k2f-policy-document

The saved excerpt is a snapshot from when this skill was listed. The external source remains the complete and most current version.

---
name: k2f-policy-document
description: Draft employee- and organization-facing policy documents—handbook sections, HR rules, conduct, benefits summaries, and compliance procedures—from user source material, outlines, or legacy policy text. Use when the user asks for a policy document, employee handbook section, HR policy, workplace rule write-up, or formal policy refresh. Create policy documents with AI-native design and the highest aesthetic standards, powered by the K2F format and engine. Purpose-built for AI agents, K2F lets them focus on semantic content and design intent while automatically handling layout, typography, and formatting complexity.
---

# K2F Policy Document — AI-Native Aesthetic Design

## Purpose

Teach agents to blueprint **policy documents** before layout: scannable rule architecture, plain-language layers for employees, and formal hierarchy—without inventing benefits, legal obligations, jurisdictions, or approval paths the user or HR/legal did not supply.

## Built on K2F

This skill produces policy documents as `.K2F` packages. Follow the **k2f** skill for authoring: unpack or init a workspace, edit the semantic tree, run pack/verify, then export. K2F keeps agents on *meaning* (sections, emphasis, tables)—not pixel coordinates—while the engine compiles layout, typography, and theme into consistent export-ready output.

Before you build, install the core skill: `npx skills add suzheng/k2f --skill k2f`.

## When to Use

- New or refreshed **single-policy** pages (PTO, remote work, expenses, conduct, travel, security, leave types).
- Handbook **sections** that must read consistently with house style but stay findable on first scan.
- Reformatting draft or pasted policy language into a publishable structure for HR or legal review.
- Not for chat-only policy Q&A, AgentRC scoring policies, or counsel-grade contracts (use **k2f-contract**).

## Workflow

### Step 1. Generate the content_and_design.md

One blueprint pairing **information architecture** with **design intent**. Label blocks the **k2f** skill can map to sections, callouts, and emphasis later.

**Sub-steps**

1. **Document frame** — Policy title, document ID if user provided, **effective date**, version, policy owner (team/role), and **applicability** (entities, locations, employment types). State whether this is **new**, **amendment**, or **supersedes** prior text; link prior version only from user inputs.

2. **Purpose & scope** — Two short blocks: why the policy exists (one paragraph max) and who/when it applies. Cut narrative that repeats the title; flag `HR VERIFY` when scope crosses regions the user did not specify.

3. **At-a-glance layer** — **Quick summary** (3–5 bullets or one short paragraph): what employees must know before reading detail. This is the scan anchor—not a second copy of every rule.

4. **Definitions** — Only terms that change compliance meaning in *this* policy; omit dictionary glossaries.

5. **Policy statements** — Numbered **rules**: clear subject, modal verb (`must` / `may not` / `requires approval`), and outcome. One obligation per bullet; no mixed “policy + how-to” in the same bullet.

6. **Procedures** — Separate numbered flows: how to request, report, escalate, or document compliance (forms, systems, timelines from user only). Procedures reference rules; they do not restate them in prose.

7. **Roles & responsibilities** — Compact matrix or bullets: employee, manager, HR, IT, finance, legal—who does what at each step.

8. **Exceptions & approvals** — Who may grant exceptions, required documentation, and where records live. If unknown, `APPROVAL PATH — USER TO CONFIRM`.

9. **Related policies & references** — Cross-links and external regs **only** when user named them; no fabricated citations.

10. **Enforcement & questions** — Consequences or progressive discipline **only if user supplied**; otherwise `LEGAL/HR TO CONFIRM`. **Who to contact** for edge cases and a single questions channel.

11. **Revision histo