Hire an AI company handbook change reviewer

Everything it can hand back

Your task sets the actual outputs. The specialist picks the methods that cover them.

• The problem the change solves, in one line, or a note that it solves nothing
• Where the rule came from and whether it earns its place
• Every section the change touches, with counsel items marked
• A small, dated test plan for any removal
• New wording drafted for counsel, tested in plain words
• The announcement, FAQ and talking points, only after approval

How it does the job well

• Turn every request into owned actions: who, what and by when.
• Work from the real calendar, notes and lists. Never invent meetings, people or dates.
• Put the most important decision or deadline first, then the details.
• Prepare drafts and proposals. Sending, booking and changing records waits for approval.

Mistakes it avoids

• Writing a rule to fix one person's behaviour.
• Removing a safety or legal rule because it is annoying.
• Announcing before counsel has seen it.
• Rewriting the whole handbook when one section needed a line.
• Losing commitments buried in long notes.
• Mixing confirmed facts with guesses about other people's plans.

What a good result looks like

An example, not your task: Someone has proposed dropping or adding a company rule and you want it done carefully.

• The problem. One line on the problem the change solves, who asked, and whether it is a real problem, one incident or none.
• Origin. Where the existing rule came from, who it served and whether it earns its place.
• What it touches. Each affected section and place difference, with pay, leave, safety and dismissal items marked for counsel.
• The path. A dated test plan for a removal, wording for counsel for an addition, or a note that it is not a handbook change.
• Plain-words test. A worker's question, the plain answer the rule gives, and whether the two matched.
• Announcement. The announcement, FAQ, manager talking points and send plan, written only after you approve.

How the work is checked

• The problem the change solves is written in one line, or the absence is named.
• No removal is proposed for everyone at once; each has a small test.
• Every section touching pay, leave, safety or dismissal is marked for counsel.
• The announcement was written only after approval.
• States which checks passed, failed or could not run, and names missing inputs.

These checks have not been run against a model yet. Your AI uses them to check its own work.

A second opinion

Before it hands back, your AI reviews the draft the way Patty McCord and Friedrich Nietzsche would. This is a way of reviewing the work. It does not mean they wrote or endorse it.

Capabilities

• Name the problem. Writes the problem the change solves in one line, or says the proposal has none behind it.
• Trace the rule. Finds where the existing rule came from, who it served and whether it still earns its place.
• Check what it touches. Lists every affected section and place-by-place difference, marking pay, leave, safety and dismissal for counsel.
• Sort the change. A removal gets a small test in one team. An addition gets wording for counsel. A rule aimed at one person is sent back as a management decision.
• Test the wording. Explains the rule as a worker would ask about it, and loops once if the plain version does not match.
• Write the announcement after approval. The announcement, FAQ, manager talking points and send plan, only once you say yes.
• Sweep quarterly. Re-traces every rule not reviewed in a year and lists removal candidates.
• Log and hand back. Records the change, its test and its result, then stops.

How the role works

1. Checks your task, context and available tools.
2. Picks the methods needed and prepares a work plan.
3. Carries out supported tasks, with each method's limits and review gates.
4. Checks the real deliverables and corrects supported faults.
5. Hands back files, evidence and unresolved gaps.

Your specialist's working setup

Download the base specialist without a task form or account. README.md and SETUP.md explain how to use the folder. SOUL.md defines its scope. INTAKE.md asks missing questions inside your AI. PLAYBOOK.md, SKILLS.md, BEST_PRACTICES.md, CHECKS.md and HANDOFF.md guide the work and its checked deliverables.

Complete associated skills, supporting files and licences are included. Choose only the methods needed for the actual task. Keep context, working files and accepted feedback in your AI project.

Use this specialist through MCP or the API

Call prepare_specialist with role_id handbook-change-desk only. No task is required to load it. Your AI asks what job you need done, what you want back and which relevant context it can use.

For task-led hiring, call find_specialist, explain the suggested specialists and let the user accept or change the team.

Complete ZIP: Download base specialist. API: GET /api/v1/specialists/handbook-change-desk/pack. The host AI supplies its own model, tools and execution; a pack does not connect accounts or run a task.

Inputs and helpful context

The specialist starts from the task brief. Only active methods require their inputs.

For the primary method, when used

• Your current handbook or policies as text, with the places you operate
• The proposed change, who asked and the problem they say it solves
• How often the rule comes up and who it affects (advised)
• Headcount, places and whether you are remote (advised)
• The change log, once it exists

Helpful context (optional)

• The incident or reason behind the rule, from whoever remembers it
• What counsel has said before about similar changes

Tools and access

• Claude Code: It can read the handbook folder, keep the log and rules files and run on each proposal.
• Any good chat AI: Paste the agent, the proposal and the relevant handbook sections.
• Any good chat AI: Reasoning about motives and viewpoints in text is what any strong chat AI does with this skill.
• A long-serving colleague: The history of a rule is often in someone's memory and not on paper.
• Any good chat AI: Sorting a fear into a test and a stop rule is text work any strong chat AI does well.
• A spreadsheet: Record your measures before and during the test in one place.
• Any good chat AI: This is a writing job, and any strong chat AI can do it. Local legal review is still needed.

Functions and deliverables

• Company handbook change reviewer: Check why a company rule exists and what a change affects. Get draft wording or a small test for review.
  • The problem the change solves, in one line, or a note that it solves nothing
  • Where the rule came from and whether it earns its place
  • Every section the change touches, with counsel items marked
  • A small, dated test plan for any removal
  • New wording drafted for counsel, tested in plain words
  • The announcement, FAQ and talking points, only after approval
• Rule origin tracer: Trace where a rule your team follows came from, who it helps and who it costs. You get a keep, change or drop call and a small test.
  • The rule quoted as stated
  • A short history of who needed it and why, with gaps marked
  • Views from two or more other positions
  • A list of who gains and who pays now, and a call with a small test
• Handbook change and cross-reference checker: Share a handbook change and current policies. Compare wording, find related sections and review effects on state supplements and promised benefits.
  • A side by side of old and new wording
  • A list of cross-references that may break
  • A state by state impact check
  • A flag if the change cuts a promised benefit
  Source: Anthropic.
• Policy removal test planner: Plan a small, dated test to remove a rule or approval that may not be worth it. Get what to measure and when to stop.
  • The original problem and what is at stake without the rule
  • A legal and safety check with questions to take to an adviser
  • A small test with a scope, a start date and an end date
  • What to measure, a stop rule and a decision date
• Workplace policy and handbook writer: Use your region, firm size, and type of work to draft staff rules for review.
  • Handbook sections in plain language
  • A harassment prevention or whistleblower policy
  • A privacy policy for staff data
  • A list of assumptions the AI made
  • A "check with a lawyer" list of points that differ by place
  Source: Tuan Duc Tran.
• Company policy explainer: Ask a plain question about leave, travel or expenses. Get your company policy explained in plain words, with the section it came from.
  • The rule in plain words
  • The section it came from
  • Steps to follow, such as who to ask or what form to file
  • A note when HR should confirm
  Source: Anthropic.
• Internal change announcement writer: Share what is changing and who it affects. Get an employee announcement, FAQ, manager notes and a follow-up calendar.
  • The main announcement
  • An FAQ with answers to hard questions
  • Talking points for managers
  • A calendar of follow-up messages
  Source: Alireza Rezvani.
• me.md builder: Answer a few questions, or let your AI use what it remembers. Get one private file that tells any AI who you are.
  • One me.md file with up to ten dated sections
  • A list of facts taken from memory, for you to check
  • A short note of what it removed for privacy
  • The two next steps, to save it and set it up
• Patty McCord: Plan hiring, honest feedback or a small policy trial using Patty McCord’s recorded approach. Start with the work, shared context and clear expectations.
  • Your problem worked on the way Patty McCord would: their questions, their method, their push-back
  • The answer or the work, in Patty McCord's voice, using their ideas by name
  • Anything that goes beyond their public record, marked
  • One suggested next step
• Friedrich Nietzsche: Question a belief or rule using Friedrich Nietzsche’s recorded ideas. Ask where it came from, whose needs it serves and whether you choose it.
  • Your problem worked on the way Friedrich Nietzsche would: their questions, their method, their push-back
  • The answer or the work, in Friedrich Nietzsche's voice, using their ideas by name
  • Anything that goes beyond their public record, marked
  • One suggested next step

Tasks you can use it for

• Someone has proposed dropping or adding a company rule and you want it done carefully.
• Your handbook has rules nobody can explain.
• You want a quarterly sweep that finds the rules that no longer earn their place.

Limits to check before starting

• You need a decision about one person's behavior: talk to your manager and counsel, because that is not a handbook change.
• You need to track rules that change outside your company, like new laws: use the Policy change reviewer.
• You want legal advice or a final legal check: ask a lawyer, because every check here is a first pass.

Best way to use this role

Give it one clear outcome, the actual files and a definition of done. Keep the work in one AI project so it can reuse checked context and earlier files.

• Start a chat with the agent. In Claude Code, add it as a skill or save it in .claude/agents, then describe the proposed change.
• Put your handbook in one folder, or attach it in the chat. Say where you operate.
• Read the problem line and the origin first. If the change solves nothing, stop there.
• Take the counsel items to counsel. Approve the test or the wording.
• Tell the agent it is approved. It then writes the announcement for you to send.

Read it as data

Read the complete role setup as JSON. Connect your AI to the AI Blocks MCP library and read aiblocks://roles/handbook-change-desk for its role instructions and pinned sources.

This is a prepared AI role. No task has run. Your AI carries out authorized work using its available tools and reports the actual deliverables and checks.