shutdownlisted

@markdownai # /shutdown — Willow 2.0 Graceful Close Stack position: this skill is the **end-of-session persistence** layer — it absorbed the former `/handoff` skill, so the handoff write is step 2 of this sequence, not a separate skill. See the Persistent memory section in `willow.md` for the full 4-layer stack. ## Sequence 1. **Close scan (mechanized)** — run the deterministic pre-handoff scan: ``` willow_run(app_id, allow_net=True, task="python3 -m willow.fylgja.close_scan {AGENT} --apply") ``` One pass does what used to be three model loops: closes provably-finished `process-*` flags (pid gone), reconciles every PR-shaped open thread against live `gh pr view`, and lints MEMORY.md for entries missing KB atom IDs. Read its JSON: - `flags.still_running` → update each flag with current progress (`soil_put`) so the next session reads accurate state. `flags.ambiguous` (no pid; log tail attached) → judge from the log tail; close or annotate manually. A handoff over a stale flag is a lie. - `threads.drop` → these PRs are MERGED/CLOSED and must **not** appear in the handoff. If one left genuine follow-up work, write *that follow-up* as the thread — never the merge. - `threads.keep` → carry forward with the refreshed `pr_status`. `threads.no_pr_ref` → carry or close on judgment. - `memory.missing_atom_id` → fix in step 3. 2. **Write the handoff** — this runs early by design: it is the one artifact that must survive even if th