Recovery evidence — anonymized historical field note
Fieldwork: September 2026. Founder-assisted work, not an automated Hetalos run.

Observed: development and production backups restored into separate, new
temporary projects. Selected catalog fields matched the respective observations.
Temporary destination cleanup was read back. The original production database
was not overwritten by these restores.

The production request-to-observed-online interval was approximately 4m55s.
This was not service-ready RTO. Recovered-data cutoff and achieved RPO were unknown.
Customer-row completeness, application function, full first-start isolation and
tenant isolation were not established. Containment was required after creation
of the production trial; original production settings were unchanged.

Technical-result review, adopted recovery scope/targets, minimum service-function
validation, control closure and customer/auditor acceptance remained separate.
No current provider status or customer recovery guarantee is represented.