# Over

> Use as load-bearing work lands, or when the user asks to debrief, hand off recent work, or check what they actually own versus what AI carried. Triggers on "debrief", "what do I own", "hand over", "hand it back", "over", and recently shipping a load-bearing decision the human has not reasoned through.

- Skill: `majiayu000/over` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds add majiayu000/over`
- Raw SKILL.md: https://api.skillmd.com/api/skills/majiayu000/over/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: majiayu000 (https://skillmd.com/u/majiayu000)
- Updated: 2026-09-09
- Page: https://skillmd.com/skills/majiayu000/over

---


Hand a load-bearing decision back the moment it lands — in-flow, never banked. The human reasons it into their own model; you confirm once. The signal is the work just done, not a stored ledger.

## The gate

Run before anything else. Survey the work just done for decisions that pass the handover test (below) AND were not already handed over — a build-time probe counts as handed over.

- **Quiet when nothing is load-bearing.** Nothing load-bearing, or every load-bearing decision already reconstructed → exit in one line ("nothing load-bearing to hand over"). Never manufacture a handover.
- **The work is the source.** Decisions come from what just happened in the thread — the work, its constraints, the human's words — never from git history or a stored ledger (there is none).

## Multiplicity

Unlike the build-time probe — one handover per delegation — an invoked sweep may hand back several at once. Take every un-reconstructed load-bearing decision, highest-blast-radius first, so the attention spent lands on the biggest if the human stops early. Each is its own single teach-back.

<!-- doc-gen FILE src=../handover.md -->
## The handover

A held mental model can't be handed to another mind — only built there by reasoning. When verified work carries a decision worth owning, hand it over: the human reasons it into their own model, you confirm once.

- **The handover test.** A decision is worth a handover only when it is load-bearing: the recoverability test holds — code indifferent between ≥2 paths, the human's goal (not a retrievable fact) broke the tie — AND it is long-lived / load-bearing. Mechanical, re-derivable, or local-and-cheap decisions are never handovers.
- **Probe forward, generated from the work.** Pose ONE question per handover, built from the decision's novel move and the human's own words — never a fixed template. Aim it forward: what they would weigh the next time this pattern returns, or what they would tell someone meeting this code in six months — not "was this right?". The forward question elicits the rule; re-justifying the past does not.
- **Confirm by reframing, once.** The human answers; you reframe their answer through the decision's actual tradeoff — name the risk it accepted — and correct once. Not a yes/no stamp. One question in, one reframe out.
- **Stateless.** The exchange lives in the thread; nothing is stored — no rule saved, no profile kept. All signal is live.
<!-- end-doc-gen -->

## Invisibility

Never surface scores, streaks, mastery labels, or counts. A handover that gets graded stops measuring understanding and starts manufacturing performance.