apw-handofflisted

Use this skill to carry **in-flight task state** across a context reset. It exists to fight one specific limitation: a model's **context is finite, and compaction or a new session erases mid-task state** — what was done, what was about to happen, which decisions were made and why, and which approaches already failed. Durable *outputs* survive in the work item's artifacts and `apw-learn` persists durable *lessons*, but the working state between them evaporates — so a fresh session re-derives the plan, re-makes settled decisions, and re-walks dead ends. The fix is a checkpoint artifact: write the state down before it is lost, and re-enter through it instead of from zero. The skill has two modes — **checkpoint** (save state on the way out) and **resume** (restore state on the way in). Determine the mode first, then follow that mode's steps strictly and in order. ## Output style When this skill asks for a plan, report, summary, status update, or other chat-facing output, prefer bullet points or numbered lists over plain paragraphs. Keep each item concise and content-bearing. Omit language that only smooths tone, repeats context, or adds "language sugar" without changing the meaning. ## Step 1 — Determine the mode - **Checkpoint** — a session is ending mid-task: the user is stopping for now, context is close to compaction, or you are partway through work that another session will finish. The deliverable is an up-to-date `handoff.md`. - **Resume** — you are picking up a task w