compose setup, manual configure, and go live.

Managed Kubernetes Reliability

Built on Sealos Managed Kubernetes for high availability, auto-scaling, and self-healing by default.

Automatic HTTPS & Security

Every deployment includes a secure domain with automatic SSL. We handle certificates and text protection for you.

Persistent Storage

Attach persistent volumes with ease. Your data stays safe, durable, and always accessible.

Scale when needed

Adjust resources as your app grows, no downtime required.

You Get the Whole Stack

Sealos provisions and connects every resource your application needs. Everything is ready to use.

One-click, fully provisionedSecure by defaultProduction ready

App Service

Your application is running and ready to serve requests.

Public HTTPS URL

Secure, global endpoint to access your application.

Database

Managed database instance for your application.

Persistent Volume

Durable storage for uploads and application data.

Environment Variables

Configuration and secrets injected securely into your app.

Logs & Metrics

Centralized logs and basic metrics are enabled.

README

README.md

Deploy and Host OpenClaw on Sealos

OpenClaw is an AI agent gateway for browser-based agent control and chat-channel automation. This template deploys OpenClaw as a single StatefulSet with persistent state on Sealos Cloud.

OpenClaw Screenshot

About Hosting OpenClaw

OpenClaw runs a Gateway and Control UI on port 18789. The Gateway manages agent configuration, model provider access, device pairing, and workspace files for browser-driven agent workflows.

This Sealos template follows the official Docker and Kubernetes runtime model: one OpenClaw container, Gateway token authentication, /healthz and /readyz probes, and PVC-backed local state. The deployment stores OpenClaw config, workspace data, and browser pairing profiles across restarts.

Common Use Cases

  • Personal AI Operator: Run a browser-based assistant that can use your configured model provider.
  • Team Agent Console: Host a shared Control UI for agent setup, testing, and operational review.
  • Chat-Channel Automation: Connect agents to messaging channels after creating the required platform tokens.
  • Workflow Prototyping: Keep agent workspace files and instructions persistent while iterating on tasks.

Dependencies for OpenClaw Hosting

The template includes the OpenClaw Gateway image and persistent volumes. You provide a compatible model endpoint, API key, and default model id at deployment time.

Deployment Dependencies

Implementation Details

Architecture Components:

This template deploys one OpenClaw Gateway service:

  • OpenClaw Gateway: Serves the Control UI, WebSocket Gateway, and health endpoints on port 18789.
  • State PVC: Mounts /home/node/.openclaw for openclaw.json, workspace files, and agent state.
  • Profile PVC: Mounts /home/node/.config/openclaw for browser device pairing profile data.

Configuration:

At first boot, the template writes openclaw.json with:

  • Gateway token authentication enabled
  • Gateway bind mode set for Sealos ingress access
  • Control UI origin restricted to your generated Sealos HTTPS URL
  • A default model provider built from provider_kind, base_url, api_key, and model
  • A default agent workspace at ~/.openclaw/workspace

Persistence is PVC-backed local state, matching the official Docker and Kubernetes deployment guides.

License Information:

OpenClaw is licensed under the MIT License. This Sealos template is provided under the same license terms as the templates repository.

Why Deploy OpenClaw on Sealos?

Sealos is an AI-assisted Cloud Operating System built on Kubernetes. By deploying OpenClaw on Sealos, you get:

  • One-Click Deployment: Open the template page, enter the model provider values, and deploy.
  • Managed HTTPS Access: Sealos creates the public URL and TLS certificate for the Gateway.
  • Persistent Agent State: PVCs keep workspace, configuration, and profile data across restarts.
  • Resource Controls: The template maps the official Kubernetes baseline of 1 CPU and 2Gi memory to the Sealos resource ladder.
  • Canvas Operations: Adjust environment variables, resources, and replica settings from the Sealos Canvas after deployment.

Deployment Guide

  1. Open the OpenClaw template and click Deploy Now.
  2. Configure the parameters in the popup dialog:
    • provider_kind: Select openai_compat or anthropic_compat.
    • base_url: Enter the provider base URL, such as an OpenAI-compatible /v1 endpoint.
    • api_key: Enter the provider API key.
    • model: Enter the model id, such as claude-opus-4-6 or gpt-5.2.
  3. Wait for deployment to complete, typically 2-3 minutes. After deployment, Sealos redirects you to the Canvas. For later changes, describe your request in the AI dialog or open the relevant resource card.
  4. Open the OpenClaw App card URL. The URL includes the generated Gateway token as a fragment.
  5. Complete first-device pairing when the Control UI asks for approval:
openclaw devices list
openclaw devices approve <requestId>

Run those commands from the OpenClaw container terminal in Canvas. After approval, refresh the Control UI and continue setup.

Initial Setup and Login

The App URL includes #token=<gateway-token>, which lets the Control UI authenticate against the Gateway. Browser device pairing protects remote access: approve the first browser once, then the profile PVC keeps that pairing data across restarts.

After opening the Control UI:

  1. Confirm that the default provider and model are listed in model settings.
  2. Open the default assistant workspace and send a short prompt.
  3. Add messaging-channel credentials from the Control UI or by editing environment variables on the StatefulSet resource card.

Configuration

You can update OpenClaw after deployment through:

  • Control UI: Manage agents, model providers, workspace files, and channel setup.
  • Canvas AI Dialog: Describe configuration changes and let Sealos apply them.
  • StatefulSet Resource Card: Edit provider values, API keys, and resource limits.

For provider changes, keep the provider protocol aligned with the base URL:

  • openai_compat for OpenAI-compatible chat-completions endpoints
  • anthropic_compat for Anthropic Messages-compatible endpoints

Scaling

OpenClaw keeps local state in PVCs, so the template runs one replica. To adjust resources:

  1. Open the Canvas for your OpenClaw deployment.
  2. Open the StatefulSet resource card.
  3. Adjust CPU or memory using the Sealos resource controls.
  4. Apply the update and wait for the Pod to become Ready.

The template starts with 1 CPU and 2048Mi memory, matching the official Kubernetes manifest and the Sealos resource ladder.

Troubleshooting

Control UI Shows a Pairing Screen

Open the OpenClaw container terminal from Canvas, run openclaw devices list, then approve the shown request id with openclaw devices approve <requestId>.

Model Calls Fail

Confirm that provider_kind, base_url, api_key, and model match your provider. For OpenAI-compatible providers, the base URL usually ends with /v1.

Gateway Health Fails

Open the StatefulSet logs and check that the Gateway is listening on port 18789. The probes call /healthz and /readyz, matching the official container healthcheck.

Additional Resources

License

This Sealos template is provided under the templates repository license. OpenClaw is licensed under the MIT License.

Related templates

Explore more production-ready templates.

More