document-backup-strategy

Document Backup Strategy

Produce the living document that every future "how do we back this up" question refers back to. Written once the user has picked an option from 03-options.md.

Required sections

Write to backup-plan/BACKUP-STRATEGY.md (capitalised so it's visible). Include:

1. Summary — one paragraph: what's protected, where it goes, who owns it.
2. Scope — what is in scope, what is explicitly out of scope (and why).
3. Data classes and targets — table mapping each asset → RPO → RTO → tool → destination → retention.
4. Schedule — when each job runs (cron expression or equivalent), approximate runtime, when it must not run (backup windows).
5. Retention policy — grandfather-father-son or forget-policy in tool-native terms (e.g. restic --keep-daily 7 --keep-weekly 4 --keep-monthly 12 --keep-yearly 5). Align with compliance requirements.
6. Encryption & key management — where keys live, who can decrypt, what happens if the primary operator is unavailable.
7. Monitoring & alerting — how success/failure is detected, where alerts go, the SLO for responding to a failed job.
8. Access and credentials — which accounts/keys each job uses, least-privilege notes (e.g. restricted B2 application key, IAM policy scoped to one bucket/prefix).
9. Testing cadence — when restores are drilled (see restore-plan skill). Untested backups are not backups.
10. Change log — initial entry with today's date and approving user.

Quality bar

• Every schedule is concrete (no "regularly" / "periodically").
• Every destination references a real entry from the backup-infra memory.
• Every asset from 02-data-inventory.md is either covered or listed as out-of-scope with justification.
• No secrets in the document. Reference credential locations by name only.

After writing, append a pointer to the main project README.md under a "Backups" heading so future contributors find it.