Open Agent Harness Start

Operations

From local loop to production runbook.

Operations deserves its own lane: deployment shape, configuration, production readiness, Kubernetes rollout, and incident recovery should not be hidden inside architecture or roadmap pages.

Embedded first Keep local development simple while preserving the split-deployment path.
Stateful by design Treat PostgreSQL, Redis, and workspace backing store as explicit operational dependencies.
Runbook owned Production docs stay close to rollout checks and failure recovery steps.
01 Configure
02 Validate
03 Deploy
04 Observe
05 Recover
Track Local operation

Start the daemon, WebUI, and TUI, then validate the full workspace loop.

Track Deployment shape

Choose embedded workers first, then split API, controller, queue, and worker planes when scale needs it.

Track Production readiness

Review storage, credentials, observability, backups, rollout, and failure-mode requirements.

Track Kubernetes rollout

Use the checklist and runbook as the deployable operational contract.

Runbook shelf

Keep deploy-time facts close to failure recovery.

These pages form the practical operating surface: configuration, split deployment, queue reliability, and the Kubernetes runbook.