Workplace Writing skill 08 of 13

Policies and Guidelines SKILL.md

The Policies and Guidelines skill makes Claude, Codex or another AI agent write workplace policies and guidelines people understand. It states purpose and scope, writes numbered rules with must and should, adds examples of allowed and not allowed, sets the exception route, responsibilities and proportionate consequences, puts a short summary first and flags where legal, HR or compliance review is needed.

policies-guidelines/SKILL.md, version 1.0.0, 1,472 words.

Policies tell people what they must or must not do; guidelines tell them how to do something well. Both often fail the same way: long legal-style text that people skim, unclear rules mixed with background, no examples, and nobody sure who the policy applies to or what happens if it is broken. When a policy is hard to understand, people either ignore it or follow it wrongly.

This skill writes and rewrites workplace policies and guidelines (such as expenses, remote work, travel, IT and data use, AI use, social media, leave and security) in plain language: purpose, scope, the rules stated clearly, examples, exceptions and approvals, responsibilities, consequences, related documents, an owner and a review date. It flags where legal or HR review is needed.

Operating principles

  1. Rules first, reasons second. State what people must do, then why.
  2. Plain language. Short sentences, "you" and "we", defined terms only where needed.
  3. Examples make rules real. Show allowed and not-allowed cases.
  4. Clear scope and exceptions. Who it applies to, what is excluded, how to ask for an exception.
  5. Legal review where it matters. Employment, data protection, health and safety and regulated areas need qualified review.

Policy versus guideline versus procedure

DocumentSaysExample
PolicyWhat must or must not happen"Customer data must not be pasted into public AI tools."
GuidelineHow to do something well (recommended)"Write prompts without personal names where possible."
Procedure (SOP)Steps to follow"1. Open the approved AI tool. 2. ..."

When to use, and when not to

Use it to write new policies, rewrite unclear ones, create guidelines (style, brand, AI use, social media) and prepare policy summaries for staff.

Don't use it as legal advice; employment contracts, regulated policies and anything with legal effect should be reviewed by HR, legal or compliance professionals. For step-by-step processes, use sops-process-docs.

Inputs

  • The topic and why a policy is needed now.
  • Existing policy, laws or standards that apply (to be confirmed by professionals).
  • Who it applies to and any regional differences.
  • Decisions already made by leadership.
  • Approval and review process.

Raw SKILL.md file

.md

This is the exact file agents read, including the credit header and licence notice. Save it as policies-guidelines/SKILL.md.

---
name: policies-guidelines
description: "Write workplace policies and guidelines people understand: purpose and scope, numbered rules using must and should, examples of allowed and not allowed, exceptions and approvals, responsibilities and proportionate consequences, a summary first, owner and review date, and flags for legal, HR or compliance review. Use whenever someone needs a workplace policy, guideline, AI use policy, expenses, remote work, travel, IT, data or social media policy, or wants to rewrite a long policy in plain language."
license: CC-BY-4.0
metadata:
  author: Afeez Adebayo
  author_url: https://afeezadebayo.com
  author_linkedin: https://www.linkedin.com/in/afeez-adebayo/
  author_x: https://x.com/za_icon
  source: https://afeezadebayo.com/skills-md/productivity/workplace-writing/policies-guidelines/
  category: productivity
  subcategory: workplace-writing
  version: "1.0.0"
  created: "2026-10-10"
  updated: "2026-10-13"
---

# Policies and Guidelines

Policies tell people what they must or must not do; guidelines tell them how to do something well. Both often fail the same way: long legal-style text that people skim, unclear rules mixed with background, no examples, and nobody sure who the policy applies to or what happens if it is broken. When a policy is hard to understand, people either ignore it or follow it wrongly.

This skill writes and rewrites workplace policies and guidelines (such as expenses, remote work, travel, IT and data use, AI use, social media, leave and security) in plain language: purpose, scope, the rules stated clearly, examples, exceptions and approvals, responsibilities, consequences, related documents, an owner and a review date. It flags where legal or HR review is needed.

## Operating principles
1. **Rules first, reasons second.** State what people must do, then why.
2. **Plain language.** Short sentences, "you" and "we", defined terms only where needed.
3. **Examples make rules real.** Show allowed and not-allowed cases.
4. **Clear scope and exceptions.** Who it applies to, what is excluded, how to ask for an exception.
5. **Legal review where it matters.** Employment, data protection, health and safety and regulated areas need qualified review.

## Policy versus guideline versus procedure
| Document | Says | Example |
|---|---|---|
| Policy | What must or must not happen | "Customer data must not be pasted into public AI tools." |
| Guideline | How to do something well (recommended) | "Write prompts without personal names where possible." |
| Procedure (SOP) | Steps to follow | "1. Open the approved AI tool. 2. ..." |

## When to use, and when not to
Use it to write new policies, rewrite unclear ones, create guidelines (style, brand, AI use, social media) and prepare policy summaries for staff.

Don't use it as legal advice; employment contracts, regulated policies and anything with legal effect should be reviewed by HR, legal or compliance professionals. For step-by-step processes, use `sops-process-docs`.

## Inputs
- The topic and why a policy is needed now.
- Existing policy, laws or standards that apply (to be confirmed by professionals).
- Who it applies to and any regional differences.
- Decisions already made by leadership.
- Approval and review process.

## Workflow

### Step 1. Define purpose and scope
Why the policy exists, who it covers and what it does not cover.

### Step 2. List the decisions
The actual rules leadership has decided. Where decisions are missing, list them as questions rather than inventing rules.

### Step 3. Write the rules
Numbered, short, using "must" for requirements and "should" for recommendations.

### Step 4. Add examples
Allowed and not allowed, or a table of common situations.

### Step 5. Exceptions and approvals
How to request an exception, who approves and how quickly.

### Step 6. Responsibilities and consequences
Who does what (employees, managers, owners) and what happens if the policy is not followed, in proportionate language.

### Step 7. Add a summary
Five to seven bullet points at the top for people who read nothing else.

### Step 8. Review and publish
Legal, HR or compliance review where needed; owner, version and review date; announce with the summary and where to ask questions.

## Policy template
```
Title: <policy name> | Owner: <role> | Version: <x> | Effective: <date> | Review: <date>
Summary: <5 to 7 bullets>
Purpose: <why this policy exists>
Scope: Applies to <who> | Does not cover <what>
Rules
1. You must ...
2. You must not ...
3. You should ...
Examples: <allowed and not allowed>
Exceptions: <how to request, who approves>
Responsibilities: <employees, managers, policy owner>
If the policy is not followed: <proportionate consequences>
Related documents: <links>
Questions: <contact>
Change log: <date, change>
```

## Example: AI use policy rules
```
1. Use only approved AI tools for work tasks (list: <tools>).
2. You must not paste customer personal data, passwords or confidential financial data into AI tools that are not approved for it.
3. You are responsible for checking AI output for accuracy before sharing it.
4. You should tell readers when content was largely written by AI, where that matters to them.
5. Ask <team> before using AI for decisions about people, such as hiring.
```

## Situation tables
| Situation | Allowed? | Notes |
|---|---|---|
| Taxi home after 9pm from the office | Yes | Up to \u00a330; keep the receipt |
| Client dinner | Yes | Up to \u00a360 per person; list attendees |
| Monitor for home office | Yes, once | Manager approval first |
| Alcohol without clients | No | |
Tables answer most questions faster than paragraphs and reduce requests for clarification.

## Writing rules clearly
| Unclear | Clear |
|---|---|
| Employees are expected to exercise appropriate judgement regarding expenditure. | You must get approval before spending more than \u00a3500. |
| Data should be handled in line with relevant regulations. | You must not email customer data outside the company. |
| Remote working may be permitted subject to agreement. | You can work remotely up to three days a week with your manager's agreement. |

## Announcing a policy
Share the summary, the effective date, what changed from the previous version, where to find the full policy and who to ask. Give people time to adjust before enforcement, and answer common questions in a short FAQ.

## Guidelines
Guidelines help people do things well rather than set rules. Keep them practical: principles, dos and don'ts, examples and templates. Brand, writing, social media and AI guidelines work best with many real examples and few abstract statements.

## Reviewing policies
Review each policy on its date, after relevant law changes, after incidents and when people repeatedly ask the same question. Archive old versions and keep a change log so people know what changed.

## Using an AI agent
The agent can turn leadership decisions into numbered rules, rewrite legal-style text in plain language, build situation tables, draft the summary and FAQ and list questions that need legal or HR review. It must not invent rules or state legal positions; missing decisions are listed as questions.

## Edge cases and judgement calls
| Situation | What to do |
|---|---|
| Legal requirements unclear | Mark for legal review; do not state legal positions as fact |
| Different rules by country | One core policy with country annexes |
| Policy conflicts with practice | Find out why; change the practice or the policy deliberately |
| Very long existing policy | Summary first; split into policy plus procedures |
| Sensitive topic (discipline, health) | HR review; respectful, neutral wording |
| Guideline rather than rule | Use "should" and examples; avoid "must" |

## Handoffs to related skills
- Procedures: `sops-process-docs`
- Plain language: `plain-english`
- Team writing standards: `team-style-guide`
- Staged writing: `writing-process`
- Email security rules: the Email and Inbox `email-security` skill

## Worked example
A company's 14-page expenses policy generated constant questions. The rewrite put a seven-bullet summary first, cut the rules to 12 numbered lines using "must" and "should", added a table of 10 common situations (taxis after 9pm, client meals, home office equipment), set a clear exception route with a two-day response time, and moved procedures to a linked SOP. Finance reviewed limits and HR reviewed consequences. Questions to finance fell sharply in the next quarter, and late expense claims dropped.

## Common mistakes
- Background before the rules.
- Legal jargon.
- No examples.
- Unclear scope or exception process.
- No owner or review date.

## Self-check before you answer
- Is there a summary and clear scope?
- Are rules numbered with must and should?
- Are examples and the exception route included?
- Is legal or HR review flagged where needed?

---
Created by Afeez Adebayo, Marketing, SEO & Growth Expert
Website: https://afeezadebayo.com
LinkedIn: https://www.linkedin.com/in/afeez-adebayo/
X: https://x.com/za_icon
Source: https://afeezadebayo.com/skills-md/productivity/workplace-writing/policies-guidelines/

Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). You may copy, share and adapt this skill, including for commercial use, as long as you credit Afeez Adebayo and link to the source URL above. Keep this notice in any copy or adapted version.

Workflow

Step 1. Define purpose and scope

Why the policy exists, who it covers and what it does not cover.

Step 2. List the decisions

The actual rules leadership has decided. Where decisions are missing, list them as questions rather than inventing rules.

Step 3. Write the rules

Numbered, short, using "must" for requirements and "should" for recommendations.

Step 4. Add examples

Allowed and not allowed, or a table of common situations.

Step 5. Exceptions and approvals

How to request an exception, who approves and how quickly.

Step 6. Responsibilities and consequences

Who does what (employees, managers, owners) and what happens if the policy is not followed, in proportionate language.

Step 7. Add a summary

Five to seven bullet points at the top for people who read nothing else.

Step 8. Review and publish

Legal, HR or compliance review where needed; owner, version and review date; announce with the summary and where to ask questions.

Policy template

Title: <policy name> | Owner: <role> | Version: <x> | Effective: <date> | Review: <date>
Summary: <5 to 7 bullets>
Purpose: <why this policy exists>
Scope: Applies to <who> | Does not cover <what>
Rules
1. You must ...
2. You must not ...
3. You should ...
Examples: <allowed and not allowed>
Exceptions: <how to request, who approves>
Responsibilities: <employees, managers, policy owner>
If the policy is not followed: <proportionate consequences>
Related documents: <links>
Questions: <contact>
Change log: <date, change>

Example: AI use policy rules

1. Use only approved AI tools for work tasks (list: <tools>).
2. You must not paste customer personal data, passwords or confidential financial data into AI tools that are not approved for it.
3. You are responsible for checking AI output for accuracy before sharing it.
4. You should tell readers when content was largely written by AI, where that matters to them.
5. Ask <team> before using AI for decisions about people, such as hiring.

Situation tables

SituationAllowed?Notes
Taxi home after 9pm from the officeYesUp to \u00a330; keep the receipt
Client dinnerYesUp to \u00a360 per person; list attendees
Monitor for home officeYes, onceManager approval first
Alcohol without clientsNo

Tables answer most questions faster than paragraphs and reduce requests for clarification.

Writing rules clearly

UnclearClear
Employees are expected to exercise appropriate judgement regarding expenditure.You must get approval before spending more than \u00a3500.
Data should be handled in line with relevant regulations.You must not email customer data outside the company.
Remote working may be permitted subject to agreement.You can work remotely up to three days a week with your manager's agreement.

Announcing a policy

Share the summary, the effective date, what changed from the previous version, where to find the full policy and who to ask. Give people time to adjust before enforcement, and answer common questions in a short FAQ.

Guidelines

Guidelines help people do things well rather than set rules. Keep them practical: principles, dos and don'ts, examples and templates. Brand, writing, social media and AI guidelines work best with many real examples and few abstract statements.

Reviewing policies

Review each policy on its date, after relevant law changes, after incidents and when people repeatedly ask the same question. Archive old versions and keep a change log so people know what changed.

Using an AI agent

The agent can turn leadership decisions into numbered rules, rewrite legal-style text in plain language, build situation tables, draft the summary and FAQ and list questions that need legal or HR review. It must not invent rules or state legal positions; missing decisions are listed as questions.

Edge cases and judgement calls

SituationWhat to do
Legal requirements unclearMark for legal review; do not state legal positions as fact
Different rules by countryOne core policy with country annexes
Policy conflicts with practiceFind out why; change the practice or the policy deliberately
Very long existing policySummary first; split into policy plus procedures
Sensitive topic (discipline, health)HR review; respectful, neutral wording
Guideline rather than ruleUse "should" and examples; avoid "must"
  • Procedures: sops-process-docs
  • Plain language: plain-english
  • Team writing standards: team-style-guide
  • Staged writing: writing-process
  • Email security rules: the Email and Inbox email-security skill

Worked example

A company's 14-page expenses policy generated constant questions. The rewrite put a seven-bullet summary first, cut the rules to 12 numbered lines using "must" and "should", added a table of 10 common situations (taxis after 9pm, client meals, home office equipment), set a clear exception route with a two-day response time, and moved procedures to a linked SOP. Finance reviewed limits and HR reviewed consequences. Questions to finance fell sharply in the next quarter, and late expense claims dropped.

Common mistakes

  • Background before the rules.
  • Legal jargon.
  • No examples.
  • Unclear scope or exception process.
  • No owner or review date.

Self-check before you answer

  • Is there a summary and clear scope?
  • Are rules numbered with must and should?
  • Are examples and the exception route included?
  • Is legal or HR review flagged where needed?

When to use this skill, and when not to

Use it for expenses, remote work, travel, IT and data use, AI use, social media, leave and security policies, and for guidelines such as brand or writing standards.

Long legal-style policies get skimmed or followed wrongly. Clear rules with examples and a simple exception route are followed far more often.

This skill is not legal advice; policies with legal effect need professional review. For step-by-step processes, use SOPs and Process Documents.

Example prompts

  • “Write an AI use policy for our company.”
  • “Rewrite our 14-page expenses policy in plain language.”
  • “Create a remote work policy with examples.”
  • “Turn these leadership decisions into numbered rules.”
  • “Write a social media guideline for staff.”

Example output

The skill returns a fixed structure. A shortened example for an imaginary site:

Summary: 7 bullets
Rules: 12 numbered lines using must and should
Situations table: 10 common cases
Exceptions: request route, 2-day response
Review flags: limits (finance), consequences (HR)
Owner, version, effective and review dates set

Rules are never invented; missing decisions are listed as questions for leadership, and legal points are flagged for review.

What the agent needs from you

The skill asks for the topic and why a policy is needed, any existing policy and standards, who it applies to and regional differences, decisions leadership has made and the approval and review process.

How this skill hands off to related skills

Policies and Guidelines connects to SOPs and Process Documents, Plain English and the Team Style Guide, plus Email Security for security rules.

How to run Policies and Guidelines step by step

  1. Install the skill as policies-guidelines/SKILL.md.
  2. Define purpose and scope.
  3. List leadership decisions.
  4. Write rules with examples.
  5. Add exceptions and a summary.
  6. Get professional review where needed.

Situations this skill handles

Policies and Guidelines handles these situations:

SituationWhat the skill does
Unclear legal requirementsMarked for review
Rules differ by countryCore policy plus annexes
Policy conflicts with practiceDeliberate change
Very long policySummary and split
Sensitive topicsHR review, neutral wording
Guideline not ruleShould and examples

What makes this skill different

Most policies are written for lawyers. This skill writes for the people who must follow them: rules first, plain words, real examples and a clear exception route, while flagging what needs professional review.

Practical tips from real projects

  • Rules first, reasons second.
  • Plain language.
  • Examples make rules real.
  • Legal review where it matters.

A worked example: a company's 14-page expenses policy generated constant questions. The rewrite put a seven-bullet summary first, cut the rules to 12 numbered lines using must and should, added a table of ten common situations such as taxis after nine at night, client meals and home office equipment, set an exception route with a two-day response time and moved procedures into a linked SOP. Finance reviewed limits and HR reviewed consequences, and questions to finance fell sharply the next quarter.

Policies, guidelines and procedures differ: a policy says customer data must not be pasted into public AI tools, a guideline recommends writing prompts without personal names, and a procedure lists the steps in the approved tool. The template covers title, owner, version, effective and review dates, summary, purpose, scope, rules, examples, exceptions, responsibilities, consequences, related documents, questions and a change log.

Unclear rules become clear ones: employees are expected to exercise appropriate judgement regarding expenditure becomes you must get approval before spending more than five hundred pounds. Situation tables answer most questions faster than paragraphs, showing what is allowed, limits and notes for common cases.

An AI use policy example shows the style: use only approved tools, never paste customer personal data, passwords or confidential financial data into unapproved tools, check AI output before sharing it, say when content was largely AI-written where that matters and ask before using AI for decisions about people. The workflow defines purpose and scope, lists the decisions leadership has made, writes the rules, adds examples, sets exceptions and approvals, states responsibilities and proportionate consequences, writes the summary and arranges legal, HR or compliance review before publishing with an owner and review date. Announcements share the summary, effective date, what changed and who to ask, with a short FAQ and time to adjust before enforcement. Policies are reviewed on their date, after law changes and incidents and whenever the same questions keep arriving, with old versions archived. An AI agent can turn decisions into numbered rules, rewrite legal-style text plainly, build situation tables and list questions for professional review, but it never invents rules or states legal positions.

Edge cases are handled with care. Where legal requirements are unclear, points are marked for review rather than stated as fact. Rules that differ by country use one core policy with country annexes. When a policy conflicts with everyday practice, the team finds out why and changes one or the other deliberately. Very long policies get a summary and are split into a policy plus procedures, sensitive topics such as discipline or health get HR review and neutral wording, and guidelines use should and examples rather than must. Common mistakes include background before rules, legal jargon, no examples, unclear scope or exception routes and no owner or review date.

Guidelines differ from policies because they help people do things well rather than set rules. Brand, writing, social media and AI guidelines work best with principles, dos and don'ts, templates and many real examples. The final check confirms the policy has a summary and clear scope, numbered rules using must and should, examples and an exception route, and that legal or HR review is flagged wherever the topic needs it.

Clear policies reduce questions, mistakes and disputes, and make expectations fair for everyone.

Frequently asked questions

A policy sets what must or must not happen; a guideline recommends how to do something well.

Numbered, short, using must for requirements and should for recommendations, with examples.

Yes, five to seven bullets at the top for people who read nothing else.

Employment, data protection, health and safety and regulated topics should be reviewed by professionals.

Through a clear route naming who approves and how quickly.

On a set date, after law changes or incidents and when the same questions keep coming up.

Changelog

  • 1.0.0, 10 October 2026: first release.

Use, adapt and share this skill under CC BY 4.0: credit Afeez Adebayo and link to this page.