crewship reset
Reset requires a stopped installation and an explicit local root. It never selects a target from cwd, a CLI profile or the fallback user home.
Stop using the installation’s supervisor, run reset, start through that same
supervisor, wait for
/readyz, then run crewship seed --server <explicit-url>.
Provider-backed seeding requires credentials. --offline-demo seeds UI fixtures
without model execution. Reset itself does not stop or restart services.
Configuration, secrets, logs, backups and installation identity are retained.
Application rows return to the migration defaults of this binary; the database
schema, migration metadata and identity nonce survive. Configured managed stores
and installed chats/skills are cleared; Bolt contents are reset.
Only exact installation-labelled containers, volumes and networks are removed.
Foreign or unlabelled Docker resources, images and build cache are retained.
Docker inventory is required; failures prevent clearing application data.
External stores, unproven custom layouts, checkout roots, hard-linked database
files, symlink escapes and overlaps with preserved management data are refused.
Reset holds lifetime installation, SQLite and identity locks. If it fails after
beginning cleanup, startup refuses the installation until reset completes.
Keep it stopped and rerun the same command. No lock files or reset markers should
be manually deleted to bypass this protection. SQL row replacement is
transactional; Docker and filesystem cleanup can have completed before a later
failure. Apple container cleanup is currently unsupported.
This command is available in the full server binary, not crewship-cli.