Deployment Management
Learn how to deploy new agent configurations, monitor rollout progress, and roll back deployments in WorkClaw.
What is a deployment in WorkClaw?
A deployment is the process of pushing a new build to one or more runtimes. When you change your agent's skills, connections, or configuration, those changes are packaged into a build and then deployed to the runtime hosts where your Claw operates.
Each deployment is tracked with a unique identifier, timestamp, and status so you always know what version of your agent is running.
How do I trigger a deployment?
Navigate to Admin > Deployments and click Deploy. WorkClaw will automatically create a fresh build from your current configuration and begin the rollout using your selected deploy strategy. You can also trigger deployments through the API.
Before triggering, confirm your build is healthy by checking the Build System page for any errors.
How do I monitor a deployment in progress?
The deployment detail view shows real-time progress including:
- Phase -- which stage of the deploy strategy is active (e.g., shifting traffic, draining old containers).
- Runtime status -- per-host health indicators showing which runtime hosts have received the new build.
- Error count -- any failures encountered during rollout.
For deeper observability, see Monitoring & Observability.
How do I roll back a deployment?
If a deployment introduces problems, click Rollback on the deployment detail page. WorkClaw will redeploy the previous healthy build using the same deploy strategy. Rollbacks follow the same phased process as forward deployments, ensuring zero downtime when using blue/green or rolling strategies.
What is the deploy cadence?
The deploy cadence controls how frequently automatic deployments occur. You can set it to manual (deploy only when triggered), or configure automatic intervals. Adjust the cadence under Admin > Settings > Deploy Cadence.
Can I deploy to a subset of runtimes?
Yes. When using batched deployments, you can target specific runtime groups. This is useful for canary-style releases where you validate changes on a small set of hosts before rolling out broadly.