Rescue-Tag-Before-Destructive-Operation

These examples and illustrative results are independently authored teaching materials, not measured model results.

Use case

An authorized teaching history rewrite has staged, unstaged and new files. A tag references a commit, not uncommitted data.

Mechanism

Record HEAD/branch and dirty-state inventory. Create a unique verified recovery ref without overwriting existing refs. Separately preserve index/worktree deltas and required new/ignored files with sizes/digests/restoration steps. Test restoration in an isolated checkout before the authorized destructive operation; insufficient backups leave original state intact.

Bad example

Create a rescue tag and assume uncommitted files, Git configuration and external data are all protected.

Good example

Bind a unique recovery ref to original HEAD and separately save staged/unstaged diffs and required new files. Reconcile the inventory and test recovery in a copy, declaring uncovered state. Recovery artifacts do not expand deletion permission.

Why the change matters

Commit refs and working files occupy different layers. Separate backups/restoration tests prevent returning to an old commit while losing current work.

Observable expectation

The teaching restoration copy has original HEAD, both deltas and new-file bytes/hashes. A correct tag with missing new files is incomplete. Rehearse in isolation, not by destructive cleanup in the original tree.

Limits

No frozen source is confirmed; this is an independent teaching construction. Refs/backups can fail or disappear, and secrets/external state/unlisted ignored files are not automatically protected. Backups do not grant destructive authority.