restic Restore

Guided restore of files, volumes, or databases. Restore is destructive when applied in place — confirm before overwriting anything.

Parse [--what <path>] [--snapshot <id|latest>] [--target <dir>] from "$ARGUMENTS" (defaults: snapshot latest, target /restore).

Process

Follow the disaster-recovery skill. For a full server rebuild, hand off to the restic-backup-engineer agent.

1. Confirm access — set -a; . /root/.restic/r2.env; set +a then restic snapshots. If the repo can't be read (no password / keys), stop and explain that recovery is impossible without them.
2. Locate — restic ls <snapshot> / restic find <pattern> to confirm what will be restored.
3. Plan & confirm — show the user exactly what will be restored and where. Restore to a staging target by default (/restore), not in place. Get explicit confirmation before any in-place / --overwrite restore.
4. Restore files: restic restore <snapshot> --target <dir> [--include <what>].
5. Restore a database (if requested): stop app containers (not the DB), then restic dump <snapshot> <dump.gz> | zcat | docker exec -i <db> ... (see disaster-recovery).
6. Verify — diff a known file, check DB rows / app health. Report what was restored.

Never run restic init here — reconnecting to an existing repo uses snapshots, not init.