Backup Now Directions¶
Take an immediate manual snapshot of work-buddy's vital SQLite databases as an anchor point before a risky operation.
When to use¶
User wants a fresh point-in-time snapshot of the task store and the other vital SQLite databases. Typical anchor points: right before a destructive operation, before swapping branches, before running a long script that writes to the store, or just because it's been a while.
Slash command: /wb-backup-now
Directions¶
Take an immediate manual snapshot of work-buddy's vital SQLite databases. Useful right before doing something risky (a destructive operation, an experimental migration, etc.) so a known-good point-in-time exists to restore to if things go sideways.
Manual snapshots get a -manual suffix on their snapshot ID and live in a separate retention bucket from the hourly rolling snapshots, so this won't be swept away by routine cron churn (cap: 20 manual snapshots; deliberately small -- anchor points, not archival).
If backups.github.repo is configured AND gh is authenticated, the snapshot also pushes to GitHub Releases. Otherwise the snapshot stays local-only and a warning surfaces in the Settings -> Backups card.
How¶
Call the capability directly:
`mcp__work-buddy__wb_run("data_backup", {"manual": true})`
The response shows the snapshot ID, tarball path, size, and whether the remote push succeeded.
When to use¶
- About to run a
mcp_registry_reloadafter editing a sensitive capability. - About to run
task_archiveor another high-fan-out task-mutating operation. - About to swap branches, run a long-running script that writes to the store, or restore an older state.
- Just want a fresh off-machine copy because it has been a while.
For automatic protection between manual snapshots, the hourly cron (sidecar_jobs/data-backup.md) takes a rolling snapshot every hour. See architecture/backups for the full subsystem.