Readiness and cleanup
Check Docker readiness and remove resources left by crashed runs
stashbase agent docker doctor # docker CLI, daemon, and default image status; non-zero exit if not ready
stashbase agent docker status # list leftover per-run networks
stashbase agent docker cleanup # remove leftovers after confirmation (--yes to skip)When a check fails, doctor says how to fix it, for example sudo systemctl enable --now docker to start the daemon on Linux.
doctor also shows how old the default image is and which Claude Code and Codex versions it contains. To update them, see Update Claude Code and Codex.
doctor and status accept --json. For a Docker-backend profile, agent validate checks Docker readiness instead of the native backend.
Every run's network and containers are named after its session ID (ags_...), so you can match leftovers to their audit logs. A normal exit or Ctrl+C removes them. Leftovers appear only after a crash or SIGKILL. cleanup skips networks that belong to live local sessions. A --remote session has no local record, so cleanup shows creation times to help you avoid removing one that's still running.