Recovering an instance

Three console commands are the supported way back into an instance with no reachable administrator. Run them in the web container (docker exec, or the platform’s console). All three are idempotent: running one against an account that is already in the desired state prints “already …” and exits 0.

Command	What it does
app:admin:create <email>	Ensures the email is a verified administrator, creating the account if it does not exist. Options: --full-name, --password. With no --password it prompts (or, non-interactively, generates one and prints it once). An existing account is promoted and verified in place and keeps its password.
app:user:promote <email>	Grants ROLE_ADMIN to an existing account, keeping any other roles.
app:user:verify <email>	Marks the account’s email verified and burns any outstanding verification token — including on an account that was already verified, since that link logs its bearer straight in. The escape hatch when outbound mail never arrives: an unverified account is parked on the check-email page and cannot reach the admin area.

Non-interactive first admin on a fresh instance:

docker exec <web-container> bin/console app:admin:create you@example.com
# → Created administrator you@example.com.
# → Generated password (shown once): …

A shell-recovered instance has no feature-flag rows at all, and that is fine: every install flag falls back to exactly the value the wizard would have written (registration on, no cap, billing and social login off, a 14-day trial), so the instance behaves like a wizard-installed one with the defaults accepted. To change any of them, visit the admin Feature flags page — its scan view lists every flag the code references but the database does not define, and creates the rows on request.