Manage the Space with the CLI
Use the lifecycle command that matches your intent: reconcile the release, stop workloads without deleting data, resume them, or permanently reset the complete Local Space.
Update the Space or the standalone CLI
Run shimpz update to apply a newer atomic Local release to the complete managed Shimpz Space,
including its release-bound CLI. The atomic Local release is the only update authority for that managed
executable. The command checks the replacement before switching to it. If the candidate fails that admission
check, the installed release stays unchanged and nothing is remembered, so the next update checks the same
release again after you fix the reported cause. If an admitted candidate replaces the installed release but does
not become healthy, Shimpz attempts to restore the previous working release and remembers the failed release, so
later updates report it instead of retrying it until a different release is selected. The command reports whether
that restoration succeeded; if it reports that the rollback also failed, follow the printed cause before updating
again. Success reports Shimpz Space is ready with the local Admin address and active
release. When no newer release exists, it reports that the installed release is current and changes nothing.
shimpz update Update never resumes a Space you stopped with shimpz stop. For a stopped Space it only reports
whether a newer release is available and preserves the stopped intent; run shimpz start when you
want to resume the Space and apply the current release. Update also does not repair the installed release. Use shimpz install to install or repair the complete Local Space; it clears a stopped intent and starts
the workloads.
To upgrade a standalone CLI used outside a managed Local Space, reinstall it with cargo install --locked shimpz-cli (Rust 1.97.1 or newer). shimpz upgrade prints this
command and never replaces the CLI itself. A standalone upgrade never updates Admin, Team, Brain, network
boundaries, or managed Space state. A Space-managed CLI is updated only through shimpz update.
cargo install --locked shimpz-cli If the CLI-owned Local Space exists but its current Docker runtime fails validation, the CLI names the failure, lists the exact owned scope, and asks whether to remove that corrupt Local state and install a fresh Space. Yes permanently removes every listed resource and the installed Local state before continuing in the same terminal. Any listed Team, Assistant, credential, setting, conversation, or Docker data is lost. No, Enter, or a non-interactive run leaves it unchanged. This recovery choice is available only for a corrupt CLI-owned Space. A running, attested Space requires a fresh, two-minute host authorization projected by the CLI. While the current Supervisor authentication record is usable, reset also requires that password. Before initial setup, Admin and Team verify that no Supervisor identity exists and reset continues without a password prompt. If the authentication record itself requires recovery, the host authorization can proceed only through the existing Supervisor identity or Team's independent proof that none exists.
Stop workloads without deleting the Space
shimpz stop stops the owned Space services and installed Assistant workloads. It keeps every Team,
Assistant, Integration, setting, conversation, container, network, volume, and release binding. Stopping changes
only availability, so it does not require the Supervisor password. Automatic maintenance leaves an intentionally
stopped Space stopped.
shimpz stop shimpz status The stop is complete when status prints Shimpz Space is stopped. Run shimpz start to resume the same Space. Start reconciles the current atomic release and restores each valid installed
Assistant; if local configuration changed while stopped, it is reconciled before workloads start.
Reset only when you want to delete everything
Reset removes the managed Assistants, Teams, Integrations, Admin state, and local Shimpz data. It cannot be undone. Disconnect provider Integrations first when possible. When the owned Admin is running, reset attests its loopback listener and projects a fresh, exact-Space host authorization that expires after two minutes. While the current Supervisor authentication record is usable, reset requires that password as well before asking Admin to delete domain state. Before initial password setup, Admin and Team verify that no Supervisor identity exists and reset continues without asking for one. If the authentication record requires recovery, host authorization uses the existing Supervisor identity when valid; otherwise Team independently requires that no identity exists. If no managed runtime remains, the CLI can finish removing only the validated scheduler and files, plus any owned Linux encrypted-storage residue, without a password. Ambiguous or foreign state inside a Shimpz-owned resource stops reset with an error. Unrecognized content outside the owned scope is preserved and listed in the successful result. Stopping first does not bypass reset authorization.
shimpz reset Reset is complete only when the CLI prints Shimpz Space was reset successfully.