updatelisted

# arlo:update — re-sync arlo with the ground it stands on arlo's whole reason to exist is that stored command knowledge rots: a hand-maintained table said `--parent` for four months after the real flag became `--parent-id`, and everyone who trusted it created orphans. A `.arlo/cards.json` left alone becomes exactly that table. So cards are **a regenerated projection of live ground truth, never stored truth** — and this skill is how you regenerate them. Run it periodically, and after anything changes the project's commands. Update reconciles **this project** against the arlo you already have installed. It does *not* fetch a newer arlo — that is **/arlo:upgrade** (pull the latest arlo from upstream, then run this to reconcile). Keep the two axes distinct: `upgrade` changes *arlo*; `update` re-reads *your project*. `REFERENCE.md` (bundled beside this file) carries the full harvest rules (§2) and gradient. ## Steps 1. **Sync the project's core to the installed arlo.** Copy the core the lights-out `arlo` CLI runs (under arlo's home) from the *currently installed* plugin, so the project runs the arlo you have. (If you want a *newer* arlo — new harvest rules, a new rung, a sharper ranking — run **/arlo:upgrade** first; it fetches the latest arlo, then runs this step.) 2. **Re-harvest from the *current* live sources.** Re-run the harvest named by the project's card spec against the sources as they are right now — the same generation as setup, run fresh — into a n