# Leader OS — the Governance Sprint, made agentic

The leader pathway's promise is a nurse manager with a **defensible method** — not a vendor demo. This pack is that method as a working system: skills that draft your unit communications for your approval, scaffold QI work, analyze only what's safe to analyze, keep a decision ledger, and make you the unit's AI-governance leader with the review-gate machinery from the Governance Kit at your fingertips. The Governance Kit is a separate download: grab `governance-kit.zip` from the Resource Hub's Safety & governance section and unzip it into your `My-Nurse-AI-OS/` folder, so it sits alongside this pack as `../governance-kit/`.

**What's inside**

| File | What it does | Tier |
|---|---|---|
| `Weekly-Brief.SKILL.md` | Drafts the unit brief from your notes — staff-facing, so nothing goes out without you | 🟡 Yellow |
| `QI-Coach.SKILL.md` | PDSA/A3 scaffolds, honest measure definitions, run-chart literacy | 🟢 Green |
| `Staffing-Lens.SKILL.md` | Workforce-planning analysis on anonymized aggregates only — refuses named-staff analysis | 🟡 Yellow |
| `Policy-Drafter.SKILL.md` | Policy and education drafts with cited source standards; adoption stays with your committee | 🟡 Yellow |
| `Decision-Journal.SKILL.md` | The leader's ledger: decisions, rationale, and review dates, kept without drama | 🟢 Green |
| `Governance-Facilitator.SKILL.md` | Tier classification, gate design, the rubber-stamp check — your governance co-pilot | 🟢 Green |
| `Leader-Boundary-Header.md` | The shared boundary all six skills load | — |
| `First-Governed-Workflow-Week.md` | The one-week sprint: one governed workflow + your unit's AI accountability charter | — |
| `Unit-Incident-Response-One-Pager.md` | When an AI-assisted workflow goes wrong: detect → contain → learn, sized for a unit | — |

**Setup (2 minutes).** Tell your AI: *"Load the six skills in 17-Leader-OS, including the Leader-Boundary-Header they share. Apply my Do-Not-Remember rules with the Manager/Leader extension. Then open First-Governed-Workflow-Week.md and start Day 1."*

**The rules that make this safe to lead with.**
- **People decisions are never the AI's** (🔴, permanently): hiring, evaluation, discipline, and anything that affects an individual's employment is a human decision, humanly delivered.
- **Aggregates only.** Skills here work on anonymized, aggregate, non-identifiable data. Named-staff performance analysis is refused by design — proxy metrics launder bias into people decisions.
- **Staff-facing and org-facing communication is 🟡 minimum** — drafted, then approved by you before anyone else sees it.
- **Employer-data integrations are 🟠 deferred**: no HRIS, EHR, scheduling-system, or employer Teams/Slack connections without institutional/compliance approval, routed through the [vendor & integration questions](../08-Integrations/Skill-and-MCP-Vetting-Checklist.md).
- If you've approved every AI draft unchanged for a month, that's not efficiency — that's a rubber stamp. `Governance-Facilitator` will say so (the Governance Kit's gate-health rule).

**Outcomes worth tracking** (adoption signals only — never staffing-outcome, safety-outcome, or compliance claims): governed workflows in use, gate health (edits and rejections vs. unchanged approvals), time-to-draft you reclaim, decision-journal completeness.

> Agents propose. Humans judge. Nurses steward. · No PHI, no personnel files, ever.