# Init

> Read the two skills this system is entered by, follow every skill either one names to the end of the chain, and run the waking check. The start of a conversation, a project's governing file directing that a stretch of work opens here, "memento:init", "init", "start here", "orient", "load memento".

- Skill: `ryzhakar/init` (Agent Skill)
- Install (CLI): `npx skillmds@latest add ryzhakar/init`
- Raw SKILL.md: https://api.skillmd.com/api/skills/ryzhakar/init/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: ryzhakar (https://skillmd.com/u/ryzhakar)
- Updated: 2026-09-21
- Page: https://skillmd.com/skills/ryzhakar/init

---


<read-every-skill-in-the-chain>
Read `pat-down` and `span-closure`; never read one without the other.

Read every skill either one names, and every skill those name, to the end of the chain; never stop at the first skill named.

Skip a skill already read in this stretch of work; never read one twice.
</read-every-skill-in-the-chain>

<separate-reading-from-running>
Separate reading a skill from running it; never take a skill read for a skill run.

Run `pat-down`; never defer it past this reading.

Hold `span-closure` unrun at a start; never run it on reading it.

Enter `span-closure` where the end of a stretch of work is signalled ahead of it; never enter it at a start.
</separate-reading-from-running>