Recovery test press / v0.1.0

A backup badge is not a restore.

Run the real restore. Check what your application needs. Destroy the isolated target. Keep the proof—not the data.

Run your first drill Read the source

LocalBackup never leaves your environment
ExplicitExact target confirmation on every run
PortablePlain TOML in, plain JSON out

A four-stage halftone proof press restores a database cylinder, checks it, destroys the copy, and stamps an evidence sheet. — Fig. 01 — Restore → check → destroy → attest

Put it beside your restore tooling

One binary. Your commands.

Attestor does not store backups or speak to your provider. It orchestrates the restore path you already own.

shell / restore-drill

$ cargo install restore-drill-attestor
$ restore-drill validate --config restore-drill.toml
$ restore-drill run --config restore-drill.toml \
    --confirm local-restore-drill --output ./attestations

Recorded-output simulator

See the evidence, not the rows.

The browser demonstration uses invented timings and never connects to a database. The CLI writes the same evidence shape after real commands run locally.

Ready to run an isolated sample.

RDA / ATTESTATIONNOT RUN

Drill: primary-postgres
Target: local-restore-drill
Evidence: outcomes + durations only

1Prepare targetWaiting—
2Restore backupWaiting—
3Run 3 checksWaiting—
4Destroy targetWaiting—

No commands · No stdout · No stderr · No data values

Compact by design

Proof you can hand over.

The attestation shows whether recovery worked, how long each operation took, the configuration fingerprint, and when the drill becomes stale.

includedPass/fail state

includedStage durations

includedFresh-until date

excludedDatabase values

excludedCommand output

excludedSecrets

A hard boundary, not a checkbox

Production-looking targets are refused.

The config must declare an isolated target. The run must repeat its exact ID. Names containing prod, production, or live are rejected before any command starts.

Attestor proves restore mechanics. It does not provide network isolation or ransomware protection; those remain your responsibility.

Optional one-time unlock

The CLI stays useful. The pack standardizes the team.

Operator Pack adds local scheduling recipes, a recovery review agenda, and a multi-database evidence policy template. Pay once; keep the files.

GitHub Actions and cron scheduling recipes
30-day evidence review checklist
Multi-database ownership matrix template

Sociobot/Dodo is the merchant of record. Refunds are handled there and revoke the license automatically. Terms · Privacy

Operator Pack

^$39one-time purchase Buy the Operator Pack

LICENSE ACTIVEOperator Pack / local copy

Recovery evidence policy

cadence_days: 30
failure_owner: on-call-platform
required_checks: [row_count, schema, application]
evidence_retention_days: 365
escalate_after_failed_drills: 1

GitHub Actions schedule

schedule:
  - cron: "17 3 1 * *"
workflow_dispatch:

Monthly review agenda

Confirm each protected database has evidence under 30 days old.
Review restore and cleanup durations for drift.
Assign every failed drill before closing the review.

Do not wait for the incident.

Make restore confidence a file you can show.

Install Restore Drill Attestor