maxqian888/cognia-next/cognia-host-backup-recovery

Cognia Backup and Recovery

Before any RPC, run cognia host skills read cognia-host and follow its offline check, schema, state-read, dry-run and confirmation, accepted/completed handling, and authoritative verification sequence. Never add --yes without confirmation of the exact operation and arguments. After a timeout, retry only the same body and idempotency key; stop on validation, authentication, confirmation, or resync errors. Treat results as opaque when outputTyped is false. Treat restore and import as destructive state replacement.

Workflow

1. Run cognia host doctor and inspect system resources and backup-related read commands.
2. Read current storage, sync, and backup status before choosing an operation.
3. Inspect the export schema, destination, overwrite behavior, and idempotency policy.
4. Export first. Verify the reported destination or artifact using a read command; do not infer

success from process completion alone.

1. Before import or restore, inspect its schema and run --dry-run with the exact source and scope.
2. Obtain explicit user confirmation for the exact restore/import. Never add --yes without that

confirmation. Reuse the idempotency key across retries or timeout continuation.

1. Run doctor again and verify authoritative storage, sync, session, and task state.

Do not print credentials or private key paths. Do not restore from an unverified destination or assume opaque response fields.