Runner Management

Managed runners are secure OpenClaw containers that run AI agents in isolated sandbox environments. The dashboard provides a full lifecycle management interface — create, configure, monitor, and destroy runners without touching the command line.

Runner Lifecycle

Runners follow a state machine lifecycle:

Creating → Provisioning → Running → Stopping → Stopped → Destroying → Destroyed
                                       ↓
                                    Errored

Creating a Runner

From the dashboard sidebar, navigate to Runners and click Create Runner.

Configuration options:

Runbooks:

Tool profiles:

Monitoring Runners

The runner detail page shows:

• Status — current lifecycle state with uptime
• Resource usage — CPU, memory, disk consumption
• Session logs — real-time log streaming
• Cost tracker — cumulative session cost vs budget
• Activity feed — agent actions, tool calls, messages

Desktop Streaming (private beta)

For runners with browser or desktop capabilities, click Desktop to view the runner’s screen in real-time via VNC. Supports:

• Live screen viewing (up to 2 concurrent viewers)
• Mouse and keyboard input
• Region screenshots for debugging

Managing Multiple Runners

The Runners list page shows all runners with:

• Status indicators (green = running, yellow = provisioning, red = errored)
• Cost per runner (cumulative session cost)
• Last activity timestamp
• Quick actions (start, stop, destroy)

Fleet View (private beta)

Fleet orchestration is in private beta. For organizations running multiple runners under an allowlisted plan, the fleet view aggregates:

• Total runners by status
• Aggregate cost across all runners
• Capacity — how many more runners your plan allows
• Health — runners with errors or high resource usage

Snapshots

Create point-in-time snapshots of a running container for:

• Backup — save state before risky operations
• Branching — fork a runner to test different approaches
• Rollback — restore to a previous known-good state

From the runner detail page, click Snapshots to create, restore, or branch.

Skills Management

Install skills from ClawHub directly from the dashboard:

1. Navigate to the runner detail page
2. Click Skills
3. Browse or search the ClawHub registry
4. Click Install — the skill is installed on the running container

The security scanner checks every skill before installation for malicious patterns.

Runner Configuration

Update runner settings from the detail page under Settings:

• Governance policy — override org-level governance for this runner
• Environment variables — inject secrets and configuration
• Network policy — control egress (which external services the runner can reach)
• Idle timeout — auto-stop to save costs
• Session budget — per-session cost ceiling

Troubleshooting

Next Steps

• Runners Quickstart — create your first runner
• Runners Security — sandbox isolation and network policies
• Connect Your Machine (BYOVM) — bring your own compute
• Runner Costs — cost tracking in the dashboard
• Runbook: Runner Stuck — troubleshoot stuck runners