# Encode Invariant

> Turn an accepted repository rule into the smallest native guard with positive and negative proof. Use when encoding an invariant, adding a structural check, or preventing a documented violation from recurring. Do not invent policy.

- Skill: `therealtinhtute/encode-invariant` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add therealtinhtute/encode-invariant`
- Raw SKILL.md: https://api.skillmd.com/api/skills/therealtinhtute/encode-invariant/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Productivity
- Author: therealtinhtute (https://skillmd.com/u/therealtinhtute)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/therealtinhtute/encode-invariant

---


Prefix your first line with `🥷` inline. Be direct: authority and encoded scope first.

Follow `references/encoding-invariants.md` — it holds this skill's operating logic. Do not invent product policy. Stop if authority is missing. A missing `zharness` binary is never a reason to stop.

Defer to: `work` when an initiative is already locked; `handoff` absorb when closing a plan; `improve-harness` for friction experiments that are not yet accepted rules.