sdd-archive

Imported from IrvinngB/keel (adapters/codex/skills/sdd-archive/SKILL.md). Install upstream with npx skills add IrvinngB/keel --skill sdd-archive. Copyright stays with the author.

You are the SDD orchestrator. Delegate to the phase sdd-archive subagent; do NOT archive inline.

1. Resolve the change (the user's arguments or the single active change; ask if ambiguous).
2. LOAD <change>/verify-report. If missing or FAIL → STOP, tell the user to run command verify (and fix CRITICALs) first.
3. Launch phase sdd-archive: it merges the <change>/spec/* deltas into specs/<capability> and SAVEs <change>/archive-report. Re-running is safe: if <change>/archive-report already exists the merge is skipped and only the remaining steps run (a retry after an interrupted run just finishes the state flag and layout). Then SAVE <change>/state with status: archived and archived_on (files backend: finalize by moving the change folder, per its doc).
4. Present the closure summary: capabilities synced, task completion, where the archive lives (backend-specific).
5. Suggest command postmortem — the closed cycle is exactly what to learn from.
6. Launch phase sdd-steer to refresh steering docs with the new reality (and apply any previously APPROVED lesson proposals).
7. If the backend is shareable (files), suggest committing the store so the team gets the updated source-of-truth specs.

The archive is an audit trail — never delete or modify archived changes.