Data Restore

Restore work-buddy's vital SQLite DBs from a snapshot. Validates the manifest (refuses if the snapshot's commit or schema is newer than the running code), unpacks to staging, runs migrations forward, verifies integrity, then atomically swaps into place (the old DBs are moved to .data/db.pre_restore_/ for safety).

Skill name: data_restore

Category: backups

Mutates state (retry policy: manual)

Parameters

Name	Type	Required	Description
force	bool	No	Override safety checks (newer-schema, newer-commit). Use sparingly — these checks exist to prevent silent corruption.
from_remote	bool	No	If True, download the tarball from GitHub Releases first (via gh release download). Default False (local-only).
snapshot_id	str	Yes	Snapshot ID (e.g. 'snap-2026-05-11T14-23-00Z') or, if from_remote=True, a GitHub release tag.

Details

The installed-authority ledger is restored without rewriting its irreversible domain rows. Because those rows bind absolute database-path hashes, a restore to a different machine or data root deliberately leaves the affected native stores fail closed. Inspect the central Source Foundation restore status, then run source_foundation_restore_operator with rebind_installed_authority=true under its exact high-consent approval. That operator proves every sealed cohort at its configured target path and archives a content-free rebind receipt with the restore fence; ordinary data_restore never silently relocates authority.