What problem does it solve? After a rollback, rebase, or partial restore, teams often lose track of which features silently disappeared or regressed. This Skill provides a disciplined workflow to compare a checkpoint against the current branch, distinguish genuinely missing behavior from intentional differences, and restore only what is actually broken without overwriting current improvements. ## Core Features & Use Cases - Protected baseline creation: Creates a named branch and tag at the current HEAD before any edits so recovery work is always reversible. - Audit-driven recovery: Runs the repository's checked-in recovery audit (pnpm run audit:recovery) and classifies each manifest entry as PASS, MISSING, or intentional difference before restoring anything. - Selective incremental restoration: Restores only confirmed missing features in small slices, updating contracts, generated clients, tests, and runtime wiring, then re-runs the audit after each slice. - Use Case: After a bad merge drops a report route's regression test, use this Skill to preserve a baseline, run the audit, confirm the test is genuinely missing (not intentionally removed), restore the smallest focused test, and produce an evidence-backed completion report. ## Quick Start Ask the AI to recover any behavior lost since a specific checkpoint by running the recovery audit, preserving a named baseline, and restoring only confirmed missing features with evidence.
