Q3 E2E Synthetic Data and Cleanup

Principles

  1. Synthetic data must be attributable to exactly one LTI run.
  2. Mutable scenarios target acceptance, sandbox, test mode, or development only.
  3. Cleanup narrows by run-owned identifiers and creation time. It never widens after a lookup failure.
  4. Secrets and raw DR tokens are runtime inputs, not evidence.
  5. A cleanup failure is a visible run failure or manual-review state.

Run Identity

Each run records:

Do not use personal employee or supporter addresses as synthetic identities.

Campaign Fixtures

The current baseline contains 450 seeded acceptance petitions and 90 sandbox donation forms across 30 production segment definitions. Backup 1784327649439 predates that seed.

DR Tokens

Donations and Providers

Cleanup Order

  1. Stop any active browser/provider step and record terminal evidence.
  2. Complete provider-specific cleanup only when explicitly supported.
  3. Remove exact acceptance member/signature rows only after verifying the unique synthetic email and creation window.
  4. Remove run-owned temporary files and local browser state.
  5. Record complete, partial, manual_review, or not_applicable cleanup status.

Broad SQL deletion, table reset, truncation, wildcard email deletion, and production cleanup are prohibited.

Retention

Recovery

If seeded acceptance content or schema becomes invalid, stop mutable runs, create a fresh backup if the database remains coherent, and restore the last verified acceptance backup. Restoration requires operator approval and must verify database identity before execution.